paper(gqa): single-rank decode-streaming study — composite's masked bandwidth win

Adds the memory-bound mirror of the compute-bound prefill figure. New single-rank decode sweep (T_q=1, M=8) reuses the prefill_compute_bound kernels, sweeping per-rank S_kv with NO cross-CUBE reduce — isolating the local attention that the 64-way Case-6 decode masks under its reduce tail.

Finding: even memory-bound decode benefits from the composite command. Its scheduler-streamed concurrent per-tile DMAs reach ~233 GB/s (91% of the 256 GB/s per-rank roofline) while the primitive's blocking tl.dot serializes one tile DMA and plateaus at ~166 GB/s — a 25-28% latency win that widens with context. This refines the paper's 'decode is latency-neutral' claim: neutral at 64-way production scale (reduce-dominated), but the composite extracts bandwidth at the local level. The bandwidth roofline here mirrors the MAC roofline in prefill.

Adds: gqa_decode_streaming.py sweep (wired into the milestone umbrella as GQA_1H_SWEEPS=decode_streaming), paper_plot_gqa_decode_streaming.py, the figure, and a new fig:gqa-decode-stream + paragraph in 05-gqa.tex; rebuilt main.pdf. Prefill (Experiment B) re-verified byte-identical (794.1/668.1/646.9us).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

docs/report/1H-codesign-paper/sections/03-gemm.tex

@@ -172,16 +172,16 @@ kernel of \S\ref{sec:gqa} reaches for next — keeping the right
 working set on-chip so the composite pipeline lands in the
 compute-rich regime rather than the BW-bound one.
 
-\subsection{Why composite, and not user-orchestrated async loading?}
+\subsection{Why composite, and not kernel-orchestrated async loading?}
 \label{sec:gemm-vs-async}
 
 A reader familiar with double-buffered GEMM kernels on conventional
 hardware may ask: why a hardware-side composite command at all? Why
-isn't the obvious user-level pattern --- async-load each operand,
+isn't the obvious kernel-level pattern --- async-load each operand,
 overlap with compute, accumulate --- sufficient?
 
 To answer this concretely we contrast composite against two
-user-orchestrated baselines that have access to the same single-op
+kernel-orchestrated baselines that have access to the same single-op
 primitives the platform exposes (\textsf{tl.load} for async DMA into
 TCM, \textsf{tl.dot} for a single-op GEMM command on TCM-resident
 operands, \textsf{tl.store} for a DMA write-back). Both baselines
@@ -199,7 +199,7 @@ way at the runtime API surface to express ``start computing on
 tile 0 of $B$ while tile 1 is still in flight.'' Load-of-$B$ and
 GEMM therefore serialize.
 
-\paragraph{Async-tiled (chunked prefetch).} The user-level workaround is to
+\paragraph{Async-tiled (chunked prefetch).} The kernel-level workaround is to
 split $B$ along $K$ into \textsf{TILE\_K}-sized chunks, issue async
 \textsf{tl.load}s for those chunks, issue one \textsf{tl.dot} per
 chunk (each blocking only on its own $b_i$), and accumulate via
@@ -227,7 +227,7 @@ $K_{\text{KV}}=4096, d_{\text{head}}=128$ already needs
 --- past the cap. async-full and queue-all async-tiled are therefore
 not just slower than composite but \emph{architecturally infeasible}
 at LLM context length. The depth-2 async-tiled kernel is the only
-user-level
+kernel-level
 variant whose peak TCM footprint stays
 $O(2 \cdot \textsf{TILE\_K} \cdot N)$ regardless of $K$, the same
 order as composite's per-tile streaming buffer. It is the apples-to-apples
@@ -249,7 +249,7 @@ onto $K / \textsf{TILE\_K} = 48$ hardware tiles, so per-tile costs
 amplify into the largest measurable gap. The work content is
 identical for all four kernels: $\sim$6.3 M f16 MACs and
 $\sim$386 KiB of $B$ traffic from HBM, which together require
-$\sim$786 ns of GEMM-engine compute and $\sim$750 ns of DMA on a
+$\sim$786 ns of GEMM-engine compute and $\sim$781 ns of DMA on a
 saturated per-PE link. What differs is the number of host commands
 the same work is decomposed into --- 2 for composite (one
 \textsf{tl.load(A)} plus one composite), 4 for async-full, and 192
@@ -257,7 +257,7 @@ for either async-tiled variant (48 $A$-loads + 48 $B$-loads + 48
 \textsf{tl.dot}s + 47 elementwise adds + 1 store). The
 engine-pipeline-window throughput tracks that decomposition closely:
 composite reaches \SI{7.18}{\tera\flop\per\second} (post-overlap
-limit, only \SI{12}{\percent} below the \SI{8}{\tera\flop\per\second}
+limit, only \SI{10}{\percent} below the \SI{8}{\tera\flop\per\second}
 per-PE GEMM peak), async-full \SI{3.91}{\tera\flop\per\second}
 (DMA and compute serialize on a single big dot), and both async-tiled
 variants $\sim$\SI{2.53}{\tera\flop\per\second} (192 commands' worth of
@@ -272,7 +272,7 @@ specific simulator mechanisms.
 four issuance patterns: composite (one command, scheduler streams
 per-tile internally), async-full (one \textsf{tl.dot} on
 fully-loaded $B$), async-tiled with depth-2 double-buffer
-(TCM-bounded; the only user-level variant that scales to LLM
+(TCM-bounded; the only kernel-level variant that scales to LLM
 context length), and async-tiled with depth-$\infty$
 (all B-tiles queued up front; included as a sanity check that the
 prefetch depth is \emph{not} what separates the async-tiled kernel from
@@ -293,7 +293,7 @@ and all three async kernels beat it.}
 \end{figure*}
 
 \paragraph{Decomposing the gap.} Three structural mechanisms separate
-composite from the user-level baselines, and they layer.
+composite from the kernel-level baselines, and they layer.
 
 \emph{1. Inter-engine token routing happens below the host-side
 dispatch path.} The composite encodes the full
@@ -309,7 +309,7 @@ hand-offs total --- behind a single command from the host's point of
 view.
 
 \emph{2. \textsf{tl.dot} cannot replicate that per-tile pipeline at the
-user level.} A single-op GEMM command is handled on the GEMM engine as
+kernel level.} A single-op GEMM command is handled on the GEMM engine as
 a single monolithic compute timeout for the supplied $M{\times}K{\times}N$;
 there is no internal token loop that would let a streaming DMA of
 $B[i{+}1]$ overlap with the GEMM of $B[i]$ inside one
@@ -388,7 +388,7 @@ the four kernels to combine (a) macro-command dispatch at the host
 boundary (amortizing the structural CPU cost across all the work a
 single GEMM does), (b) scheduler-internal per-HW-tile streaming of
 DMA$\rightleftarrows$compute, and (c) TCM-bounded streaming buffer.
-User-orchestrated async kernels can have any two of those, not all
+Kernel-orchestrated async kernels can have any two of those, not all
 three: async-full pays one host dispatch (a) but forfeits per-tile
 overlap (b) and pins all of $B$ in TCM (c); depth-$\infty$ async-tiled
 achieves inter-chunk overlap but at $N_K$ host dispatches and

docs/report/1H-codesign-paper/sections/04-allreduce.tex

@@ -127,6 +127,63 @@ across whatever inter-device topology the configuration specifies,
 with the IPCQ ring buffer placed in on-PE TCM, PE-local HBM, or
 cube-shared SRAM---the third knob the results section sweeps.
 
+\subsection{Design alternatives}
+\label{sec:ipcq-alternatives}
+
+PE\_IPCQ is one point in a small space of hardware mechanisms for
+moving a short message from one PE to a neighbor and signalling its
+arrival. Three established alternatives anchor the space, each the
+HW realization of a familiar host-networking idea: a \emph{doorbell +
+polling} scheme (the classic MMIO doorbell---write the payload by DMA,
+write a doorbell, let the peer poll or take an interrupt); a
+\emph{hardware message queue} (HMQ, the NVLink-style descriptor engine
+that pushes a queue entry to the peer, with large payloads still
+riding a second DMA); and a \emph{completion-queue} design (RDMA-CQ,
+the InfiniBand/RoCE pattern where a DMA write auto-posts a completion
+entry the peer's CQ polls). PE\_IPCQ is the fourth: a hardware ring
+with credit return, splitting the control plane into PE\_IPCQ and the
+data plane into PE\_DMA, with head updates riding the payload and tail
+updates riding a 16\,B side-channel credit (\S\ref{sec:allreduce}).
+
+\begin{figure*}[t]
+\centering
+\includegraphics[width=0.78\linewidth]{ipcq_alternatives_architecture_flow.png}
+\caption{Per-send data and control flow for the four PE-to-PE
+signalling mechanisms (sender\,$\rightarrow$\,NoC\,$\rightarrow$\,receiver).
+Doorbell and RDMA-CQ each issue two fabric transactions (payload then
+doorbell / completion) and leave the peer polling or taking an
+interrupt; HMQ adds a dedicated descriptor engine but still moves large
+payloads on a second DMA; PE\_IPCQ folds head-pointer signalling into
+the payload flit train and returns the tail credit on a side channel,
+so a send is one MMIO write and a receive is a flip-flop read. This is
+a \emph{design schematic}, not a measured comparison.}
+\label{fig:ipcq-arch}
+\end{figure*}
+
+\begin{figure}[t]
+\centering
+\includegraphics[width=\linewidth]{ipcq_alternatives_decision_matrix.png}
+\caption{Why the ring+credit design was chosen, across five criteria:
+single-send latency, whether the host CPU sits on the critical path,
+whether the receiver must poll or take a wake-up interrupt, whether the
+control and data datapaths are duplicated, and whether the mechanism is
+right-sized for single-owner PE-to-PE traffic (rather than a
+multi-tenant fabric). PE\_IPCQ is the only design that clears every
+criterion. The accompanying per-send step-count tally
+($\sim$28 control events for IPCQ versus $\sim$38 for HMQ, $\sim$53 for
+RDMA-CQ, and $\sim$56 for doorbell+polling) is an \emph{illustrative}
+order-of-magnitude comparator over hand-counted pipeline steps---not a
+simulator measurement. The measured, simulator-grounded results follow
+in the next subsection.}
+\label{fig:ipcq-decision}
+\end{figure}
+
+The qualitative comparison motivates the design but is not a
+quantitative claim: the cycle-step tallies above are hand-counted
+control events, deliberately separated from the measured latencies that
+follow. Everything in the results subsection runs on the PE\_IPCQ
+substrate and is simulator-grounded.
+
 \subsection{Results}
 
 All measurements in this section run on the PE\_IPCQ substrate
@@ -138,8 +195,21 @@ the collective sweep builds its own six-device (six-SIP, $2\times3$)
 configurations---distinct from the two-SIP default of
 Table~\ref{tab:hw}---and measures all-reduce latency as a function of
 payload size for three inter-device topologies: a 1D ring, a 2D mesh
-(no wrap), and a 2D torus. Table~\ref{tab:allreduce} and
+(no wrap), and a 2D torus (Figure~\ref{fig:allreduce-topo}).
-Figure~\ref{fig:allreduce-cmp} report the result.
+Table~\ref{tab:allreduce} and Figure~\ref{fig:allreduce-cmp} report the
+result.
+
+\begin{figure}[t]
+\centering
+\includegraphics[width=\linewidth]{allreduce_topology.png}
+\caption{The three six-device ($2\times3$) inter-device topologies the
+collective sweep runs over, and the hierarchical local-reduce /
+global all-reduce-broadcast schedule mapped onto each: a 1D ring, a 2D
+mesh (no wrap-around), and a 2D torus (wrap-around links on both axes).
+The torus's wrap links shorten the worst-case reduction path, which is
+what the latency sweep below rewards.}
+\label{fig:allreduce-topo}
+\end{figure}
 
 \begin{table}[t]
 \centering

docs/report/1H-codesign-paper/sections/05-gqa.tex

@@ -27,15 +27,83 @@ the established baseline rather than being re-derived.
 \subsection{Data Placement Policy}
 \label{sec:gqa-placement}
 
-% TODO: compare data-placement options that apply across both the
+Long-context decode is bound by the KV cache, so the first-order design
-% short- and long-context regimes. Candidate axes:
+question is how to place that cache---and the running softmax state it
-%   - KV cache:  per-CUBE shard vs. replicate; per-PE shard vs. replicate
+feeds---across the two hardware axes the machine exposes: the CUBEs and,
-%   - Q / W_qkv / W_o weights:  static partition across CUBEs and PEs
+within each CUBE, the PEs (here $C{=}8$ CUBEs $\times$ $P{=}8$ PEs, for
-%   - Workspace (m, l, O softmax state):  scratch arena placement
+$C\!\cdot\!P{=}64$ attention engines over one KV-head group on the
-% The 4-case taxonomy used for long-context decode in
+LLaMA-3.1-70B target). Each axis can \emph{replicate} the KV cache or
-% \S\ref{sec:gqa-long} (Cube-{SP,Repl} x PE-{TP,SP}) is one instantiation
+\emph{shard} it, and a shard can run along the sequence dimension
-% of this framework; the short-context mapping in \S\ref{sec:gqa-short}
+$S_{kv}$ or the head dimension $d_{\text{head}}$. The cross product is a
-% is another.
+small, enumerable taxonomy; six placements span its meaningful corners
+(Figure~\ref{fig:gqa-kv-sharding}).
+
+\begin{figure}[t]
+\centering
+\includegraphics[width=\linewidth]{gqa_long_ctx_6cases_kv_sharding_diagram.png}
+\caption{The six KV-placement strategies, drawn on the
+$S_{kv}\!\times\!d_{\text{head}}$ KV tensor (rows = sequence, columns =
+head dimension). Cube colour bands and dashed PE dividers show which
+axis each level shards. Cases~1--3 either replicate the cache or shard
+it on a single axis (8-way at most); Cases~4--6 reach a full 64-way
+split, three different ways: Case~4 splits $S_{kv}$ across CUBEs and
+$d_{\text{head}}$ across PEs, Case~5 the mirror, and Case~6~$\star$
+splits $S_{kv}$ on \emph{both} axes.}
+\label{fig:gqa-kv-sharding}
+\end{figure}
+
+Two quantities decide which placement is viable, and they pull against
+each other. The first is \textbf{per-PE KV memory}. With a per-PE HBM
+budget of \SI{6.0}{\giga\byte} and \SI{1.76}{\giga\byte} of attention
+weights resident, the KV headroom is \SI{4.24}{\giga\byte} per PE. At a
+production context of $S_{kv}{=}1\,\text{M}$ tokens the unsharded cache
+is \SI{40}{\giga\byte}/PE (Case~1), an 8-way shard is \SI{5}{\giga\byte}
+(Cases~2--3)---both \emph{over} the headroom---while only the 64-way
+placements bring it to \SI{640}{\mega\byte}/PE (Cases~4--6), comfortably
+inside budget. Memory alone therefore eliminates Cases~1--3 at long
+context. The second quantity is \textbf{communication per token}, and it
+is what separates the three survivors. Sharding $d_{\text{head}}$
+(Cases~4--5) makes each PE hold only a slice of every head, so the
+$Q\!\cdot\!K^{\top}$ score is \emph{partial} and must be all-reduced
+across the slice owners on every token---a reduction whose volume scales
+with $S_{kv}$ ($\sim$\SI{166}{\mega\byte}/token analytically, intra-CUBE
+on the NoC for Case~4, inter-CUBE on UCIe for Case~5). Case~6~$\star$
+instead shards $S_{kv}$ on both axes, so every PE computes a
+\emph{complete} score over its own token range and only the small running
+softmax state $(m,\ell,O)$ is merged across PEs
+($\sim$\SI{6.2}{\mega\byte}/token)---a $\sim$27$\times$ lighter collective
+than the $d_{\text{head}}$-split designs.
+
+\begin{figure}[t]
+\centering
+\includegraphics[width=\linewidth]{gqa_long_ctx_6cases_summary.png}
+\caption{Long-context placement analysis at $S_{kv}{=}1\,\text{M}$ tokens.
+\emph{Left:} the per-PE HBM budget---\SI{1.76}{\giga\byte} of attention
+weights leave \SI{4.24}{\giga\byte} of KV headroom (red line).
+\emph{Middle:} per-PE KV memory per case (log scale); only the 64-way
+placements (Cases~4--6, \SI{640}{\mega\byte}) clear the headroom, while
+the unsharded (\SI{40}{\giga\byte}) and 8-way (\SI{5}{\giga\byte}) cases
+overflow. \emph{Right:} analytical communication per token (log scale);
+the $d_{\text{head}}$-split Cases~4--5 pay a partial-score all-reduce
+($\sim$\SI{166}{\mega\byte}/token) that the both-axes-$S_{kv}$ split of
+Case~6~$\star$ avoids ($\sim$\SI{6.2}{\mega\byte}/token, merging only the
+softmax state). On these two axes Case~6 (marked $\star$ in the figure)
+is the single placement that lands both inside the memory budget and at
+low per-token communication; whether that combination is the right one to
+pick is a regime-specific question taken up next.}
+\label{fig:gqa-budget}
+\end{figure}
+
+These two costs---per-PE KV memory and per-token communication---are
+intrinsic properties of each placement, fixed by how it shards the cache
+and independent of the workload regime. They do not by themselves name a
+winner: a short prompt where the whole cache fits on one PE values low
+communication and tolerates replication, whereas a million-token decode
+is bound by the memory wall and will pay communication to escape it.
+Which placement is appropriate is therefore a per-regime question, which
+the short- and long-context subsections that follow answer by running the
+options on the simulator and reading off latency, traffic, and the
+redundant compute each one induces.
 
 \subsection{Inference with Short-Context Length}
 \label{sec:gqa-short}
@@ -54,9 +122,10 @@ overlaps score computation; and per-tile \emph{scratch recycling} that
 keeps the running softmax accumulators ($m,\ell,O$) in a persistent arena
 while freeing per-tile temporaries, so the kernel fits the
 \SI{1}{\mebi\byte} scratch budget across many tiles. A further refinement
-that restructures the decode step into two stateful composites (a named
+restructures the decode step so the per-tile matrix products and the
-\textsf{softmax\_merge} recipe) is designed but not yet wired into the
+online-softmax merge are issued as \emph{composite} commands rather than
-measured path; results below reflect the implemented kernel only.
+hand-tiled primitives; \S\ref{sec:gqa-composite} measures it against the
+primitive baseline at long context.
 
 % TODO: CUBE <-> KV-head mapping diagram for the short-context regime
 % (h_kv=8 KV heads -> 8 CUBEs, 1:1; intra-CUBE PE usage).
@@ -75,78 +144,248 @@ measured path; results below reflect the implemented kernel only.
 
 % TODO: prefill long-context performance figure.
 
-The four headline panels above stress the kernel at moderate context
+Long-context decode is the regime where the KV cache, not attention
-lengths. Long-context decode---the regime where KV cache size, not
+compute, sets serving cost, so the placement question of
-attention compute, sets serving cost---turns the choice of how to
+\S\ref{sec:gqa-placement} becomes decisive here. To pick the right
-parallelize across cubes and PEs into a first-order design knob. We
+placement for this regime we run each of the six options as a fused
-compare four strategies on the LLaMA-3.1-70B single-KV-head-group
+decode kernel on the simulator (one decode step on the LLaMA-3.1-70B
-target (8 CUBEs $\times$ 8 PEs, one KV-head group):
+single-KV-head-group target, $C{=}8$ CUBEs $\times$ $P{=}8$ PEs) at a
+tractable $S_{kv}{=}8192$, and read off end-to-end latency, on-device op
+traffic, and the redundant compute each one induces. The swept context is
+small enough that all six fit in memory at $S_{kv}{=}8192$; the
+placements the long-context memory budget rules out (Cases~1--3) are
+drawn in red, run here only to expose their issue and communication
+structure.
 
-\begin{itemize}\setlength\itemsep{1pt}
+\begin{figure}[t]
-  \item \textbf{Case 1} (Cube-SP $\times$ PE-TP): KV split by $S_{kv}$
+\centering
-        across CUBEs; PEs tensor-parallel on the batch dimension
+\includegraphics[width=\linewidth]{gqa_decode_long_ctx_6cases_latency.png}
-        (wastes PE-TP work at $B{=}1$).
+\caption{Measured end-to-end decode latency per placement
-  \item \textbf{Case 2} (Cube-Repl $\times$ PE-TP): full KV
+($S_{kv}{=}8192$; red = ruled out by the long-context memory budget,
-        replicated to every CUBE; PEs tensor-parallel on batch.
+blue = the predicted Pareto choice). The fastest raw latency belongs to
-  \item \textbf{Case 3} (Cube-Repl $\times$ PE-SP): full KV
+Case~3 (\SI{17.8}{\micro\second})---but Case~3 replicates the full KV
-        replicated; PEs sequence-parallel on $S_{kv}$ with an
+cache into every CUBE, so it overflows the per-PE budget at production
-        intra-CUBE all-reduce.
+context and wastes 8$\times$ the compute
-  \item \textbf{Case 4} ($\star$, Cube-SP $\times$ PE-SP): KV split
+(Figure~\ref{fig:gqa-6cases-par}). Among the placements that actually fit
-        64-way (across both CUBEs and PEs) with a two-phase all-reduce
+1\,M-token memory (Cases~4--6), Case~6~$\star$ is the fastest
-        on the running softmax state $(m, \ell, O)$.
+(\SI{30.6}{\micro\second}, versus \SI{31.4}{} and \SI{34.5}{\micro\second}
+for the $d_{\text{head}}$-split Cases~4 and~5)---making it the placement
+of choice for long-context decode.}
+\label{fig:gqa-6cases-lat}
+\end{figure}
+
+\begin{figure}[t]
+\centering
+\includegraphics[width=\linewidth]{gqa_decode_long_ctx_6cases_parallelism.png}
+\caption{Redundant compute per placement, measured as active-PE
+$\times$ $S_{\text{local}}$ (PE-tokens; lower means less wasted work).
+The minimum is \num{8192} PE-tokens---one pass over the sequence.
+Case~3 inflates this 8$\times$ to \num{65536} by replicating the KV
+cache across all eight CUBEs so every CUBE redundantly re-attends the
+whole sequence; the $d_{\text{head}}$-split Cases~4--5 likewise carry
+\num{65536} because each token is processed across eight head slices.
+Case~6~$\star$ achieves the full 64-way split at the minimal
+\num{8192} PE-tokens---fully parallel, no replication.}
+\label{fig:gqa-6cases-par}
+\end{figure}
+
+\begin{figure}[t]
+\centering
+\includegraphics[width=\linewidth]{gqa_decode_long_ctx_6cases_traffic.png}
+\caption{Measured on-device op traffic per placement. The unsharded
+Case~1 issues no IPCQ copies (each PE has the full cache, nothing to
+reduce); the single-axis Case~3 charges 168. Among the 64-way splits,
+the $d_{\text{head}}$-split Cases~4--5 charge the most---280 IPCQ copies
+and 8 DMA writes each, the partial-score all-reduce that head-slicing
+forces---while Case~6~$\star$ needs only 189 IPCQ copies and a single DMA
+write, because it merges just the running softmax state $(m,\ell,O)$
+rather than partial scores. This is the on-device collective traffic that
+PE\_IPCQ and the torus links of \S\ref{sec:allreduce} are provisioned to
+absorb at link speed.}
+\label{fig:gqa-6cases-traffic}
+\end{figure}
+
+The measurements select the right placement for this regime. The fastest
+raw latency (Case~3, \SI{17.8}{\micro\second}) comes from replicating the
+full KV cache into every CUBE---which is exactly the placement the
+long-context memory budget forbids, and which the parallelism panel shows
+wastes 8$\times$ the compute. Restricting attention to the placements that
+fit production-context memory (the 64-way splits, Cases~4--6), the choice
+is Case~6~$\star$: it is the fastest of the three
+(\SI{30.6}{\micro\second}), and the op-count panel shows why---its
+softmax-state-only reduction charges 189 IPCQ copies and one DMA write
+against the 280 copies and 8 DMA writes the $d_{\text{head}}$-split
+Cases~4--5 pay for their partial-score all-reduce. For long-context
+decode, then, the appropriate data placement is the both-axes sequence
+shard (Case~6): it is the cheapest-communicating member of the only
+memory-feasible family, and the cross-PE softmax reduction it does pay is
+precisely the traffic the communication-side codesign of this report is
+built to move quickly.
+
+\subsection{Use of Composite Commands}
+\label{sec:gqa-composite}
+
+The decode kernel of \S\ref{sec:gqa-long} issues its local attention as
+primitive operations: it walks each PE's $S_{\text{local}}$ token slice in
+\SI{1024}{}-token tiles, and for every tile issues a $Q\!\cdot\!K^{\top}$
+\textsf{dot}, the online-softmax primitives, and a $P\!\cdot\!V$
+\textsf{dot}, merging the running $(m,\ell,O)$ state by hand. The number
+of PE\_CPU commands this costs grows with the context. At a production
+context of $S_{kv}{=}1\,\text{M}$ tokens the Case-6 64-way split gives
+each PE $S_{\text{local}}{=}16384$ tokens, so its local attention is a
+$Q\!\cdot\!K^{\top}$ of $(8,128)\!\cdot\!(128,16384)$ and a
+$P\!\cdot\!V$ of $(8,16384)\!\cdot\!(16384,128)$---sixteen hand-issued
+tiles, each a fresh batch of CPU commands.
+
+The composite command lets the kernel hand that tiling to PE\_SCHEDULER.
+We compare three command forms of the \emph{same} Case-6 kernel---identical
+placement and identical $(m,\ell,O)$ reduce, differing only in how the
+local attention is issued:
+\begin{itemize}
+  \item \textbf{primitive}---the hand-tiled \textsf{dot}/softmax kernel
+    above (the baseline of \S\ref{sec:gqa-long}).
+  \item \textbf{composite}---each matrix product is one coarse
+    \textsf{composite} GEMM over the \emph{whole} $S_{\text{local}}$, with
+    $K$ and $V$ passed as HBM references so PE\_SCHEDULER streams and
+    tiles them on the fixed $32\!\times\!64\!\times\!32$ MAC tile; the
+    softmax stays primitive.
+  \item \textbf{composite\,+\,softmax\_merge}---additionally folds the
+    online-softmax merge and $P\!\cdot\!V$ into a single stateful
+    \textsf{softmax\_merge} recipe composite.
 \end{itemize}
 
 \begin{figure}[t]
 \centering
-\includegraphics[width=\linewidth]{gqa_decode_long_ctx_4cases_latency.png}
+\includegraphics[width=\linewidth]{gqa_decode_long_ctx_composite.png}
-\caption{End-to-end decode latency per parallelism strategy
+\caption{Three command forms of the Case-6 decode kernel, swept over
-(LLaMA-3.1-70B single-KV-head group, 8 CUBEs $\times$ 8 PEs).
+context length ($S_{\text{local}}{=}S_{kv}/64$ per PE). \emph{Right:}
-Replication into CUBEs (Cases 2/3) wins the latency race
+PE\_CPU commands issued. The hand-tiled primitive kernel rises
-(\SI{20.2}{\micro\second} for Case 3), but Case~4 ($\star$, KV split
+$O(n_{\text{tiles}})$---from 96 commands at one tile to 426 at the
-64-way) finishes within \SI{14}{\micro\second} of the leader while
+1\,M-token, sixteen-tile production point---while both composite forms
-paying a different cost---visible in Figure~\ref{fig:gqa-4cases-mem}.}
+issue a context-\emph{independent} $O(1)$ count (94 and 98) that
-\label{fig:gqa-4cases-lat}
+\emph{saturates}: one coarse descriptor offloads the entire per-tile
+fan-out. \emph{Left:} the consequence for wall-clock latency is none---all
+three land on the same curve (\SI{30.6}{}, \SI{231}{},
+\SI{461}{\micro\second} at 8\,K\,/\,64\,K\,/\,128\,K), because decode is
+bound by streaming the KV cache, not by issue. Command-count is measured
+at emit time (exact, to 1\,M); latency on the data-mode engine over the
+tractable range.}
+\label{fig:gqa-composite}
 \end{figure}
 
+Figure~\ref{fig:gqa-composite} reads off the two quantities that matter,
+and they point in opposite directions. The PE\_CPU command count (right)
+collapses from a context-growing $O(n_{\text{tiles}})$ to a flat $O(1)$:
+at 1\,M tokens the composite form issues \num{94} commands against the
+primitive kernel's \num{426}, a $4.5\times$ reduction
+($4.2\times$ in modeled dispatch cost), and---crucially---that number no
+longer grows with context. The wall-clock latency (left), by contrast,
+is unchanged across all three forms: decode is bound by streaming the KV
+cache out of HBM, so the command form does not move the critical path.
+
+That juxtaposition is the point. The composite command is not a latency
+optimization for this memory-bound decode \emph{at the 64-way production scale} (a
+single-rank caveat follows, Figure~\ref{fig:gqa-decode-stream}); it is a
+\emph{CPU-issue}
+optimization. Its value is removing the per-tile dispatch work that would
+otherwise grow without bound as context grows, freeing PE\_CPU to run
+ahead and keep the engines fed---which is exactly what lets the
+data-movement cost analyzed next show through as the true bottleneck
+rather than being masked by issue overhead. The \textsf{softmax\_merge}
+recipe folds the online merge into the same descriptor; on this
+memory-bound path its marginal cost over the plain GEMM composite is small
+(98 vs.\ 94 commands), and like the plain composite it keeps the issued
+count flat as context scales.
+
+\paragraph{Isolating the rank: the masked streaming win.} That neutrality
+is a property of the \emph{full 64-way} critical path, not of the local
+attention: at production scale the inter-CUBE $(m,\ell,O)$ reduce tail and
+shared-HBM contention set the wall clock, so a faster local attention does
+not surface. Stripping those away---a single rank, no cross-CUBE reduce,
+swept over the per-rank context $S_{kv}$ (so $S_{kv}{=}16$\,K here is the
+per-PE load of a 1M-token, 64-way-sharded decode)---exposes the local
+attention directly (Figure~\ref{fig:gqa-decode-stream}), and the composite
+\emph{does} win, by \SI{25}{}--\SI{28}{\percent}. The reason is the
+memory-bound mirror of prefill: its scheduler-streamed concurrent per-tile
+DMAs keep the HBM pipeline full and reach \SI{233}{\giga\byte\per\second}
+---\SI{91}{\percent} of the per-rank \SI{256}{\giga\byte\per\second}
+roofline---whereas the primitive kernel's blocking \textsf{tl.dot}
+serializes one tile DMA at a time and plateaus at
+\SI{166}{\giga\byte\per\second}. So even for memory-bound decode the
+composite is not \emph{only} a CPU-issue optimization---it also extracts
+bandwidth---but that latency benefit materializes only when the local
+attention is on the critical path, which at 64-way production scale it is
+not.
+
 \begin{figure}[t]
 \centering
-\includegraphics[width=\linewidth]{gqa_decode_long_ctx_4cases_memory.png}
+\includegraphics[width=\linewidth]{gqa_decode_streaming.png}
-\caption{Per-CUBE KV memory footprint for the four cases. Cases 1
+\caption{Single-rank memory-bound decode ($T_q{=}1$, $M{=}8$), three
-and 4---both with the KV cache split across cubes (Cube-SP)---hold
+command forms, swept over per-rank context. \emph{Left:} end-to-end
-only \SI{0.5}{\mebi\byte} of KV state per CUBE; Cases 2 and 3, which
+latency---the composite forms run \SI{25}{}--\SI{28}{\percent} below the
-replicate the full KV, hold \SI{4}{\mebi\byte} per CUBE, an
+primitive, a gap that widens with context. \emph{Right:} achieved HBM
-\textbf{8$\times$} blowup at this configuration that scales linearly
+bandwidth against the per-rank \SI{256}{\giga\byte\per\second} roofline.
-with context length.}
+The primitive's blocking load$\rightarrow$dot serializes the KV stream and
-\label{fig:gqa-4cases-mem}
+plateaus at \SI{166}{\giga\byte\per\second}; the composite forms pipeline
+concurrent per-tile DMAs through the scheduler and reach
+\SI{233}{\giga\byte\per\second}. This is the memory-bound mirror of the
+prefill result (Figure~\ref{fig:gqa-prefill-cb}): there the composite
+approaches the MAC roofline, here the bandwidth roofline. Capped at
+$16$\,K---the plain composite materializes the full $(M,S_{kv})$ scores in
+TCM, so beyond that only the \textsf{softmax\_merge} recipe, which tiles
+the softmax, stays within scratch.}
+\label{fig:gqa-decode-stream}
 \end{figure}
 
+\paragraph{The compute-bound mirror: prefill.} Decode's \emph{production}
+verdict---command form is latency-neutral at 64-way scale---is a property
+of its regime, not of the
+composite command. A decode step has $T_q{=}1$, so its score and context
+products are skinny ($M{=}G\,T_q{=}8$): the MAC array is barely fed and
+the kernel is bound by streaming the KV cache. Prefill is the opposite
+corner. It processes a block of query positions at once, so $M{=}G\,T_q$
+is large and tile-filling, the GEMMs carry real arithmetic intensity
+($\sim$$M$ flops/byte, well above the roofline ridge), and the kernel is
+\emph{compute-bound}. This is the regime the composite command was built
+for (\S\ref{sec:gemm}): it streams the per-HW-tile
+DMA$\rightleftarrows$compute pipeline so the MAC array stays fed, whereas
+the primitive kernel's blocking \textsf{tl.dot} serializes each tile's
+load and compute and starves the array between tiles. We run the same
+three command forms on a single-rank compute-bound prefill (FlashAttention
+$Q$-block $\times$ $S_{kv}$-tile, online softmax) and sweep the context
+length (Figure~\ref{fig:gqa-prefill-cb}).
+
 \begin{figure}[t]
 \centering
-\includegraphics[width=\linewidth]{gqa_decode_long_ctx_4cases_traffic.png}
+\includegraphics[width=\linewidth]{gqa_prefill_compute_bound.png}
-\caption{Per-case op-count breakdown. The replicated-KV PE-TP design
+\caption{Compute-bound prefill, three command forms, swept over context
-(Case 2) avoids almost all on-device communication
+length ($M{=}8\,T_q$ tile-filling). \emph{Left:} end-to-end latency.
-(\textasciitilde0 IPCQ copies), but at the cost of KV memory.
+\emph{Right:} MAC utilization (achieved $\div$ the
-Case~4's two-phase reduce charges \textasciitilde190 IPCQ copies and
+\SI{8}{\tera\flop\per\second} per-PE peak). The hand-tiled primitive sits
-\textasciitilde190 DMA reads---this is the traffic that PE\_IPCQ
+flat at $\sim$\SI{68}{\percent}---its serial load$\to$dot path leaves the
-(\S\ref{sec:allreduce}) is built to absorb at on-device speed.}
+MAC array idle between tiles regardless of context. The composite forms
-\label{fig:gqa-4cases-traffic}
+climb with context (\SI{67}{}$\to$\SI{80}{\percent} plain,
+\SI{67}{}$\to$\SI{83}{\percent} with the recipe) because a deeper $P\!\cdot
+\!V$ reduction gives more HW tiles to pipeline, and they convert that into
+wall-clock: at \num{1024} the recipe form is \SI{646.9}{} vs.\
+\SI{794.1}{\micro\second} (\SI{19}{\percent} faster). The margin
+\emph{grows} with context---the compute-bound mirror of the GEMM result of
+\S\ref{sec:gemm}.}
+\label{fig:gqa-prefill-cb}
 \end{figure}
 
-Three things stand out. First, the fastest case in pure latency
+The two studies together state the composite command's value precisely. It
-(Case~3, \SI{20.2}{\micro\second}) is also the most memory-hungry,
+has two distinct benefits, and which one matters is set by the workload's
-requiring the full KV state on every CUBE---an option that fails to
+roofline position. The first is \emph{host-issue offload}: one macro
-scale once context length blows past the per-CUBE budget. Second,
+command in place of $O(n_{\text{tiles}})$ fine ones, which removes
-Case~4's KV-split design gives back roughly \SI{14}{\micro\second}
+PE\_CPU dispatch work and is regime-independent (it shows in the decode
-versus Case~3 in exchange for an \textbf{8$\times$} KV-memory
+command count). The second is \emph{MAC-array feeding}: the
-reduction; for practical long-context serving where KV capacity is
+scheduler-internal per-tile DMA$\rightleftarrows$compute pipeline, which
-the binding constraint, this is the trade the design chooses
+only converts to latency when the workload is compute-bound enough to have
-(marked $\star$). Third, Case~4 pays its way in
+a MAC array worth keeping busy (it shows in the prefill utilization). A
-\emph{communication}: the op-count panel shows \textasciitilde190
+memory-bound decode exercises only the first; a compute-bound prefill
-IPCQ copies and \textasciitilde190 DMA reads, precisely the
+exercises both. The composite command is the single mechanism that
-on-device collective traffic that PE\_IPCQ and the torus links of
+delivers each where it applies.
-\S\ref{sec:allreduce} are provisioned to move quickly---so the
-``slower'' strategy is in fact the one that fully cashes in the
-communication-side codesign work of this report.
 
 \subsection{Comprehensive Analysis}
 \label{sec:gqa-analysis}

scratch_dot_tiling_experiment.py

@@ -0,0 +1,204 @@
+#!/usr/bin/env python3
+"""SCRATCH EXPERIMENT (not production; do not commit).
+
+Question: does charging the *primitive* decode kernel for per-HW-tile
+(16x16x16) CPU dispatch flip the "composite gives no decode-latency
+benefit" conclusion?
+
+We monkeypatch TLContext.dot so that, in the primitive kernel, every
+tl.dot whose (M,K,N) exceeds the HW GEMM tile (mac_m/mac_k/mac_n) is
+split by the CPU into ceil(M/mac_m)*ceil(K/mac_k)*ceil(N/mac_n)
+HW-tile-sized GemmCmds.  Each tile GemmCmd is emitted through the normal
+_emit() path, so it (a) charges PE_CPU dispatch overhead via
+_charge_dispatch (PeCpuOverheadCmd), and (b) blocks like a normal
+single-op GemmCmd on PE_GEMM at the cycle-accurate ceil-product latency.
+
+We also inject mac_m/mac_k/mac_n into every pe_gemm topology node so BOTH
+the primitive-tiled and the composite variants run on the *same*
+cycle-accurate engine (fair comparison).  Composite is left untouched:
+the CPU emits ONE CompositeCmd, and PE_SCHEDULER tiles internally (no
+per-HW-tile CPU dispatch).
+
+Data correctness:
+  Inputs are ctx.zeros (q/k/v), so every matmul result is zeros and the
+  DataExecutor replay is trivial.  To keep replay numerically correct
+  regardless, exactly ONE emitted tile per dot carries the *real* full
+  operands+output handles (so the DataExecutor computes the true (M,N)
+  result via the recorded handle shapes), while its timing fields
+  (m,k,n) are the HW-tile size so the engine charges exactly one tile of
+  cycle time.  The remaining n_tiles-1 emitted tiles are timing-only
+  GemmCmds (16x16x16) writing to throwaway scratch.  Net: n_tiles tiles
+  of engine time + n_tiles dispatch charges, and a correct final output.
+
+Engine mode: enable_data=True (same as the production sweep's
+_engine_latency_ns), op_log end-to-end latency.
+"""
+from __future__ import annotations
+
+import json
+from math import ceil
+from pathlib import Path
+
+# --- HW GEMM tile under test -------------------------------------------------
+MAC_M = 16
+MAC_K = 16
+MAC_N = 16
+# Legacy alt to also try: (8, 16, 32)
+
+S_KV_LATENCY = (8192, 32_768, 65_536, 131_072)
+
+ROOT = Path(__file__).resolve().parent
+SWEEP_JSON = (
+    ROOT / "src" / "kernbench" / "benches" / "1H_milestone_output"
+    / "gqa" / "long_ctx" / "sweep_decode_composite.json"
+)
+
+
+# --- mac-dim topology override -----------------------------------------------
+def _topo_with_mac(mac_m: int, mac_k: int, mac_n: int):
+    """Compiled topology with mac dims injected into every pe_gemm node."""
+    from kernbench.topology.builder import resolve_topology
+
+    handle = resolve_topology("topology.yaml")
+    g = handle.topology_obj
+    n = 0
+    for node in g.nodes.values():
+        if node.kind == "pe_gemm":
+            node.attrs["mac_m"] = mac_m
+            node.attrs["mac_k"] = mac_k
+            node.attrs["mac_n"] = mac_n
+            n += 1
+    print(f"    injected mac=({mac_m},{mac_k},{mac_n}) into {n} pe_gemm nodes")
+    return handle
+
+
+# --- tiling monkeypatch for TLContext.dot ------------------------------------
+def _make_tiled_dot(orig_dot, mac_m: int, mac_k: int, mac_n: int):
+    from kernbench.common.pe_commands import GemmCmd
+
+    def tiled_dot(self, a, b):
+        if len(a.shape) < 2 or len(b.shape) < 2:
+            return orig_dot(self, a, b)
+        m, k = a.shape[-2], a.shape[-1]
+        k2, n = b.shape[-2], b.shape[-1]
+        if k != k2:
+            raise ValueError(f"dot shape mismatch: a.K={k} != b.K={k2}")
+
+        n_tiles = ceil(m / mac_m) * ceil(k / mac_k) * ceil(n / mac_n)
+
+        out_shape = (*a.shape[:-2], m, n)
+        out = self._make_compute_out(shape=out_shape, dtype=a.dtype)
+        self._await_pending(a, b)
+
+        if n_tiles <= 1:
+            self._emit(GemmCmd(a=a, b=b, out=out, m=m, k=k, n=n))
+            return out
+
+        # One real-data tile: full handles (so DataExecutor computes the
+        # true result), but timing fields = HW tile (one tile of cycles).
+        self._emit(GemmCmd(a=a, b=b, out=out, m=mac_m, k=mac_k, n=mac_n))
+        # Remaining timing-only tiles: throwaway scratch, 16x16x16.
+        scratch = self._make_compute_out(shape=(mac_m, mac_n), dtype=a.dtype)
+        for _ in range(n_tiles - 1):
+            self._emit(GemmCmd(a=a, b=b, out=scratch,
+                               m=mac_m, k=mac_k, n=mac_n))
+        return out
+
+    return tiled_dot
+
+
+# --- latency runner (replicates sweep's _engine_latency_ns) ------------------
+def _engine_latency_ns(variant: str, S_kv: int, topo) -> float:
+    from kernbench.benches.gqa_helpers.long_ctx.gqa_decode_long_ctx_composite import (  # noqa: E501
+        _end_to_end_ns, _run_panel_fn,
+    )
+    from kernbench.runtime_api.bench_runner import run_bench
+    from kernbench.runtime_api.types import resolve_device
+    from kernbench.sim_engine.engine import GraphEngine
+
+    result = run_bench(
+        topology=topo, bench_fn=_run_panel_fn(variant, S_kv),
+        device=resolve_device(None),
+        engine_factory=lambda t, d: GraphEngine(
+            getattr(t, "topology_obj", t), enable_data=True,
+        ),
+    )
+    if not result.completion.ok:
+        raise RuntimeError(
+            f"{variant}@{S_kv} failed: {result.completion}"
+        )
+    return _end_to_end_ns(result.engine.op_log)
+
+
+def _emit_dispatch(variant: str, S_kv: int) -> int:
+    """PE_CPU command count at the center rank (cube 6, pe 0)."""
+    from kernbench.benches.gqa_helpers.long_ctx.gqa_decode_long_ctx_composite import (  # noqa: E501
+        _emit_dispatch as prod_emit,
+    )
+    return prod_emit(variant, S_kv)[0]
+
+
+def main() -> None:
+    import kernbench.triton_emu.tl_context as tlc
+
+    # Baseline (A): primitive UNTILED latencies from the production sweep
+    # (mac=0 / TFLOPS model). Read straight off the committed sweep JSON.
+    sweep = json.loads(SWEEP_JSON.read_text())
+    base_A = {}
+    for r in sweep["rows"]:
+        if r["variant"] == "primitive" and r["latency_ns"] is not None:
+            base_A[r["S_kv"]] = r["latency_ns"]
+
+    print(f"== mac tile = ({MAC_M},{MAC_K},{MAC_N}) ==")
+
+    # --- command-count sanity (emit-time, mac-independent) ---------------
+    orig_dot = tlc.TLContext.dot
+    print("\n[dispatch counts @ S_kv=131072]")
+    n_prim_untiled = _emit_dispatch("primitive", 131072)
+    tlc.TLContext.dot = _make_tiled_dot(orig_dot, MAC_M, MAC_K, MAC_N)
+    try:
+        n_prim_tiled = _emit_dispatch("primitive", 131072)
+        n_comp = None
+    finally:
+        tlc.TLContext.dot = orig_dot
+    n_comp = _emit_dispatch("composite", 131072)
+    print(f"  primitive UNTILED PE_CPU cmds : {n_prim_untiled}")
+    print(f"  primitive TILED   PE_CPU cmds : {n_prim_tiled}  "
+          f"(x{n_prim_tiled / max(n_prim_untiled,1):.0f})")
+    print(f"  composite         PE_CPU cmds : {n_comp}")
+
+    # --- latency sweep ----------------------------------------------------
+    rows = []
+    topo = _topo_with_mac(MAC_M, MAC_K, MAC_N)
+
+    for S_kv in S_KV_LATENCY:
+        A = base_A.get(S_kv)
+
+        # (C) composite on the mac engine (untouched dot path)
+        C = _engine_latency_ns("composite", S_kv, topo)
+
+        # (B) primitive TILED on the mac engine
+        tlc.TLContext.dot = _make_tiled_dot(orig_dot, MAC_M, MAC_K, MAC_N)
+        try:
+            B = _engine_latency_ns("primitive", S_kv, topo)
+        finally:
+            tlc.TLContext.dot = orig_dot
+
+        gap_pct = (B - C) / C * 100.0 if C else float("nan")
+        rows.append((S_kv, A, B, C, gap_pct))
+        print(f"  S_kv={S_kv:>7}: A(untiled)={A!s:>12}  "
+              f"B(tiled)={B:12.2f}  C(comp)={C:12.2f}  (B-C)/C={gap_pct:+6.1f}%")
+
+    # --- final table ------------------------------------------------------
+    print("\n==================== RESULT TABLE ====================")
+    print(f"{'S_kv':>8} | {'A untiled(ns)':>14} | {'B tiled(ns)':>14} | "
+          f"{'C comp(ns)':>14} | {'(B-C)/C':>9}")
+    print("-" * 72)
+    for S_kv, A, B, C, gap in rows:
+        a_s = f"{A:.2f}" if A is not None else "n/a"
+        print(f"{S_kv:>8} | {a_s:>14} | {B:>14.2f} | {C:>14.2f} | "
+              f"{gap:>+8.1f}%")
+
+
+if __name__ == "__main__":
+    main()

scripts/paper/measure_gqa_decode_placement_comm.py

@@ -0,0 +1,362 @@
+"""Measured comm overlay for all 6 GQA decode KV placements.
+
+Runs the simulator for Cases 1-6 (the same 6 placements the chart in
+paper_plot_gqa_4cases_summary.py covers analytically) at S_kv = 64 K,
+sums actual IPCQ-copy bytes from the engine op_log, projects to
+per-token (x80 layers), scales the partial-score-AR component of
+Cases 4/5 from S_kv = 64 K -> S_kv = 1 M (linear in S_kv; other
+cases are S_kv-independent), adds the constant Wo + FFN AR
+(1.25 MB / token), and writes the result to JSON for
+paper_plot_gqa_4cases_summary.py to overlay on the analytical bars.
+
+Single layer of decode attention only — the projection × 80 takes
+the per-layer measurement to a per-token total.
+
+Usage:
+  python scripts/paper/measure_gqa_decode_placement_comm.py
+"""
+from __future__ import annotations
+
+import json
+import os
+import sys
+from pathlib import Path
+
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_repl_pe_sp import (
+    gqa_attention_decode_long_ctx_cube_repl_pe_sp_kernel as _case3_kernel,
+)
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_repl_pe_tp import (
+    gqa_attention_decode_long_ctx_cube_repl_pe_tp_kernel as _case1_kernel,
+)
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_sp_pe_sp import (
+    gqa_attention_decode_long_ctx_cube_sp_pe_sp_kernel as _case6_kernel,
+)
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_sp_pe_tp import (
+    gqa_attention_decode_long_ctx_cube_sp_pe_tp_kernel as _case2_kernel,
+)
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_sp_pe_tp_dhead import (
+    gqa_attention_decode_long_ctx_cube_sp_pe_tp_dhead_kernel as _case4_kernel,
+)
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_tp_dhead_pe_sp import (
+    gqa_attention_decode_long_ctx_cube_tp_dhead_pe_sp_kernel as _case5_kernel,
+)
+from kernbench.benches.gqa_helpers.shared._gqa_panel_helpers import _ccl_cfg
+from kernbench.ccl.sfr_config import configure_sfr_intercube_multisip
+from kernbench.policy.placement.dp import DPPolicy
+from kernbench.runtime_api.bench_runner import run_bench
+from kernbench.runtime_api.types import resolve_device
+from kernbench.sim_engine.engine import GraphEngine
+from kernbench.topology.builder import resolve_topology
+
+_C = 8
+_P = 8
+_N_LAYERS = 80
+_S_KV_MEAS = 64 * 1024       # per-run simulator S_kv (1/16th of headline)
+_S_KV_HEADLINE = 1 << 20     # 1 Mi tokens, the chart's headline S_kv
+
+# Per-cube S_kv share for the d_head-TP partial-score AR cost.
+#   Case 4 (Cube-SP × PE-TP_dhead): per-cube = S_kv / C
+#   Case 5 (Cube-TP_dhead × PE-SP): per-cube = S_kv (KV replicated across
+#       cubes for the cube-axis d_head-TP), so partial-score AR scales
+#       with the full S_kv.
+# Headline / measured per-cube ratios give the partial-score-AR scale-up
+# from S_kv = 64 K to S_kv = 1 M. (m,ℓ,O) AR is S_kv-independent.
+_PARTIAL_SCORE_SCALE = _S_KV_HEADLINE / _S_KV_MEAS   # = 16
+
+# Per-token Wo + FFN AR (constant across all cases, comes from the
+# attn-output and FFN-down all-reduces NOT measured by the attention-
+# only kernel run here).
+_WO_PER_LAYER_BYTES = 8 * 1024
+_FFN_PER_LAYER_BYTES = 8 * 1024
+_WO_FFN_PER_TOKEN_BYTES = (
+    (_WO_PER_LAYER_BYTES + _FFN_PER_LAYER_BYTES) * _N_LAYERS
+)   # 1.25 MB
+
+# Total PE count in one KV-head group — average per-PE comm = total / N.
+_NUM_PES = _C * _P
+
+# Total partial-score-AR slice produced by the attention compute when
+# d_head is sharded. Used to split measured IPCQ traffic into the
+# S_kv-scaling component (partial scores) vs the S_kv-independent
+# component ((m,ℓ,O) merge). Same as the analytical formula in
+# paper_plot_gqa_4cases_summary.py: h_q · S_q · per_cube_S_kv · 2 bytes.
+_H_Q = 8
+_S_Q = 1
+_BYTES_PER_ELEM = 2
+
+_PARAMS = dict(C=_C, P=_P, T_q=_S_Q, S_kv=_S_KV_MEAS,
+               d_head=128, h_q=_H_Q, h_kv=1)
+
+
+def _bench_fn_case1(ctx):
+    """Case 1: Cube-Repl x PE-repl (PE-TP doesn't shard KV)."""
+    p = _PARAMS
+    configure_sfr_intercube_multisip(ctx.engine, ctx.spec, _ccl_cfg())
+    dp = DPPolicy(cube="replicate", pe="replicate",
+                  num_cubes=p["C"], num_pes=p["P"])
+    q = ctx.zeros((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp, name="q_c1")
+    k = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp, name="k_c1")
+    v = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp, name="v_c1")
+    o = ctx.empty((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp, name="o_c1")
+    ctx.launch("case1_repl_repl", _case1_kernel,
+               q, k, v, o,
+               p["T_q"], p["S_kv"], p["h_q"], p["h_kv"],
+               p["d_head"], p["C"], p["P"],
+               _auto_dim_remap=False)
+
+
+def _bench_fn_case2(ctx):
+    """Case 2: Cube-SP x PE-repl (PE-TP doesn't shard KV)."""
+    p = _PARAMS
+    configure_sfr_intercube_multisip(ctx.engine, ctx.spec, _ccl_cfg())
+    dp_full = DPPolicy(cube="replicate", pe="replicate",
+                       num_cubes=p["C"], num_pes=p["P"])
+    dp_kv = DPPolicy(cube="row_wise", pe="replicate",
+                     num_cubes=p["C"], num_pes=p["P"])
+    q = ctx.zeros((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp_full, name="q_c2")
+    k = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp_kv, name="k_c2")
+    v = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp_kv, name="v_c2")
+    o = ctx.empty((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp_full, name="o_c2")
+    ctx.launch("case2_sp_repl", _case2_kernel,
+               q, k, v, o,
+               p["T_q"], p["S_kv"], p["h_q"], p["h_kv"],
+               p["d_head"], p["C"], p["P"],
+               _auto_dim_remap=False)
+
+
+def _bench_fn_case3(ctx):
+    """Case 3: Cube-Repl x PE-SP."""
+    p = _PARAMS
+    configure_sfr_intercube_multisip(ctx.engine, ctx.spec, _ccl_cfg())
+    dp_full = DPPolicy(cube="replicate", pe="replicate",
+                       num_cubes=p["C"], num_pes=p["P"])
+    dp_kv = DPPolicy(cube="replicate", pe="row_wise",
+                     num_cubes=p["C"], num_pes=p["P"])
+    q = ctx.zeros((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp_full, name="q_c3")
+    k = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp_kv, name="k_c3")
+    v = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp_kv, name="v_c3")
+    o = ctx.empty((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp_full, name="o_c3")
+    ctx.launch("case3_repl_sp", _case3_kernel,
+               q, k, v, o,
+               p["T_q"], p["S_kv"], p["h_q"], p["h_kv"],
+               p["d_head"], p["C"], p["P"],
+               _auto_dim_remap=False)
+
+
+def _bench_fn_case4(ctx):
+    p = _PARAMS
+    configure_sfr_intercube_multisip(ctx.engine, ctx.spec, _ccl_cfg())
+    dp_full = DPPolicy(cube="replicate", pe="column_wise",
+                       num_cubes=p["C"], num_pes=p["P"])
+    dp_kv = DPPolicy(cube="row_wise", pe="column_wise",
+                     num_cubes=p["C"], num_pes=p["P"])
+    q = ctx.zeros((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp_full, name="q_c4")
+    k = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp_kv, name="k_c4")
+    v = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp_kv, name="v_c4")
+    o = ctx.empty((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp_full, name="o_c4")
+    ctx.launch("case4_dhead_tp", _case4_kernel,
+               q, k, v, o,
+               p["T_q"], p["S_kv"], p["h_q"], p["h_kv"],
+               p["d_head"], p["C"], p["P"],
+               _auto_dim_remap=False)
+
+
+def _bench_fn_case5(ctx):
+    p = _PARAMS
+    configure_sfr_intercube_multisip(ctx.engine, ctx.spec, _ccl_cfg())
+    dp_q = DPPolicy(cube="column_wise", pe="replicate",
+                    num_cubes=p["C"], num_pes=p["P"])
+    dp_kv = DPPolicy(cube="column_wise", pe="row_wise",
+                     num_cubes=p["C"], num_pes=p["P"])
+    q = ctx.zeros((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp_q, name="q_c5")
+    k = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp_kv, name="k_c5")
+    v = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp_kv, name="v_c5")
+    o = ctx.empty((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp_q, name="o_c5")
+    ctx.launch("case5_dhead_tp_inter", _case5_kernel,
+               q, k, v, o,
+               p["T_q"], p["S_kv"], p["h_q"], p["h_kv"],
+               p["d_head"], p["C"], p["P"],
+               _auto_dim_remap=False)
+
+
+def _bench_fn_case6(ctx):
+    p = _PARAMS
+    configure_sfr_intercube_multisip(ctx.engine, ctx.spec, _ccl_cfg())
+    dp_full = DPPolicy(cube="replicate", pe="replicate",
+                       num_cubes=p["C"], num_pes=p["P"])
+    dp_kv = DPPolicy(cube="row_wise", pe="row_wise",
+                     num_cubes=p["C"], num_pes=p["P"])
+    q = ctx.zeros((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp_full, name="q_c6")
+    k = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp_kv, name="k_c6")
+    v = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp_kv, name="v_c6")
+    o = ctx.empty((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp_full, name="o_c6")
+    ctx.launch("case6_sp_sp", _case6_kernel,
+               q, k, v, o,
+               p["T_q"], p["S_kv"], p["h_q"], p["h_kv"],
+               p["d_head"], p["C"], p["P"],
+               _auto_dim_remap=False)
+
+
+def _sum_ipcq_bytes(op_log) -> int:
+    """Sum nbytes across all ipcq_copy records."""
+    return sum(
+        r.params.get("nbytes", 0)
+        for r in op_log
+        if r.op_kind == "memory" and r.op_name == "ipcq_copy"
+    )
+
+
+def _partial_score_slices(case: int) -> int:
+    """Divisor that splits S_kv into partial-score tiles, per the
+    analytical model in paper_plot_gqa_4cases_summary.py:
+
+      partial_score_per_PE = h_q * S_q * (s_kv / slices) * bytes
+
+    Case 4 (Cube-SP x PE-TP-dhead): intra-cube AR over d_head shards
+        on PE axis -> partial tile per PE has per-cube S_kv = s_kv/C.
+    Case 5 (Cube-TP-dhead x PE-SP): inter-cube AR over d_head shards
+        on cube axis -> partial tile per PE has per-PE S_kv = s_kv/P.
+    Cases 1, 2, 3, 6: no partial-score AR (only (m,l,O) merge).
+    """
+    if case == 4:
+        return _C
+    if case == 5:
+        return _P
+    return 0
+
+
+def _split_attn_layer_bytes(case: int, total_ipcq_bytes: int,
+                            s_kv: int) -> tuple[int, int]:
+    """Split per-layer attention-time IPCQ bytes into:
+      (partial_score_component, mlo_component).
+
+    The partial-score component scales with s_kv (so it must be scaled
+    when projecting from the measure-time s_kv to the headline s_kv);
+    the (m,l,O) component is constant in s_kv.
+
+    Partial-score size is analytically known per-case (formula in
+    _partial_score_slices); the remainder is treated as (m,l,O) + any
+    other S_kv-independent overhead. Per-PE = total / NUM_PES.
+    """
+    per_pe_total = total_ipcq_bytes // _NUM_PES
+    slices = _partial_score_slices(case)
+    if slices == 0:
+        # No partial-score AR for this case.
+        return 0, per_pe_total
+    partial_score_per_pe = (
+        _H_Q * _S_Q * (s_kv // slices) * _BYTES_PER_ELEM
+    )
+    partial_score_per_pe = min(partial_score_per_pe, per_pe_total)
+    mlo_per_pe = per_pe_total - partial_score_per_pe
+    return partial_score_per_pe, mlo_per_pe
+
+
+_KERNELS = (
+    (1, "Case 1 (Cube-Repl x PE-repl)",      _bench_fn_case1),
+    (2, "Case 2 (Cube-SP   x PE-repl)",      _bench_fn_case2),
+    (3, "Case 3 (Cube-Repl x PE-SP)",        _bench_fn_case3),
+    (4, "Case 4 (Cube-SP   x PE-TP d_head)", _bench_fn_case4),
+    (5, "Case 5 (Cube-TP d_head x PE-SP)",   _bench_fn_case5),
+    (6, "Case 6 (Cube-SP   x PE-SP)  [*]",   _bench_fn_case6),
+)
+
+
+def main() -> int:
+    topology = os.environ.get("GQA_1H_TOPOLOGY", "topology.yaml")
+    topo = resolve_topology(topology)
+
+    out: dict = {
+        "S_kv_measured": _S_KV_MEAS,
+        "S_kv_headline": _S_KV_HEADLINE,
+        "n_layers": _N_LAYERS,
+        "num_pes": _NUM_PES,
+        "wo_ffn_per_token_bytes": _WO_FFN_PER_TOKEN_BYTES,
+        "cases": {},
+    }
+
+    print(f"Measuring at S_kv={_S_KV_MEAS:,} ; scaling partial-score AR "
+          f"to S_kv={_S_KV_HEADLINE:,} (×{int(_PARTIAL_SCORE_SCALE)})")
+    print()
+
+    for case_id, label, bench_fn in _KERNELS:
+        try:
+            res = run_bench(
+                topology=topo, bench_fn=bench_fn,
+                device=resolve_device(None),
+                engine_factory=lambda t, d: GraphEngine(
+                    getattr(t, "topology_obj", t), enable_data=True,
+                ),
+            )
+        except Exception as e:
+            print(f"  {label:<42} FAIL: {type(e).__name__}: {e}")
+            return 1
+        if not res.completion.ok:
+            print(f"  {label:<42} ENGINE FAIL: {res.completion}")
+            return 1
+
+        total_ipcq = _sum_ipcq_bytes(res.engine.op_log)
+        partial_pe, mlo_pe = _split_attn_layer_bytes(
+            case_id, total_ipcq, _S_KV_MEAS,
+        )
+        # Per-token attention-time comm at S_kv = 1 M:
+        #   (partial_score_per_layer × scale  +  mlo_per_layer) × 80 layers
+        scaled_partial_per_token = (
+            partial_pe * int(_PARTIAL_SCORE_SCALE) * _N_LAYERS
+        )
+        mlo_per_token = mlo_pe * _N_LAYERS
+        attn_per_token = scaled_partial_per_token + mlo_per_token
+        total_per_token = attn_per_token + _WO_FFN_PER_TOKEN_BYTES
+
+        out["cases"][str(case_id)] = {
+            "label": label,
+            "total_ipcq_bytes_one_layer": total_ipcq,
+            "per_pe_partial_score_bytes_one_layer": partial_pe,
+            "per_pe_mlo_bytes_one_layer": mlo_pe,
+            "per_pe_attn_bytes_per_token_at_1M": attn_per_token,
+            "per_pe_total_bytes_per_token_at_1M": total_per_token,
+        }
+
+        print(f"  {label:<42} "
+              f"ipcq_total={total_ipcq:>10,}  "
+              f"per_pe_attn(1L)={(partial_pe + mlo_pe):>9,}  "
+              f"per_pe_total/tok@1M={total_per_token / (1<<20):>7.2f} MB")
+
+    out_path = (
+        Path(__file__).resolve().parents[2]
+        / "src" / "kernbench" / "benches"
+        / "1H_milestone_output" / "gqa" / "long_ctx"
+        / "gqa_long_ctx_6cases_measured_comm.json"
+    )
+    out_path.parent.mkdir(parents=True, exist_ok=True)
+    out_path.write_text(json.dumps(out, indent=2))
+    print()
+    print(f"wrote {out_path}")
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

scripts/paper/paper_plot_gqa_4cases_summary.py

@@ -114,6 +114,8 @@ Output PNG:
 """
 from __future__ import annotations
 
+import json
+import textwrap
 from pathlib import Path
 
 import matplotlib.patches as mpatches
@@ -186,9 +188,36 @@ _WV_GB = _WV_MB_PER_LAYER * _N_LAYERS / 1000    # 0.16 GB
 _WO_GB = _WO_MB_PER_LAYER * _N_LAYERS / 1000    # 0.16 GB
 _WEIGHTS_GB = _WQ_GB + _WK_GB + _WV_GB + _WO_GB  # 1.76 GB
 
-# (m, ℓ, O) AR payload — used by Cases 1, 3, 4 (different axes per case).
+# (m, ℓ, O) AR cost per layer, derived from the kernel topology.
-_MLO_INTRA_BYTES_PER_LAYER = 32 * 1024   # ~32 KB intra-cube AR
+#
-_MLO_INTER_BYTES_PER_LAYER = 32 * 1024   # ~32 KB inter-cube AR
+# Per-PE payload  T  for one (m, ℓ, O) merge step:
+#     O  — h_q · S_q · d_head · 2 bytes  (FP16)
+#     m  — h_q · S_q · 4 bytes           (FP32)
+#     ℓ  — h_q · S_q · 4 bytes           (FP32)
+#     T  = h_q · S_q · (d_head · 2 + 8)  ≈ 2.1 KB
+#
+# Reduce algorithm: the decode kernels use hierarchical reduce-only
+# (chain / tree) rather than ring all-reduce, because the merged result
+# only needs to land on the cube that runs the downstream Wo gemm —
+# not on every PE.  For a chain of N participants the total traffic is
+# (N-1)·T and the per-PE average is (N-1)/N · T.
+#
+# The previous version used a hard-coded 32 KB / layer placeholder which
+# overestimated the per-PE cost by ~6× (and ~3× even under a hypothetical
+# ring-AR assumption).  See `gqa_long_ctx_6cases_measured_comm.json` for
+# the measured numbers this matches against.
+_MLO_PAYLOAD_BYTES = _H_Q * _S_Q * (_D_HEAD * _BYTES_PER_ELEM + 8)
+
+
+def _reduce_chain_per_pe_bytes(n_participants: int) -> int:
+    """Per-PE average bytes for a single-stage chain/tree reduce of T."""
+    if n_participants <= 1:
+        return 0
+    return _MLO_PAYLOAD_BYTES * (n_participants - 1) // n_participants
+
+
+_MLO_INTRA_BYTES_PER_LAYER = _reduce_chain_per_pe_bytes(_P)  # PE-axis
+_MLO_INTER_BYTES_PER_LAYER = _reduce_chain_per_pe_bytes(_C)  # cube-axis
 
 # Cases — renumbered in MEMORY-DESCENDING ORDER (left to right):
 #   Case 1   (40 GB) : no sharding
@@ -224,11 +253,11 @@ _CASE_COLOR = {
 }
 _ATTN_DESC = {
     1: "none",
-    2: "online-softmax\nmerge of\n(m,ℓ,O)\ninter-cube",
+    2: "online-softmax (m,ℓ,O) — inter-cube",
-    3: "online-softmax\nmerge of\n(m,ℓ,O)\nintra-cube",
+    3: "online-softmax (m,ℓ,O) — intra-cube",
-    4: "partial\nscores\n+ (m,ℓ,O)\nmerge\n(d_head-TP)",
+    4: "partial scores + (m,ℓ,O) merge  (d_head-TP)",
-    5: "partial\nscores\n+ (m,ℓ,O)\nmerge\n(d_head-TP)",
+    5: "partial scores + (m,ℓ,O) merge  (d_head-TP)",
-    6: "online-softmax\nmerge of\n(m,ℓ,O)\nintra + inter",
+    6: "online-softmax (m,ℓ,O) — intra + inter",
 }
 
 _WO_COLOR   = "#9EC5E8"
@@ -240,6 +269,22 @@ _OUT_DIR = (
     / "src" / "kernbench" / "benches"
     / "1H_milestone_output" / "gqa" / "long_ctx"
 )
+_MEASURED_JSON = _OUT_DIR / "gqa_long_ctx_6cases_measured_comm.json"
+
+
+def _load_measured() -> dict[int, float] | None:
+    """Load measured per-PE comm bytes (already scaled to S_kv=1M).
+
+    Returns {case_id: per_token_total_bytes} or None if JSON missing.
+    Produced by scripts/paper/measure_gqa_decode_placement_comm.py.
+    """
+    if not _MEASURED_JSON.exists():
+        return None
+    data = json.loads(_MEASURED_JSON.read_text())
+    return {
+        int(cid): info["per_pe_total_bytes_per_token_at_1M"]
+        for cid, info in data["cases"].items()
+    }
 
 
 # ── Formulae ────────────────────────────────────────────────────────
@@ -388,51 +433,118 @@ def _plot_memory(ax) -> None:
     ax.legend(loc="upper right", fontsize=9)
 
 
-def _plot_comm(ax) -> None:
+def _plot_comm(ax, *, mode: str = "analytical") -> None:
-    wo_mb = []
+    """Per-PE comm panel.
-    ffn_mb = []
+
-    attn_mb = []
+    mode = "analytical": single solid bars from per_token_bytes formula.
+    mode = "paired"    : analytical (solid) + simulator-measured
+                         (hatched) side-by-side per case, when the
+                         measurement JSON is available.
+    """
+    labels = [_CASE_LABEL[c] for c in _CASES]
+
+    wo_mb_list: list[float] = []
+    ffn_mb_list: list[float] = []
+    attn_mb: list[float] = []
     for c in _CASES:
         wo, ffn, attn = per_token_bytes(c, _HEADLINE_S_KV)
-        wo_mb.append(wo / (1 << 20))
+        wo_mb_list.append(wo / (1 << 20))
-        ffn_mb.append(ffn / (1 << 20))
+        ffn_mb_list.append(ffn / (1 << 20))
         attn_mb.append(attn / (1 << 20))
 
-    labels = [_CASE_LABEL[c] for c in _CASES]
+    measured = _load_measured() if mode == "paired" else None
-    x = list(range(len(_CASES)))
+    paired = measured is not None
+    source_tag = "analytical (solid) vs simulator-measured (hatched)" \
+                 if paired else "analytical"
 
-    ax.bar(x, wo_mb, color=_WO_COLOR, edgecolor="black")
+    n_cases = len(_CASES)
-    ax.bar(x, ffn_mb, bottom=wo_mb, color=_FFN_COLOR, edgecolor="black")
+    x = list(range(n_cases))
-    bottoms_attn = [w + f for w, f in zip(wo_mb, ffn_mb)]
+    bar_w = 0.36 if paired else 0.65
-    ax.bar(x, attn_mb, bottom=bottoms_attn,
+    x_ana  = [xi - bar_w / 2 for xi in x] if paired else x
+    x_meas = [xi + bar_w / 2 for xi in x] if paired else None
+
+    # Analytical bars (solid).
+    ax.bar(x_ana, wo_mb_list,  width=bar_w,
+           color=_WO_COLOR,  edgecolor="black")
+    ax.bar(x_ana, ffn_mb_list, width=bar_w, bottom=wo_mb_list,
+           color=_FFN_COLOR, edgecolor="black")
+    bottoms_attn = [w + f for w, f in zip(wo_mb_list, ffn_mb_list)]
+    ax.bar(x_ana, attn_mb, width=bar_w, bottom=bottoms_attn,
            color=_ATTN_COLOR, edgecolor="black")
 
+    # Measured bars (hatched) — same Wo+FFN base, attn from op_log.
+    meas_attn_mb: list[float] = []
+    if paired:
+        for i, c in enumerate(_CASES):
+            meas_total = measured.get(c, 0) / (1 << 20)
+            meas_attn_mb.append(
+                max(meas_total - wo_mb_list[i] - ffn_mb_list[i], 0.0))
+        ax.bar(x_meas, wo_mb_list,  width=bar_w,
+               color=_WO_COLOR,  edgecolor="black",
+               hatch="///", alpha=0.85)
+        ax.bar(x_meas, ffn_mb_list, width=bar_w, bottom=wo_mb_list,
+               color=_FFN_COLOR, edgecolor="black",
+               hatch="///", alpha=0.85)
+        ax.bar(x_meas, meas_attn_mb, width=bar_w, bottom=bottoms_attn,
+               color=_ATTN_COLOR, edgecolor="black",
+               hatch="///", alpha=0.85)
+
     ax.set_xticks(x)
     ax.set_xticklabels(labels, fontsize=9)
     ax.set_ylabel("Comm per token per PE (MB, log)")
     ax.set_yscale("log")
-    ax.set_title(
+    title = (
         f"Per-PE communication at S_kv = {_HEADLINE_S_KV:,} tokens "
-        f"(decode S_q=1, B=1; {_N_LAYERS} layers)",
+        f"(decode S_q=1, B=1; {_N_LAYERS} layers)  —  {source_tag}"
-        fontsize=11,
     )
+    if paired:
+        title += "\n(simulator measured at S_kv = 8K; " \
+                 "partial-score AR scaled ×128 to S_kv = 1M)"
+    ax.set_title(title, fontsize=10)
     ax.grid(axis="y", ls=":", alpha=0.5, which="both")
 
-    totals = [wo_mb[i] + ffn_mb[i] + attn_mb[i] for i in range(len(_CASES))]
+    totals_ana = [wo_mb_list[i] + ffn_mb_list[i] + attn_mb[i]
-    ax.set_ylim(top=max(totals) * 10)
+                  for i in range(n_cases)]
+    ymax = max(totals_ana)
+    if paired:
+        ymax = max(ymax, max(
+            (measured.get(c, 0) / (1 << 20)) for c in _CASES))
+    ax.set_ylim(top=ymax * 22)
 
     for i, c in enumerate(_CASES):
-        total_bytes = (wo_mb[i] + ffn_mb[i] + attn_mb[i]) * (1 << 20)
+        ana_bytes = totals_ana[i] * (1 << 20)
-        ax.text(x[i], totals[i] * 1.5, _fmt_bytes(total_bytes),
+        if paired:
-                ha="center", fontsize=9, weight="bold")
+            meas_bytes = measured.get(c, 0)
+            top_y = max(totals_ana[i], meas_bytes / (1 << 20))
+            label = (f"ana: {_fmt_bytes(ana_bytes)}\n"
+                     f"sim: {_fmt_bytes(meas_bytes)}")
+        else:
+            top_y = totals_ana[i]
+            label = _fmt_bytes(ana_bytes)
+        ax.text(x[i], top_y * 1.5, label,
+                ha="center", va="bottom",
+                fontsize=8 if paired else 9,
+                weight="bold", linespacing=1.05)
+
+        # Attention-time AR descriptor — placed at the attention-segment
+        # midpoint, centered between the analytical and measured bars so
+        # it visually spans both. Coloured the same as the attention bar
+        # (no background box) so it lives within the case's bar zone.
+        # Wrapped narrow so each line fits inside the bar-pair width.
+        wrapped = textwrap.fill(_ATTN_DESC[c], width=14)
         if attn_mb[i] > 0:
             mid = bottoms_attn[i] + attn_mb[i] / 2
-            ax.text(x[i], mid, _ATTN_DESC[c],
+            ax.text(x[i], mid, wrapped,
-                    ha="center", va="center", fontsize=7,
+                    ha="center", va="center",
-                    color="black", weight="bold")
+                    fontsize=7 if paired else 8,
+                    color="black", weight="bold",
+                    linespacing=1.0)
         else:
-            ax.text(x[i], totals[i] * 0.4, _ATTN_DESC[c],
+            ax.text(x[i], totals_ana[i] * 0.4, wrapped,
-                    ha="center", fontsize=7.5, color="grey", style="italic")
+                    ha="center",
+                    fontsize=7 if paired else 8,
+                    color="grey", style="italic",
+                    linespacing=1.0)
 
     legend_handles = [
         mpatches.Patch(facecolor=_WO_COLOR, edgecolor="black",
@@ -442,40 +554,67 @@ def _plot_comm(ax) -> None:
         mpatches.Patch(facecolor=_ATTN_COLOR, edgecolor="black",
                        label="Attn-time collective"),
     ]
+    if paired:
+        legend_handles.append(
+            mpatches.Patch(facecolor="white", edgecolor="black",
+                           hatch="///", label="simulator-measured"))
     ax.legend(handles=legend_handles, loc="upper right", fontsize=8,
               framealpha=0.92)
 
 
 def main() -> Path:
     _OUT_DIR.mkdir(parents=True, exist_ok=True)
-    fig = plt.figure(figsize=(21.0, 6.0))
+
-    gs = fig.add_gridspec(1, 3, width_ratios=[0.7, 1.6, 1.6], wspace=0.22)
+    # (a) Per-PE HBM budget — standalone PNG.
-    ax_b = fig.add_subplot(gs[0, 0])
+    fig_b, ax_b = plt.subplots(figsize=(4.0, 6.0))
-    ax_m = fig.add_subplot(gs[0, 1])
-    ax_c = fig.add_subplot(gs[0, 2])
     _plot_budget(ax_b)
+    fig_b.tight_layout()
+    out_b = _OUT_DIR / "gqa_long_ctx_6cases_hbm_budget.png"
+    fig_b.savefig(out_b, dpi=150)
+    plt.close(fig_b)
+    print(f"wrote {out_b}")
+
+    # (b) Combined 3-panel summary — HBM budget + KV memory + comm.
+    fig = plt.figure(figsize=(22.0, 6.5))
+    gs = fig.add_gridspec(1, 3, width_ratios=[0.7, 1.6, 1.6], wspace=0.22)
+    ax_b2 = fig.add_subplot(gs[0, 0])
+    ax_m  = fig.add_subplot(gs[0, 1])
+    ax_c  = fig.add_subplot(gs[0, 2])
+    _plot_budget(ax_b2)
     _plot_memory(ax_m)
     _plot_comm(ax_c)
-
+    fig.tight_layout()
-    fig.suptitle(
+    out = _OUT_DIR / "gqa_long_ctx_6cases_summary.png"
-        f"GQA per-PE memory + communication  —  LLaMA-3.1-70B "
-        f"single-KV-head group  (C={_C}, P={_P}, {_N_LAYERS} layers, "
-        f"S_kv = 1 M, FP16)",
-        fontsize=12, y=0.985,
-    )
-    fig.text(
-        0.5, 0.94,
-        "Cases ordered by memory  ·  Case 1 (40 GB, no sharding)  →  "
-        "Cases 2-3 (5 GB, 1-axis sharded)  →  Cases 4-6 (640 MB, 2-axis sharded)  "
-        "·  Case 6 ★ = lowest comm among memory-feasible cases",
-        ha="center", fontsize=9.5, color="#444",
-    )
-    fig.tight_layout(rect=(0, 0, 1, 0.92))
-    out = _OUT_DIR / "gqa_4cases_summary.png"
     fig.savefig(out, dpi=150)
     plt.close(fig)
     print(f"wrote {out}")
 
+    # (c) 2-panel companion (analytical only).
+    fig2 = plt.figure(figsize=(18.0, 6.5))
+    gs2 = fig2.add_gridspec(1, 2, width_ratios=[1.0, 1.0], wspace=0.18)
+    ax_m2 = fig2.add_subplot(gs2[0, 0])
+    ax_c2 = fig2.add_subplot(gs2[0, 1])
+    _plot_memory(ax_m2)
+    _plot_comm(ax_c2, mode="analytical")
+    fig2.tight_layout()
+    out2 = _OUT_DIR / "gqa_long_ctx_6cases_memory_comm_analytical.png"
+    fig2.savefig(out2, dpi=150)
+    plt.close(fig2)
+    print(f"wrote {out2}")
+
+    # (d) 2-panel companion — analytical vs simulator-measured paired.
+    fig3 = plt.figure(figsize=(19.0, 6.5))
+    gs3 = fig3.add_gridspec(1, 2, width_ratios=[1.0, 1.0], wspace=0.18)
+    ax_m3 = fig3.add_subplot(gs3[0, 0])
+    ax_c3 = fig3.add_subplot(gs3[0, 1])
+    _plot_memory(ax_m3)
+    _plot_comm(ax_c3, mode="paired")
+    fig3.tight_layout()
+    out3 = _OUT_DIR / "gqa_long_ctx_6cases_memory_comm_paired.png"
+    fig3.savefig(out3, dpi=150)
+    plt.close(fig3)
+    print(f"wrote {out3}")
+
     # Paper-ready table to stdout.
     print()
     print(f"  {'Case':<27} {'KV/tok·PE':>12} {'KV @ 1M':>11} "

scripts/paper/paper_plot_gqa_decode_long_ctx_4cases.py

@@ -1,13 +1,25 @@
 """Comparative figures for milestone-gqa-decode-long-ctx-4cases.
 
-Reads sweep.json (emitted by ``kernbench run --bench
+Reads sweep_decode.json (emitted by the milestone-1h-gqa bench) and
-milestone-gqa-decode-long-ctx-4cases``) and writes four PNGs into
+writes four PNGs into the same bench-output dir
-``docs/report/1H-codesign-paper/figures/``:
+(src/kernbench/benches/1H_milestone_output/gqa/long_ctx/):
 
-  gqa_decode_long_ctx_4cases_latency.png       end-to-end latency per case
+  gqa_decode_long_ctx_6cases_latency.png       end-to-end latency per case
-  gqa_decode_long_ctx_4cases_traffic.png       ipcq/dma op-count breakdown
+  gqa_decode_long_ctx_6cases_traffic.png       ipcq/dma op-count breakdown
-  gqa_decode_long_ctx_4cases_memory.png        per-PE KV bytes per case
+  gqa_decode_long_ctx_6cases_memory.png        per-PE KV bytes per case
-  gqa_decode_long_ctx_4cases_parallelism.png   per-PE S_local (compute work)
+  gqa_decode_long_ctx_6cases_parallelism.png   per-PE S_local (compute work)
+
+Filename still says "4cases" for backwards compat, but the script now
+covers all SIX kv-sharding strategies from the analytical chart
+(`gqa_4cases_summary.png`) — the original 4 plus the two new
+d_head-TP variants:
+
+  Case 1   Cube-Repl × PE-repl      (PE-TP doesn't shard KV)
+  Case 2   Cube-SP   × PE-repl
+  Case 3   Cube-Repl × PE-SP
+  Case 4   Cube-SP   × PE-TP-d_head      ← NEW
+  Case 5   Cube-TP-d_head × PE-SP        ← NEW
+  Case 6 ★ Cube-SP   × PE-SP             (Pareto-best)
 
 Run (after the bench):
     GQA_DECODE_LONG_CTX_4CASES_RUN=1 python -m kernbench.cli.main run \\
@@ -32,16 +44,27 @@ _FIG_DIR = (
 )
 _SWEEP_JSON = _FIG_DIR / "sweep_decode.json"
 
-# Panel name → (short label, case ordinal for left-to-right plot order).
+# Panel name → (short label, case ordinal, accent flag) using the
+# analytical chart's memory-descending ordering. PE-TP doesn't shard
+# KV memory, so the cube_repl_pe_tp panel maps to Case 1 (no
+# sharding, KV-wise) and cube_sp_pe_tp panel maps to Case 2.
+_NORMAL, _OVERFLOW, _PARETO = "normal", "overflow", "pareto"
+
 _CASE_INFO = {
-    "single_kv_group_decode_long_ctx_gqa_cube_sp_pe_tp": (
+    # panel name                                                       label                            ord  flag
-        "Case 1\nCube-SP × PE-TP", 1),
+    "single_kv_group_decode_long_ctx_gqa_cube_repl_pe_tp":            ("Case 1\nCube-Repl × PE-repl",   1, _OVERFLOW),
-    "single_kv_group_decode_long_ctx_gqa_cube_repl_pe_tp": (
+    "single_kv_group_decode_long_ctx_gqa_cube_sp_pe_tp":              ("Case 2\nCube-SP × PE-repl",     2, _OVERFLOW),
-        "Case 2\nCube-Repl × PE-TP", 2),
+    "single_kv_group_decode_long_ctx_gqa_cube_repl_pe_sp":            ("Case 3\nCube-Repl × PE-SP",     3, _OVERFLOW),
-    "single_kv_group_decode_long_ctx_gqa_cube_repl_pe_sp": (
+    "single_kv_group_decode_long_ctx_gqa_cube_sp_pe_tp_dhead":        ("Case 4\nCube-SP × PE-TP-d_head", 4, _NORMAL),
-        "Case 3\nCube-Repl × PE-SP", 3),
+    "single_kv_group_decode_long_ctx_gqa_cube_tp_dhead_pe_sp":        ("Case 5\nCube-TP-d_head × PE-SP", 5, _NORMAL),
-    "single_kv_group_decode_long_ctx_gqa_cube_sp_pe_sp": (
+    "single_kv_group_decode_long_ctx_gqa_cube_sp_pe_sp":              ("Case 6 ★\nCube-SP × PE-SP",     6, _PARETO),
-        "Case 4 ★\nCube-SP × PE-SP", 4),
+}
+
+# Bar fill colour per flag (used by every panel).
+_FLAG_COLOR = {
+    _NORMAL:   "#888888",     # neutral grey
+    _OVERFLOW: "#c0504d",     # red — fails the per-PE HBM budget
+    _PARETO:   "#3b6ea5",     # blue — Pareto-best
 }
 
 
@@ -53,23 +76,26 @@ def _sorted_by_case(rows: list[dict]) -> list[dict]:
     return sorted(rows, key=lambda r: _CASE_INFO[r["panel"]][1])
 
 
+def _bar_colors(rows: list[dict]) -> list[str]:
+    return [_FLAG_COLOR[_CASE_INFO[r["panel"]][2]] for r in rows]
+
+
 def _plot_latency(rows: list[dict]) -> Path:
     rows = _sorted_by_case(rows)
     labels = [_CASE_INFO[r["panel"]][0] for r in rows]
     lat_us = [r["latency_ns"] / 1e3 for r in rows]
-    colors = ["#888", "#888", "#888", "#3b6ea5"]  # Case 4 highlighted
+    fig, ax = plt.subplots(figsize=(12.0, 4.8))
-    fig, ax = plt.subplots(figsize=(8.0, 4.5))
+    bars = ax.bar(labels, lat_us, color=_bar_colors(rows), width=0.6)
-    bars = ax.bar(labels, lat_us, color=colors, width=0.6)
     ax.set_ylabel("end-to-end latency (µs)")
     ax.set_title(
-        "Long-context decode 4-cases — end-to-end latency per case\n"
+        "Long-context decode 6-cases — end-to-end latency per case\n"
         "LLaMA-3.1-70B single-KV-head group (8 cubes × 8 PEs)"
     )
     ax.bar_label(bars, fmt="%.1f", padding=3, fontsize=9)
     ax.grid(axis="y", ls=":", alpha=0.5)
     ax.set_ylim(0, max(lat_us) * 1.15)
     fig.tight_layout()
-    out = _FIG_DIR / "gqa_decode_long_ctx_4cases_latency.png"
+    out = _FIG_DIR / "gqa_decode_long_ctx_6cases_latency.png"
     fig.savefig(out, dpi=150)
     plt.close(fig)
     return out
@@ -83,18 +109,18 @@ def _plot_traffic(rows: list[dict]) -> Path:
     disp = ["IPCQ copy", "DMA read", "DMA write"]
     colors = ["#c0504d", "#9bbb59", "#8064a2"]
     w = 0.25
-    fig, ax = plt.subplots(figsize=(9.0, 4.5))
+    fig, ax = plt.subplots(figsize=(11.0, 4.5))
     for i, (k, d, c) in enumerate(zip(keys, disp, colors)):
         vals = [r["op_log_summary"][k] for r in rows]
         ax.bar([xi + (i - 1) * w for xi in x], vals, width=w, label=d, color=c)
     ax.set_xticks(list(x))
     ax.set_xticklabels(labels, fontsize=9)
     ax.set_ylabel("op count")
-    ax.set_title("Long-context decode 4-cases — op-count breakdown per case")
+    ax.set_title("Long-context decode 6-cases — op-count breakdown per case")
     ax.legend(fontsize=9)
     ax.grid(axis="y", ls=":", alpha=0.5)
     fig.tight_layout()
-    out = _FIG_DIR / "gqa_decode_long_ctx_4cases_traffic.png"
+    out = _FIG_DIR / "gqa_decode_long_ctx_6cases_traffic.png"
     fig.savefig(out, dpi=150)
     plt.close(fig)
     return out
@@ -103,27 +129,41 @@ def _plot_traffic(rows: list[dict]) -> Path:
 def _s_local_per_pe(panel: str, *, S_kv: int, C: int, P: int) -> int:
     """S_local (token count) each PE attends over locally.
 
-    Encodes the cube/pe sharding axes from the panel name:
+      cube_repl_pe_tp  (Case 1): S_kv        (no sharding, KV-wise)
-      cube_sp_pe_tp (Case 1): S_kv / C       (pe=replicate within cube)
+      cube_sp_pe_tp    (Case 2): S_kv / C    (cube splits S_kv, PEs replicate)
-      cube_repl_pe_tp (Case 2): S_kv         (pe=replicate; only 1 PE works)
+      cube_repl_pe_sp  (Case 3): S_kv / P
-      cube_repl_pe_sp (Case 3): S_kv / P     (pe=row_wise within cube)
+      cube_sp_pe_tp_dhead (Case 4): S_kv / C (cube splits S_kv, PE splits d_head)
-      cube_sp_pe_sp (Case 4): S_kv / (C·P)   (★ 64-way split)
+      cube_tp_dhead_pe_sp (Case 5): S_kv / P (cube splits d_head, PE splits S_kv)
+      cube_sp_pe_sp    (Case 6 ★): S_kv / (C·P)
     """
-    S_per_cube = S_kv if "cube_repl" in panel else S_kv // C
+    cube_splits_s = "cube_sp" in panel
-    return S_per_cube // P if "pe_sp" in panel else S_per_cube
+    pe_splits_s   = "pe_sp" in panel
+    S_per_cube = S_kv // C if cube_splits_s else S_kv
+    return S_per_cube // P if pe_splits_s else S_per_cube
+
+
+def _d_head_per_pe(panel: str, *, d_head: int, C: int, P: int) -> int:
+    """d_head dims each PE owns (Cases 4 and 5 shard d_head)."""
+    if "cube_tp_dhead" in panel:    # Case 5: cube shards d_head
+        return d_head // C
+    if "pe_tp_dhead" in panel:      # Case 4: PE shards d_head
+        return d_head // P
+    return d_head                   # Cases 1, 2, 3, 6: full d_head per PE
 
 
 def _active_pe_count(panel: str, *, C: int, P: int) -> int:
     """Number of PEs doing non-idle attention work.
 
-      cube_sp_pe_tp (Case 1): C       (PE 0 of each cube; 7 PEs idle per cube)
+      cube_repl_pe_tp  (Case 1): 1     (PE-TP idle for B=1; only one PE works)
-      cube_repl_pe_tp (Case 2): 1     (only PE 0 of CUBE 0)
+      cube_sp_pe_tp    (Case 2): C     (PE 0 of each cube; 7 PEs idle per cube)
-      cube_repl_pe_sp (Case 3): C·P   (all PEs busy, but cubes are redundant)
+      cube_repl_pe_sp  (Case 3): C·P   (all PEs busy, cube-side redundant)
-      cube_sp_pe_sp (Case 4): C·P     (all 64 PEs doing unique work)
+      cube_sp_pe_tp_dhead (Case 4): C·P (PE shards d_head — all 64 active)
+      cube_tp_dhead_pe_sp (Case 5): C·P (PE shards S_kv — all active)
+      cube_sp_pe_sp    (Case 6 ★): C·P (all 64 PEs doing unique work)
     """
-    if "cube_repl" in panel and "pe_tp" in panel:
+    if "cube_repl" in panel and "pe_tp" in panel and "dhead" not in panel:
         return 1
-    if "cube_sp" in panel and "pe_tp" in panel:
+    if "cube_sp" in panel and "pe_tp" in panel and "dhead" not in panel:
         return C
     return C * P
 
@@ -132,11 +172,12 @@ def _kv_bytes_per_pe(panel: str, *, S_kv: int, h_kv: int,
                      d_head: int, C: int, P: int) -> int:
     """KV bytes a single PE references (K + V, f16, 2 B/elem)."""
     s_local = _s_local_per_pe(panel, S_kv=S_kv, C=C, P=P)
-    return 2 * s_local * h_kv * d_head * 2
+    d_local = _d_head_per_pe(panel, d_head=d_head, C=C, P=P)
+    return 2 * s_local * h_kv * d_local * 2
 
 
 def _plot_memory(rows: list[dict]) -> Path:
-    """Per-PE KV bytes — Case 4 wins (64-way split)."""
+    """Per-PE KV bytes — Case 6 ★ wins (64-way split)."""
     rows = _sorted_by_case(rows)
     labels = [_CASE_INFO[r["panel"]][0] for r in rows]
     mib_per_pe = [
@@ -146,26 +187,25 @@ def _plot_memory(rows: list[dict]) -> Path:
         ) / (1024 * 1024)
         for r in rows
     ]
-    colors = ["#888", "#c0504d", "#888", "#3b6ea5"]  # 4 highlighted, 2 marked red
+    fig, ax = plt.subplots(figsize=(12.0, 4.8))
-    fig, ax = plt.subplots(figsize=(8.0, 4.5))
+    bars = ax.bar(labels, mib_per_pe, color=_bar_colors(rows), width=0.6)
-    bars = ax.bar(labels, mib_per_pe, color=colors, width=0.6)
     ax.set_ylabel("KV bytes per PE (MiB, K + V, f16)")
     ax.set_title(
-        "Long-context decode 4-cases — KV memory per PE\n"
+        "Long-context decode 6-cases — KV memory per PE\n"
         "(one KV-head group; per-layer, per-token state)"
     )
     ax.bar_label(bars, fmt="%.3f", padding=3, fontsize=9)
     ax.grid(axis="y", ls=":", alpha=0.5)
     ax.set_ylim(0, max(mib_per_pe) * 1.15)
     fig.tight_layout()
-    out = _FIG_DIR / "gqa_decode_long_ctx_4cases_memory.png"
+    out = _FIG_DIR / "gqa_decode_long_ctx_6cases_memory.png"
     fig.savefig(out, dpi=150)
     plt.close(fig)
     return out
 
 
 def _plot_parallelism(rows: list[dict]) -> Path:
-    """Total active PE-token compute load — exposes Case 3's redundancy."""
+    """Total active PE-token compute load — exposes redundant-work cases."""
     rows = _sorted_by_case(rows)
     labels = [_CASE_INFO[r["panel"]][0] for r in rows]
     total_work = [
@@ -173,19 +213,19 @@ def _plot_parallelism(rows: list[dict]) -> Path:
         * _s_local_per_pe(r["panel"], S_kv=r["S_kv"], C=r["C"], P=r["P"])
         for r in rows
     ]
-    colors = ["#888", "#888", "#c0504d", "#3b6ea5"]  # 4 highlighted, 3 marked red
+    fig, ax = plt.subplots(figsize=(12.0, 4.8))
-    fig, ax = plt.subplots(figsize=(8.0, 4.5))
+    bars = ax.bar(labels, total_work, color=_bar_colors(rows), width=0.6)
-    bars = ax.bar(labels, total_work, color=colors, width=0.6)
     ax.set_ylabel("active-PE × S_local (PE-tokens; lower ⇒ less wasted work)")
     ax.set_title(
-        "Long-context decode 4-cases — total compute load across active PEs\n"
+        "Long-context decode 6-cases — total compute load across active PEs\n"
-        "(Case 3 replicates the full K/V across 8 cubes ⇒ 8× wasted PE-tokens)"
+        "(Case 3 replicates KV across 8 cubes → 8× wasted PE-tokens; "
+        "Case 6 ★ is fully parallel without replication)"
     )
     ax.bar_label(bars, fmt="%d", padding=3, fontsize=9)
     ax.grid(axis="y", ls=":", alpha=0.5)
     ax.set_ylim(0, max(total_work) * 1.15)
     fig.tight_layout()
-    out = _FIG_DIR / "gqa_decode_long_ctx_4cases_parallelism.png"
+    out = _FIG_DIR / "gqa_decode_long_ctx_6cases_parallelism.png"
     fig.savefig(out, dpi=150)
     plt.close(fig)
     return out

scripts/paper/paper_plot_gqa_decode_long_ctx_composite.py

@@ -0,0 +1,146 @@
+"""Comparative figure for the Case-6 composite-command decode study.
+
+Reads sweep_decode_composite.json (emitted by milestone-1h-gqa, sweep
+``composite``) and writes one two-panel PNG into the bench-output dir:
+
+  gqa_decode_long_ctx_composite.png
+    Left  — end-to-end decode latency (µs) vs context length, per command
+            form (primitive / composite / composite_extended).
+    Right — PE_CPU command count vs context length: the hand-tiled
+            primitive kernel issues O(n_tiles) commands (rises with
+            context), while the coarse composite forms issue O(1) and
+            *saturate* — PE_SCHEDULER absorbs the per-tile fan-out.
+
+The x-axis is the global context length S_kv; each PE owns
+S_local = S_kv/(C·P=64) tokens, so at the 1M production point every PE
+runs Q·Kᵀ of (G·T_q, d_head)·(d_head, 16384) and P·V of
+(G·T_q, 16384)·(16384, d_head).
+
+Run (after the bench):
+    GQA_1H_RUN=1 GQA_1H_SWEEPS=composite python -m kernbench.cli.main run \\
+        --bench milestone-1h-gqa --topology topology.yaml
+    python scripts/paper/paper_plot_gqa_decode_long_ctx_composite.py
+"""
+from __future__ import annotations
+
+import json
+from pathlib import Path
+
+import matplotlib
+
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt  # noqa: E402
+
+_REPO_ROOT = Path(__file__).resolve().parents[2]
+_FIG_DIR = (
+    _REPO_ROOT / "src" / "kernbench" / "benches"
+    / "1H_milestone_output" / "gqa" / "long_ctx"
+)
+_SWEEP_JSON = _FIG_DIR / "sweep_decode_composite.json"
+_PAPER_FIG_DIR = (
+    _REPO_ROOT / "docs" / "report" / "1H-codesign-paper" / "figures"
+)
+
+_N_RANKS = 64  # C·P for the Case-6 64-way split.
+
+# variant key → (display label, colour, marker)
+_VARIANT_STYLE = {
+    "primitive":          ("primitive (tl.dot, hand-tiled)", "#c0504d", "o"),
+    "composite":          ("composite GEMM",                 "#3b6ea5", "s"),
+    "composite_extended": ("composite + softmax_merge",      "#4f8a4f", "^"),
+}
+_ORDER = ("primitive", "composite", "composite_extended")
+
+
+def _load() -> dict:
+    return json.loads(_SWEEP_JSON.read_text())
+
+
+def _series(rows: list[dict], variant: str, key: str):
+    """Sorted (S_kv, value) series for a variant, skipping null values
+    (latency is only measured over the tractable S_kv subset)."""
+    pts = sorted(
+        ((r["S_kv"], r[key]) for r in rows
+         if r["variant"] == variant and r.get(key) is not None),
+        key=lambda t: t[0],
+    )
+    return [p[0] for p in pts], [p[1] for p in pts]
+
+
+def _xticklabels(s_kvs: list[int]) -> list[str]:
+    out = []
+    for s in s_kvs:
+        if s >= 1 << 20:
+            out.append(f"{s // (1 << 20)}M")
+        else:
+            out.append(f"{s // 1024}K")
+    return out
+
+
+def main() -> None:
+    sweep = _load()
+    rows = sweep["rows"]
+    s_kv_op = sweep["s_kv_opcount"]
+    s_kv_lat = sweep["s_kv_latency"]
+
+    fig, (ax_lat, ax_cmd) = plt.subplots(1, 2, figsize=(13.0, 4.8))
+
+    for v in _ORDER:
+        label, color, marker = _VARIANT_STYLE[v]
+        xs, lat = _series(rows, v, "latency_ns")
+        ax_lat.plot(xs, [y / 1e3 for y in lat], marker=marker,
+                    color=color, label=label, lw=2)
+        xs, cmds = _series(rows, v, "pe_cpu_cmd_count")
+        ax_cmd.plot(xs, cmds, marker=marker, color=color, label=label, lw=2)
+
+    ax_lat.set_xticks(s_kv_lat)
+    ax_lat.set_xticklabels(_xticklabels(s_kv_lat))
+    ax_cmd.set_xticks(s_kv_op)
+    ax_cmd.set_xticklabels(_xticklabels(s_kv_op), fontsize=8)
+    for ax in (ax_lat, ax_cmd):
+        ax.set_xscale("log", base=2)
+        ax.set_xlabel(
+            r"context length $S_{kv}$ "
+            r"($S_{\mathrm{local}}=S_{kv}/64$ per PE)"
+        )
+        ax.grid(True, ls=":", alpha=0.5)
+        ax.legend(fontsize=9)
+
+    ax_lat.set_ylabel("end-to-end decode latency (µs)")
+    ax_lat.set_title(
+        "Case-6 decode latency per command form\n"
+        "(memory-bound: command form does not move the critical path)"
+    )
+    ax_cmd.set_ylabel("PE_CPU commands issued")
+    ax_cmd.set_title(
+        "PE_CPU command count per command form\n"
+        "(primitive O(n$_\\mathrm{tiles}$) rises; composite O(1) saturates)"
+    )
+    ax_cmd.axvline(1 << 20, color="#888", ls="--", lw=1, alpha=0.7)
+    ax_cmd.annotate("1M production\ncontext", xy=(1 << 20, 0),
+                    xytext=(1 << 18, 0.6), fontsize=8,
+                    textcoords=("data", "axes fraction"), ha="right",
+                    color="#555")
+
+    fig.suptitle(
+        "Case-6 (Cube-SP × PE-SP) long-context decode — use of composite "
+        "commands\nLLaMA-3.1-70B single-KV-head group (8 cubes × 8 PEs), "
+        "$T_q{=}1$",
+        fontsize=11,
+    )
+    fig.tight_layout(rect=(0, 0, 1, 0.94))
+
+    out = _FIG_DIR / "gqa_decode_long_ctx_composite.png"
+    fig.savefig(out, dpi=150)
+    plt.close(fig)
+    print(f"wrote {out}")
+
+    # Mirror into the paper figures dir (derived artifact).
+    if _PAPER_FIG_DIR.is_dir():
+        dst = _PAPER_FIG_DIR / out.name
+        dst.write_bytes(out.read_bytes())
+        print(f"copied {dst}")
+
+
+if __name__ == "__main__":
+    main()

scripts/paper/paper_plot_gqa_decode_streaming.py

@@ -0,0 +1,124 @@
+"""Comparative figure for the memory-bound decode-streaming composite study.
+
+Reads sweep_decode_streaming.json (emitted by milestone-1h-gqa, sweep
+``decode_streaming``) and writes one two-panel PNG:
+
+  gqa_decode_streaming.png
+    Left  — end-to-end single-rank decode latency (µs) vs per-rank context.
+    Right — achieved HBM bandwidth (GB/s) vs context, against the
+            256 GB/s per-rank roofline.
+
+The memory-bound mirror of the compute-bound prefill figure. With T_q=1
+the GEMMs are skinny (M=8) and the kernel is bound by streaming the KV
+cache. Isolating a single rank (no inter-CUBE reduce) reveals what the
+64-way Case-6 decode masks: the composite command still wins, not by
+feeding the MAC array but by keeping the DMA pipeline full — its
+scheduler-streamed concurrent tile DMAs extract ~230 GB/s (near the
+256 GB/s roofline) while the primitive kernel's blocking tl.dot serializes
+one tile DMA at a time and plateaus at ~166 GB/s. That bandwidth gap is a
+~25-28 % latency win that grows nowhere near prefill's compute-bound
+margin but is decidedly not zero.
+
+Run (after the bench):
+    GQA_1H_RUN=1 GQA_1H_SWEEPS=decode_streaming python -m kernbench.cli.main \\
+        run --bench milestone-1h-gqa --topology topology.yaml
+    python scripts/paper/paper_plot_gqa_decode_streaming.py
+"""
+from __future__ import annotations
+
+import json
+from pathlib import Path
+
+import matplotlib
+
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt  # noqa: E402
+
+_REPO_ROOT = Path(__file__).resolve().parents[2]
+_FIG_DIR = (
+    _REPO_ROOT / "src" / "kernbench" / "benches"
+    / "1H_milestone_output" / "gqa" / "long_ctx"
+)
+_SWEEP_JSON = _FIG_DIR / "sweep_decode_streaming.json"
+_PAPER_FIG_DIR = (
+    _REPO_ROOT / "docs" / "report" / "1H-codesign-paper" / "figures"
+)
+
+# Per-rank HBM roofline: 8 pseudo-channels × 32 GB/s (topology.yaml
+# hbm_ctrl.num_pcs / pc_bw_gbs; = pe_dma_to_noc_bw_gbs).
+_PEAK_HBM_GBS = 256.0
+
+_VARIANT_STYLE = {
+    "primitive":          ("primitive (tl.dot, hand-tiled)", "#c0504d", "o"),
+    "composite":          ("composite GEMM",                 "#3b6ea5", "s"),
+    "composite_extended": ("composite + softmax_merge",      "#4f8a4f", "^"),
+}
+_ORDER = ("primitive", "composite", "composite_extended")
+
+
+def _ctx_label(c: int) -> str:
+    return f"{c // 1024}K" if c >= 1024 else str(c)
+
+
+def _series(rows, variant, key):
+    pts = sorted(((r["s_kv"], r[key]) for r in rows
+                  if r["variant"] == variant), key=lambda t: t[0])
+    return [p[0] for p in pts], [p[1] for p in pts]
+
+
+def main() -> None:
+    sweep = json.loads(_SWEEP_JSON.read_text())
+    rows = sweep["rows"]
+    ctxs = sweep["s_kv_points"]
+
+    fig, (ax_lat, ax_bw) = plt.subplots(1, 2, figsize=(13.0, 4.8))
+
+    for v in _ORDER:
+        label, color, marker = _VARIANT_STYLE[v]
+        xs, lat = _series(rows, v, "latency_ns")
+        ax_lat.plot(xs, [y / 1e3 for y in lat], marker=marker,
+                    color=color, label=label, lw=2)
+        xs, bw = _series(rows, v, "achieved_bw_gbs")
+        ax_bw.plot(xs, bw, marker=marker, color=color, label=label, lw=2)
+
+    for ax in (ax_lat, ax_bw):
+        ax.set_xscale("log", base=2)
+        ax.set_xticks(ctxs)
+        ax.set_xticklabels([_ctx_label(c) for c in ctxs])
+        ax.set_xlabel(r"per-rank context length $S_{kv}$ ($T_q{=}1$)")
+        ax.grid(True, ls=":", alpha=0.5)
+        ax.legend(fontsize=9)
+
+    ax_lat.set_ylabel("end-to-end decode latency (µs)")
+    ax_lat.set_title("Single-rank memory-bound decode latency per command form")
+
+    ax_bw.set_ylabel("achieved HBM bandwidth (GB/s)")
+    ax_bw.set_title(
+        "HBM bandwidth — composite keeps the DMA pipe full; primitive plateaus"
+    )
+    ax_bw.axhline(_PEAK_HBM_GBS, color="#888", ls="--", lw=1, alpha=0.7)
+    ax_bw.text(ctxs[0], _PEAK_HBM_GBS - 8, "256 GB/s roofline",
+               fontsize=8, color="#555", va="top")
+    ax_bw.set_ylim(0, _PEAK_HBM_GBS * 1.08)
+
+    fig.suptitle(
+        "Memory-bound decode streaming — use of composite commands\n"
+        "single-rank, GQA single-KV-head group ($h_q{=}8$, $d_{\\text{head}}"
+        "{=}128$); $M{=}8$ skinny, KV-streaming-bound",
+        fontsize=11,
+    )
+    fig.tight_layout(rect=(0, 0, 1, 0.92))
+
+    out = _FIG_DIR / "gqa_decode_streaming.png"
+    fig.savefig(out, dpi=150)
+    plt.close(fig)
+    print(f"wrote {out}")
+
+    if _PAPER_FIG_DIR.is_dir():
+        dst = _PAPER_FIG_DIR / out.name
+        dst.write_bytes(out.read_bytes())
+        print(f"copied {dst}")
+
+
+if __name__ == "__main__":
+    main()

scripts/paper/paper_plot_gqa_kv_sharding_diagram.py

@@ -0,0 +1,324 @@
+"""6-case KV-sharding tensor diagram (the slide-13 PNG export).
+
+Flat 2-D rectangles, one per sharding case, with:
+    Y axis = S_kv (vertical)        — Cube-SP / PE-SP slice it
+    X axis = d_head (horizontal)    — Cube-TP-d_head / PE-TP-d_head slice it
+
+Drops the batch axis entirely (decode: B = 1, T_q = 1). Same case set
+and visual encoding as slide 13 of GQA_full_deck.pptx; matplotlib
+renders it cleanly so the PNG sits next to the other GQA summary
+artifacts in 1H_milestone_output/gqa/long_ctx/.
+"""
+from __future__ import annotations
+
+from pathlib import Path
+
+import matplotlib.patches as mpatches
+import matplotlib.pyplot as plt
+
+_C = 8
+_P = 8
+
+_GROUP_FILLS = [
+    "#A5D8FF", "#B2F2BB", "#FFD8A8", "#FFC9C9",
+    "#D0BFFF", "#99E9F2", "#FCC2D7", "#FFEC99",
+]
+
+_ACC = {
+    "red":    "#E03131",
+    "orange": "#FD7E14",
+    "blue":   "#1C7ED6",
+    "green":  "#37B24D",
+}
+
+# (label, accent, kv, comm, overflow, encoding-flags, axis-spec)
+#   y_split   = 8 horizontal Y bands  (Cube-SP on S_kv)
+#   x_split   = 8 vertical X bands    (Cube-TP-d_head)
+#   pe_y      = 7 fine horizontal dividers within each Y band
+#   pe_x      = 7 fine vertical dividers within each X band
+#   axes      = small annotation under the chip naming the axes
+#               that the cube/PE actually shard, so the reader can
+#               parse Case 5 (where cube colour fills run X instead
+#               of Y, breaking the visual symmetry of the rest).
+_CASES = [
+    dict(label="Case 1\nCube-Repl / PE-repl", accent=_ACC["red"],
+         kv="40 GB", comm="1.2 MB", overflow=True,
+         y_split=False, x_split=False, pe_y=False, pe_x=False,
+         axes="Cube: replicated     PE: replicated"),
+    dict(label="Case 2\nCube-SP / PE-repl",   accent=_ACC["orange"],
+         kv="5 GB",  comm="3.8 MB", overflow=True,
+         y_split=True,  x_split=False, pe_y=False, pe_x=False,
+         axes="Cube → Y (S_kv)     PE: replicated"),
+    dict(label="Case 3\nCube-Repl / PE-SP",   accent=_ACC["orange"],
+         kv="5 GB",  comm="3.8 MB", overflow=True,
+         y_split=False, x_split=False, pe_y=True,  pe_x=False,
+         axes="Cube: replicated     PE → Y (S_kv)"),
+    dict(label="Case 4\nCube-SP / PE-TP-d_head", accent=_ACC["blue"],
+         kv="640 MB", comm="166 MB", overflow=False,
+         y_split=True,  x_split=False, pe_y=False, pe_x=True,
+         axes="Cube → Y (S_kv)     PE → X (d_head)"),
+    dict(label="Case 5\nCube-TP-d_head / PE-SP", accent=_ACC["blue"],
+         kv="640 MB", comm="166 MB", overflow=False,
+         y_split=False, x_split=True,  pe_y=True,  pe_x=False,
+         axes="Cube → X (d_head)     PE → Y (S_kv)"),
+    dict(label="Case 6 ★\nCube-SP / PE-SP",   accent=_ACC["green"],
+         kv="640 MB", comm="6.2 MB", overflow=False,
+         y_split=True,  x_split=False, pe_y=True,  pe_x=False,
+         axes="Cube → Y (S_kv)     PE → Y (S_kv)"),
+]
+
+_OUT_DIR = (
+    Path(__file__).resolve().parents[2]
+    / "src" / "kernbench" / "benches"
+    / "1H_milestone_output" / "gqa" / "long_ctx"
+)
+
+
+def _draw_panel(ax, cfg):
+    """Draw one case's 2-D KV-tensor rectangle into a panel ax."""
+    ax.set_xlim(0, 1)
+    ax.set_ylim(1, 0)        # Y points down (S_kv ↓)
+    ax.set_aspect("auto")
+    ax.set_xticks([])
+    ax.set_yticks([])
+
+    cube_repl = not cfg["y_split"] and not cfg["x_split"]
+    pe_repl   = not cfg["pe_y"]   and not cfg["pe_x"]
+
+    # Cube-level colour fill.
+    if cfg["y_split"] and not cfg["x_split"]:
+        # 8 horizontal Y bands.
+        for c in range(_C):
+            ax.add_patch(mpatches.Rectangle(
+                (0, c / _C), 1, 1 / _C,
+                facecolor=_GROUP_FILLS[c], edgecolor="black", linewidth=0.6))
+            ax.text(0.04, c / _C + 0.5 / _C, f"C{c}",
+                    ha="left", va="center", fontsize=8,
+                    fontweight="bold", color="#333")
+    elif cfg["x_split"] and not cfg["y_split"]:
+        # 8 vertical X bands.
+        for c in range(_C):
+            ax.add_patch(mpatches.Rectangle(
+                (c / _C, 0), 1 / _C, 1,
+                facecolor=_GROUP_FILLS[c], edgecolor="black", linewidth=0.6))
+            ax.text(c / _C + 0.5 / _C, 0.04, f"C{c}",
+                    ha="center", va="top", fontsize=8,
+                    fontweight="bold", color="#333")
+    else:
+        ax.add_patch(mpatches.Rectangle(
+            (0, 0), 1, 1,
+            facecolor="#F5F5F5", edgecolor="black", linewidth=0.8))
+        ax.text(0.5, 0.5, "× 8 cubes\nfull KV",
+                ha="center", va="center",
+                fontsize=10, fontweight="bold",
+                fontstyle="italic", color="#666")
+
+    # PE-level fine dividers — distinguished from cube boundaries by
+    # using a dashed style + slightly stronger contrast. This is what
+    # makes Case 5's PE-SP (horizontal lines across vertical cube
+    # bands) read as "different axis from the cubes" at a glance.
+    if cfg["pe_y"]:
+        outer = _C if cfg["y_split"] else 1
+        band = 1 / outer
+        for o in range(outer):
+            for p in range(1, _P):
+                y = o * band + band * p / _P
+                ax.axhline(y, color="#222", linewidth=0.8,
+                           linestyle=(0, (3, 2)), alpha=0.75)
+    if cfg["pe_x"]:
+        outer = _C if cfg["x_split"] else 1
+        band = 1 / outer
+        for o in range(outer):
+            for p in range(1, _P):
+                x = o * band + band * p / _P
+                ax.axvline(x, color="#222", linewidth=0.8,
+                           linestyle=(0, (3, 2)), alpha=0.75)
+
+    # Heavy outline on top.
+    ax.add_patch(mpatches.Rectangle(
+        (0, 0), 1, 1, facecolor="none",
+        edgecolor="black", linewidth=1.2))
+
+    # Replication badges — small text-only badges in the corners of
+    # the rectangle, no ghost-card stacking (which mis-reads as a
+    # larger enclosing tensor).
+    badges: list[str] = []
+    if cube_repl:
+        badges.append("× 8 cube copies")
+    if pe_repl and (cfg["y_split"] or cfg["x_split"]):
+        # Cube is sharded but PEs in each cube replicate that shard.
+        badges.append("× 8 PEs / cube replicate")
+    elif pe_repl and cube_repl:
+        # Both replicated — PE replication adds to the cube one.
+        badges.append("× 8 PEs / cube replicate")
+    if badges:
+        ax.text(0.98, 0.02, "\n".join(badges),
+                ha="right", va="top", fontsize=7,
+                fontweight="bold", color="#444",
+                fontstyle="italic",
+                bbox=dict(facecolor="white", edgecolor="#888",
+                          boxstyle="round,pad=0.20", linewidth=0.5))
+
+
+def _make_table_png() -> Path:
+    """Slide-14 companion table: per-PE memory + comm for all 6 cases."""
+    headers = ["Case", "Sharding", "KV / PE", "Fit",
+               "Comm/tok\n(analytical)", "Notes"]
+    rows = [
+        ("Case 1",   "Cube-Repl  ·  PE-repl",      "40 GB",  "✗",
+         "1.2 MB",
+         "no sharding —\nfull KV on every PE"),
+        ("Case 2",   "Cube-SP    ·  PE-repl",      "5 GB",   "✗",
+         "3.8 MB",
+         "cube-axis\nsharded only"),
+        ("Case 3",   "Cube-Repl  ·  PE-SP",        "5 GB",   "✗",
+         "3.8 MB",
+         "PE-axis\nsharded only"),
+        ("Case 4",   "Cube-SP    ·  PE-TP-d_head", "640 MB", "✓",
+         "166 MB",
+         "d_head split intra-cube\npartial-score AR ∝ S_kv"),
+        ("Case 5",   "Cube-TP-d_head  ·  PE-SP",   "640 MB", "✓",
+         "166 MB",
+         "d_head split inter-cube\npartial-score AR on UCIe"),
+        ("Case 6 ★", "Cube-SP    ·  PE-SP",        "640 MB", "✓",
+         "6.2 MB",
+         "S_kv split both axes\n(m,ℓ,O) AR only"),
+    ]
+    accents = [_ACC["red"], _ACC["orange"], _ACC["orange"],
+               _ACC["blue"], _ACC["blue"], _ACC["green"]]
+
+    fig, ax = plt.subplots(figsize=(15.0, 5.0))
+    ax.set_axis_off()
+
+    cell_data = [headers] + [list(r) for r in rows]
+    tbl = ax.table(cellText=cell_data,
+                   colWidths=[0.07, 0.20, 0.09, 0.05, 0.14, 0.28],
+                   cellLoc="center", loc="center")
+    tbl.auto_set_font_size(False)
+    tbl.set_fontsize(10.5)
+    tbl.scale(1.0, 2.4)
+
+    n_cols = len(headers)
+    n_rows = len(rows) + 1   # +1 header
+    # Header styling.
+    for ci in range(n_cols):
+        cell = tbl[(0, ci)]
+        cell.set_facecolor("#1F4E79")
+        cell.set_text_props(color="white", weight="bold")
+        cell.set_edgecolor("#1F4E79")
+    # Body styling.
+    for ri, row in enumerate(rows, start=1):
+        is_pareto = row[0].endswith("★")
+        row_fill = "#E8F5E9" if is_pareto else (
+            "white" if ri % 2 == 1 else "#F5F5F7")
+        # Case-name cell uses accent.
+        case_cell = tbl[(ri, 0)]
+        case_cell.set_facecolor(accents[ri - 1])
+        case_cell.set_text_props(color="white", weight="bold")
+        # Remaining cells.
+        for ci in range(1, n_cols):
+            cell = tbl[(ri, ci)]
+            cell.set_facecolor(row_fill)
+            txt_kwargs = {"weight": "bold" if is_pareto else "normal",
+                          "color": "#333"}
+            if ci == 2:                              # KV / PE
+                txt_kwargs["color"] = (
+                    "#C62828" if row[3] == "✗" else "#2E7D32")
+                txt_kwargs["weight"] = "bold"
+            if ci == 3:                              # Fit
+                txt_kwargs["color"] = (
+                    "#C62828" if row[3] == "✗" else "#2E7D32")
+                txt_kwargs["weight"] = "bold"
+            cell.set_text_props(**txt_kwargs)
+        # Last-column (Notes) cells left-aligned for readability.
+        tbl[(ri, n_cols - 1)].get_text().set_ha("left")
+
+    # Force left-align on the Notes header too.
+    tbl[(0, n_cols - 1)].get_text().set_ha("left")
+
+    fig.suptitle(
+        "GQA decode KV-sharding — per-PE memory & communication\n"
+        "(LLaMA 70B GQA single KV-head group  ·  S_kv = 1 M, FP16, "
+        "80 layers)",
+        fontsize=11.5, y=0.94,
+    )
+    out = _OUT_DIR / "gqa_long_ctx_6cases_kv_sharding_table.png"
+    fig.savefig(out, dpi=150, bbox_inches="tight")
+    plt.close(fig)
+    print(f"wrote {out}")
+    return out
+
+
+def main() -> Path:
+    _OUT_DIR.mkdir(parents=True, exist_ok=True)
+    n = len(_CASES)
+    fig = plt.figure(figsize=(20.0, 7.0))
+    # Three rows per column: case chip · axis-spec annotation · rectangle.
+    gs = fig.add_gridspec(3, n,
+                          height_ratios=[0.55, 0.32, 8.5],
+                          hspace=0.05, wspace=0.20,
+                          left=0.04, right=0.99,
+                          top=0.93, bottom=0.06)
+
+    for i, cfg in enumerate(_CASES):
+        # Top: case chip header.
+        ax_chip = fig.add_subplot(gs[0, i])
+        ax_chip.set_xticks([])
+        ax_chip.set_yticks([])
+        for spine in ax_chip.spines.values():
+            spine.set_visible(False)
+        ax_chip.add_patch(mpatches.Rectangle(
+            (0, 0), 1, 1, transform=ax_chip.transAxes,
+            facecolor=cfg["accent"], edgecolor=cfg["accent"]))
+        ax_chip.text(0.5, 0.5, cfg["label"],
+                     ha="center", va="center",
+                     fontsize=10, fontweight="bold",
+                     color="white")
+
+        # Middle: axis-spec annotation — names which axis the cube
+        # shards on and which axis the PE shards on (essential for
+        # parsing Case 5 where the cube colour fills run X instead
+        # of Y, breaking the visual symmetry of the rest).
+        ax_axes = fig.add_subplot(gs[1, i])
+        ax_axes.set_xticks([])
+        ax_axes.set_yticks([])
+        for spine in ax_axes.spines.values():
+            spine.set_visible(False)
+        ax_axes.add_patch(mpatches.Rectangle(
+            (0, 0), 1, 1, transform=ax_axes.transAxes,
+            facecolor="#F5F5F7", edgecolor="#CCCCCC",
+            linewidth=0.6))
+        ax_axes.text(0.5, 0.5, cfg["axes"],
+                     ha="center", va="center",
+                     fontsize=8.5, fontweight="bold",
+                     color="#1F4E79")
+
+        # Bottom: the tensor rectangle.
+        ax = fig.add_subplot(gs[2, i])
+        _draw_panel(ax, cfg)
+        ax.set_xlabel("X : d_head = 128  →",
+                      fontsize=9, fontweight="bold",
+                      fontstyle="italic", color="#1F4E79")
+        ax.set_ylabel("Y : S_kv = 1 M  ↓",
+                      fontsize=9, fontweight="bold",
+                      fontstyle="italic", color="#1F4E79")
+
+    fig.suptitle(
+        "GQA decode KV-tensor sharding — 6 cases  ·  "
+        "LLaMA 70B GQA single KV-head group  ·  "
+        "C = 8 cubes × P = 8 PEs  ·  S_kv = 1 M, FP16, 80 layers",
+        fontsize=12, y=0.99,
+    )
+
+    out = _OUT_DIR / "gqa_long_ctx_6cases_kv_sharding_diagram.png"
+    fig.savefig(out, dpi=150, bbox_inches="tight")
+    plt.close(fig)
+    print(f"wrote {out}")
+
+    # Companion table PNG (slide-14 export).
+    _make_table_png()
+    return out
+
+
+if __name__ == "__main__":
+    main()

scripts/paper/paper_plot_gqa_prefill_compute_bound.py

@@ -0,0 +1,111 @@
+"""Comparative figure for the compute-bound prefill composite study.
+
+Reads sweep_prefill_compute_bound.json (emitted by milestone-1h-gqa,
+sweep ``prefill_cb``) and writes one two-panel PNG:
+
+  gqa_prefill_compute_bound.png
+    Left  — end-to-end prefill latency (µs) vs context length.
+    Right — MAC utilization (achieved / 8 TFLOP·s⁻¹ per-PE peak) vs context.
+
+Unlike memory-bound decode (where command form is latency-neutral), in
+compute-bound prefill the composite command keeps the MAC array fed by
+streaming DMA↔compute per HW tile, so it wins on both latency and
+utilization — and the margin grows with context (deeper P·V reduction =
+more tiles to pipeline).
+
+Run (after the bench):
+    GQA_1H_RUN=1 GQA_1H_SWEEPS=prefill_cb python -m kernbench.cli.main run \\
+        --bench milestone-1h-gqa --topology topology.yaml
+    python scripts/paper/paper_plot_gqa_prefill_compute_bound.py
+"""
+from __future__ import annotations
+
+import json
+from pathlib import Path
+
+import matplotlib
+
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt  # noqa: E402
+
+_REPO_ROOT = Path(__file__).resolve().parents[2]
+_FIG_DIR = (
+    _REPO_ROOT / "src" / "kernbench" / "benches"
+    / "1H_milestone_output" / "gqa" / "long_ctx"
+)
+_SWEEP_JSON = _FIG_DIR / "sweep_prefill_compute_bound.json"
+_PAPER_FIG_DIR = (
+    _REPO_ROOT / "docs" / "report" / "1H-codesign-paper" / "figures"
+)
+
+_VARIANT_STYLE = {
+    "primitive":          ("primitive (tl.dot, hand-tiled)", "#c0504d", "o"),
+    "composite":          ("composite GEMM",                 "#3b6ea5", "s"),
+    "composite_extended": ("composite + softmax_merge",      "#4f8a4f", "^"),
+}
+_ORDER = ("primitive", "composite", "composite_extended")
+
+
+def _ctx_label(c: int) -> str:
+    return f"{c // 1024}K" if c >= 1024 else str(c)
+
+
+def _series(rows, variant, key):
+    pts = sorted(((r["ctx_len"], r[key]) for r in rows
+                  if r["variant"] == variant), key=lambda t: t[0])
+    return [p[0] for p in pts], [p[1] for p in pts]
+
+
+def main() -> None:
+    sweep = json.loads(_SWEEP_JSON.read_text())
+    rows = sweep["rows"]
+    ctxs = sweep["ctx_points"]
+
+    fig, (ax_lat, ax_util) = plt.subplots(1, 2, figsize=(13.0, 4.8))
+
+    for v in _ORDER:
+        label, color, marker = _VARIANT_STYLE[v]
+        xs, lat = _series(rows, v, "latency_ns")
+        ax_lat.plot(xs, [y / 1e3 for y in lat], marker=marker,
+                    color=color, label=label, lw=2)
+        xs, util = _series(rows, v, "mac_util")
+        ax_util.plot(xs, [u * 100 for u in util], marker=marker,
+                     color=color, label=label, lw=2)
+
+    for ax in (ax_lat, ax_util):
+        ax.set_xscale("log", base=2)
+        ax.set_xticks(ctxs)
+        ax.set_xticklabels([_ctx_label(c) for c in ctxs])
+        ax.set_xlabel(r"context length (= $T_q$ = $S_{kv}$)")
+        ax.grid(True, ls=":", alpha=0.5)
+        ax.legend(fontsize=9)
+
+    ax_lat.set_ylabel("end-to-end prefill latency (µs)")
+    ax_lat.set_title("Compute-bound prefill latency per command form")
+    ax_util.set_ylabel("MAC utilization (% of 8 TFLOP·s⁻¹ peak)")
+    ax_util.set_title(
+        "MAC utilization — composite keeps the array fed; primitive starves"
+    )
+    ax_util.axhline(100, color="#888", ls="--", lw=1, alpha=0.6)
+
+    fig.suptitle(
+        "Compute-bound prefill attention — use of composite commands\n"
+        "single-rank, GQA single-KV-head group ($h_q{=}8$, $d_{\\text{head}}"
+        "{=}128$); $M{=}8T_q$ tile-filling",
+        fontsize=11,
+    )
+    fig.tight_layout(rect=(0, 0, 1, 0.92))
+
+    out = _FIG_DIR / "gqa_prefill_compute_bound.png"
+    fig.savefig(out, dpi=150)
+    plt.close(fig)
+    print(f"wrote {out}")
+
+    if _PAPER_FIG_DIR.is_dir():
+        dst = _PAPER_FIG_DIR / out.name
+        dst.write_bytes(out.read_bytes())
+        print(f"copied {dst}")
+
+
+if __name__ == "__main__":
+    main()

scripts/verify_case4_dhead_tp.py

@@ -0,0 +1,148 @@
+"""Phase 1a smoke test for the Case 4 d_head-TP decode kernel.
+
+Runs `gqa_attention_decode_long_ctx_cube_sp_pe_tp_dhead_kernel` at small
+S_kv (2K) to verify:
+  1. it imports cleanly,
+  2. it runs without error under the same harness as Case 6,
+  3. captures op_log_summary (gemm/ipcq/dma counts),
+  4. compares vs analytical predictions and vs Case 6 (same memory tier).
+
+Usage:
+  python scripts/verify_case4_dhead_tp.py
+"""
+from __future__ import annotations
+
+import os
+import sys
+from pathlib import Path
+
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_sp_pe_sp import (
+    gqa_attention_decode_long_ctx_cube_sp_pe_sp_kernel as _case6_kernel,
+)
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_sp_pe_tp_dhead import (
+    gqa_attention_decode_long_ctx_cube_sp_pe_tp_dhead_kernel as _case4_kernel,
+)
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_tp_dhead_pe_sp import (
+    gqa_attention_decode_long_ctx_cube_tp_dhead_pe_sp_kernel as _case5_kernel,
+)
+from kernbench.benches.gqa_helpers.shared._gqa_panel_helpers import (
+    _ccl_cfg, _summarize_op_log,
+)
+from kernbench.ccl.sfr_config import configure_sfr_intercube_multisip
+from kernbench.policy.placement.dp import DPPolicy
+from kernbench.runtime_api.bench_runner import run_bench
+from kernbench.runtime_api.types import resolve_device
+from kernbench.sim_engine.engine import GraphEngine
+from kernbench.topology.builder import resolve_topology
+
+# Small smoke-test params (S_kv=2K is enough to exercise tile-loop + AR).
+_PARAMS = dict(C=8, P=8, T_q=1, S_kv=2_048,
+               d_head=128, h_q=8, h_kv=1)
+
+
+def _bench_fn_case4(ctx):
+    p = _PARAMS
+    configure_sfr_intercube_multisip(ctx.engine, ctx.spec, _ccl_cfg())
+    dp_full = DPPolicy(cube="replicate", pe="column_wise",
+                       num_cubes=p["C"], num_pes=p["P"])
+    dp_kv = DPPolicy(cube="row_wise", pe="column_wise",
+                     num_cubes=p["C"], num_pes=p["P"])
+    q = ctx.zeros((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp_full, name="q_c4")
+    k = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp_kv, name="k_c4")
+    v = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp_kv, name="v_c4")
+    o = ctx.empty((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp_full, name="o_c4")
+    ctx.launch("case4_dhead_tp", _case4_kernel,
+               q, k, v, o,
+               p["T_q"], p["S_kv"], p["h_q"], p["h_kv"],
+               p["d_head"], p["C"], p["P"],
+               _auto_dim_remap=False)
+
+
+def _bench_fn_case5(ctx):
+    p = _PARAMS
+    configure_sfr_intercube_multisip(ctx.engine, ctx.spec, _ccl_cfg())
+    dp_q = DPPolicy(cube="column_wise", pe="replicate",
+                    num_cubes=p["C"], num_pes=p["P"])
+    dp_kv = DPPolicy(cube="column_wise", pe="row_wise",
+                     num_cubes=p["C"], num_pes=p["P"])
+    q = ctx.zeros((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp_q, name="q_c5")
+    k = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp_kv, name="k_c5")
+    v = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp_kv, name="v_c5")
+    o = ctx.empty((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp_q, name="o_c5")
+    ctx.launch("case5_dhead_tp_inter", _case5_kernel,
+               q, k, v, o,
+               p["T_q"], p["S_kv"], p["h_q"], p["h_kv"],
+               p["d_head"], p["C"], p["P"],
+               _auto_dim_remap=False)
+
+
+def _bench_fn_case6(ctx):
+    p = _PARAMS
+    configure_sfr_intercube_multisip(ctx.engine, ctx.spec, _ccl_cfg())
+    dp_full = DPPolicy(cube="replicate", pe="replicate",
+                       num_cubes=p["C"], num_pes=p["P"])
+    dp_kv = DPPolicy(cube="row_wise", pe="row_wise",
+                     num_cubes=p["C"], num_pes=p["P"])
+    q = ctx.zeros((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp_full, name="q_c6")
+    k = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp_kv, name="k_c6")
+    v = ctx.zeros((p["S_kv"], p["h_kv"] * p["d_head"]),
+                  dtype="f16", dp=dp_kv, name="v_c6")
+    o = ctx.empty((p["T_q"], p["h_q"] * p["d_head"]),
+                  dtype="f16", dp=dp_full, name="o_c6")
+    ctx.launch("case6_sp_sp", _case6_kernel,
+               q, k, v, o,
+               p["T_q"], p["S_kv"], p["h_q"], p["h_kv"],
+               p["d_head"], p["C"], p["P"],
+               _auto_dim_remap=False)
+
+
+def main() -> int:
+    topology = os.environ.get("GQA_1H_TOPOLOGY", "topology.yaml")
+    topo = resolve_topology(topology)
+
+    print(f"Smoke params: {_PARAMS}")
+    print()
+
+    for label, bench_fn in (
+        ("Case 4 (Cube-SP × PE-TP d_head)",   _bench_fn_case4),
+        ("Case 5 (Cube-TP d_head × PE-SP)",   _bench_fn_case5),
+        ("Case 6 (Cube-SP × PE-SP S_kv)",     _bench_fn_case6),
+    ):
+        try:
+            res = run_bench(
+                topology=topo, bench_fn=bench_fn,
+                device=resolve_device(None),
+                engine_factory=lambda t, d: GraphEngine(
+                    getattr(t, "topology_obj", t), enable_data=True,
+                ),
+            )
+        except Exception as e:
+            print(f"  {label:<42} FAIL: {type(e).__name__}: {e}")
+            continue
+        if not res.completion.ok:
+            print(f"  {label:<42} ENGINE FAIL: {res.completion}")
+            continue
+        s = _summarize_op_log(res.engine.op_log)
+        lat = (res.engine.op_log[-1].t_end if res.engine.op_log else 0.0)
+        print(f"  {label:<42} "
+              f"gemm={s['gemm_count']:>4}  "
+              f"ipcq={s['ipcq_copy_count']:>4}  "
+              f"dma_r={s['dma_read_count']:>4}  "
+              f"dma_w={s['dma_write_count']:>3}  "
+              f"latency={lat:.1f} ns  "
+              f"(n_ops={len(res.engine.op_log)})")
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

src/kernbench/benches/1H_milestone_output/gemm/gemm_sweep.json

@@ -33,7 +33,7 @@
       "bytes_hbm": 6144,
       "arith_intensity": 10.666666666666666,
       "tile_count_expected": 1,
-      "sim_wall_clock_s": 0.112,
+      "sim_wall_clock_s": 0.226,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 79.0,
@@ -100,7 +100,7 @@
       "bytes_hbm": 6144,
       "arith_intensity": 10.666666666666666,
       "tile_count_expected": 1,
-      "sim_wall_clock_s": 0.108,
+      "sim_wall_clock_s": 0.186,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 71.0,
@@ -167,11 +167,11 @@
       "bytes_hbm": 6144,
       "arith_intensity": 10.666666666666666,
       "tile_count_expected": 1,
-      "sim_wall_clock_s": 0.105,
+      "sim_wall_clock_s": 0.181,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 63.0,
-          "wall_ns": 63.0,
+          "wall_ns": 51.0,
           "record_count": 3
         },
         "pe_fetch_store": {
@@ -222,7 +222,7 @@
           "record_count": 0
         }
       },
-      "pe_window_ns": 183.394,
+      "pe_window_ns": 151.394,
       "composite_window_ns": 77.38400000000001
     },
     {
@@ -234,7 +234,7 @@
       "bytes_hbm": 10240,
       "arith_intensity": 12.8,
       "tile_count_expected": 1,
-      "sim_wall_clock_s": 0.169,
+      "sim_wall_clock_s": 0.314,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 79.0,
@@ -301,7 +301,7 @@
       "bytes_hbm": 10240,
       "arith_intensity": 12.8,
       "tile_count_expected": 1,
-      "sim_wall_clock_s": 0.163,
+      "sim_wall_clock_s": 0.191,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 79.0,
@@ -368,11 +368,11 @@
       "bytes_hbm": 10240,
       "arith_intensity": 12.8,
       "tile_count_expected": 1,
-      "sim_wall_clock_s": 0.104,
+      "sim_wall_clock_s": 0.186,
       "engines": {
         "pe_dma": {
-          "occupancy_ns": 79.0,
+          "occupancy_ns": 86.0,
-          "wall_ns": 79.0,
+          "wall_ns": 66.0,
           "record_count": 3
         },
         "pe_fetch_store": {
@@ -423,7 +423,7 @@
           "record_count": 0
         }
       },
-      "pe_window_ns": 191.394,
+      "pe_window_ns": 166.394,
       "composite_window_ns": 77.38400000000001
     },
     {
@@ -435,7 +435,7 @@
       "bytes_hbm": 18432,
       "arith_intensity": 14.222222222222221,
       "tile_count_expected": 2,
-      "sim_wall_clock_s": 0.162,
+      "sim_wall_clock_s": 0.318,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 158.995,
@@ -502,7 +502,7 @@
       "bytes_hbm": 18432,
       "arith_intensity": 14.222222222222221,
       "tile_count_expected": 2,
-      "sim_wall_clock_s": 0.183,
+      "sim_wall_clock_s": 0.188,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 139.995,
@@ -569,11 +569,11 @@
       "bytes_hbm": 18432,
       "arith_intensity": 14.222222222222221,
       "tile_count_expected": 2,
-      "sim_wall_clock_s": 0.106,
+      "sim_wall_clock_s": 0.186,
       "engines": {
         "pe_dma": {
-          "occupancy_ns": 112.0,
+          "occupancy_ns": 134.0,
-          "wall_ns": 107.0,
+          "wall_ns": 98.0,
           "record_count": 3
         },
         "pe_fetch_store": {
@@ -624,7 +624,7 @@
           "record_count": 0
         }
       },
-      "pe_window_ns": 223.77800000000002,
+      "pe_window_ns": 214.77800000000002,
       "composite_window_ns": 93.76800000000003
     },
     {
@@ -636,7 +636,7 @@
       "bytes_hbm": 49152,
       "arith_intensity": 21.333333333333332,
       "tile_count_expected": 8,
-      "sim_wall_clock_s": 0.177,
+      "sim_wall_clock_s": 0.308,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 1798.8600000000001,
@@ -703,7 +703,7 @@
       "bytes_hbm": 49152,
       "arith_intensity": 21.333333333333332,
       "tile_count_expected": 8,
-      "sim_wall_clock_s": 0.22,
+      "sim_wall_clock_s": 0.263,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 819.8600000000001,
@@ -770,11 +770,11 @@
       "bytes_hbm": 49152,
       "arith_intensity": 21.333333333333332,
       "tile_count_expected": 8,
-      "sim_wall_clock_s": 0.112,
+      "sim_wall_clock_s": 0.183,
       "engines": {
         "pe_dma": {
-          "occupancy_ns": 283.0,
+          "occupancy_ns": 305.0,
-          "wall_ns": 238.07200000000012,
+          "wall_ns": 229.07200000000012,
           "record_count": 6
         },
         "pe_fetch_store": {
@@ -825,7 +825,7 @@
           "record_count": 0
         }
       },
-      "pe_window_ns": 418.0820000000001,
+      "pe_window_ns": 409.0820000000001,
       "composite_window_ns": 192.07200000000012
     },
     {
@@ -837,7 +837,7 @@
       "bytes_hbm": 395264,
       "arith_intensity": 15.917098445595855,
       "tile_count_expected": 48,
-      "sim_wall_clock_s": 0.2,
+      "sim_wall_clock_s": 0.293,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 56319.35999999987,
@@ -904,7 +904,7 @@
       "bytes_hbm": 395264,
       "arith_intensity": 15.917098445595855,
       "tile_count_expected": 48,
-      "sim_wall_clock_s": 0.237,
+      "sim_wall_clock_s": 0.292,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 20236.359999999866,
@@ -912,13 +912,13 @@
           "record_count": 50
         },
         "pe_fetch_store": {
-          "occupancy_ns": 1543.9999999999998,
+          "occupancy_ns": 1543.9999999999995,
           "wall_ns": 791.9999999999998,
           "record_count": 49
         },
         "pe_gemm": {
-          "occupancy_ns": 1219.584000000017,
+          "occupancy_ns": 1219.5840000000169,
-          "wall_ns": 786.4320000000007,
+          "wall_ns": 786.4320000000005,
           "record_count": 48
         },
         "pe_math": {
@@ -939,7 +939,7 @@
           "record_count": 1
         },
         "FETCH": {
-          "occupancy_ns": 1535.9999999999998,
+          "occupancy_ns": 1535.9999999999995,
           "wall_ns": 783.9999999999998,
           "record_count": 48
         },
@@ -949,8 +949,8 @@
           "record_count": 1
         },
         "GEMM": {
-          "occupancy_ns": 1219.584000000017,
+          "occupancy_ns": 1219.5840000000169,
-          "wall_ns": 786.4320000000007,
+          "wall_ns": 786.4320000000005,
           "record_count": 48
         },
         "MATH": {
@@ -971,10 +971,10 @@
       "bytes_hbm": 395264,
       "arith_intensity": 15.917098445595855,
       "tile_count_expected": 48,
-      "sim_wall_clock_s": 0.208,
+      "sim_wall_clock_s": 0.237,
       "engines": {
         "pe_dma": {
-          "occupancy_ns": 2310.0,
+          "occupancy_ns": 2342.0,
           "wall_ns": 1569.9999999999998,
           "record_count": 3
         },
@@ -1038,7 +1038,7 @@
       "bytes_hbm": 36864,
       "arith_intensity": 7.111111111111111,
       "tile_count_expected": 8,
-      "sim_wall_clock_s": 0.089,
+      "sim_wall_clock_s": 0.259,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 1798.8600000000001,
@@ -1105,7 +1105,7 @@
       "bytes_hbm": 36864,
       "arith_intensity": 7.111111111111111,
       "tile_count_expected": 8,
-      "sim_wall_clock_s": 0.155,
+      "sim_wall_clock_s": 0.313,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 795.8600000000001,
@@ -1172,11 +1172,11 @@
       "bytes_hbm": 36864,
       "arith_intensity": 7.111111111111111,
       "tile_count_expected": 8,
-      "sim_wall_clock_s": 0.234,
+      "sim_wall_clock_s": 0.32,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 258.0,
-          "wall_ns": 218.07200000000012,
+          "wall_ns": 206.07200000000012,
           "record_count": 6
         },
         "pe_fetch_store": {
@@ -1227,7 +1227,7 @@
           "record_count": 0
         }
       },
-      "pe_window_ns": 418.0820000000001,
+      "pe_window_ns": 386.0820000000001,
       "composite_window_ns": 192.07200000000012
     },
     {
@@ -1239,7 +1239,7 @@
       "bytes_hbm": 36864,
       "arith_intensity": 7.111111111111111,
       "tile_count_expected": 16,
-      "sim_wall_clock_s": 0.119,
+      "sim_wall_clock_s": 0.304,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 7854.6320000000005,
@@ -1306,7 +1306,7 @@
       "bytes_hbm": 36864,
       "arith_intensity": 7.111111111111111,
       "tile_count_expected": 16,
-      "sim_wall_clock_s": 0.183,
+      "sim_wall_clock_s": 0.165,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 3779.6320000000005,
@@ -1373,11 +1373,11 @@
       "bytes_hbm": 36864,
       "arith_intensity": 7.111111111111111,
       "tile_count_expected": 16,
-      "sim_wall_clock_s": 0.179,
+      "sim_wall_clock_s": 0.236,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 858.0,
-          "wall_ns": 183.0,
+          "wall_ns": 171.0,
           "record_count": 18
         },
         "pe_fetch_store": {
@@ -1428,7 +1428,7 @@
           "record_count": 0
         }
       },
-      "pe_window_ns": 511.01,
+      "pe_window_ns": 479.01,
       "composite_window_ns": 405.0
     },
     {
@@ -1440,7 +1440,7 @@
       "bytes_hbm": 1572864,
       "arith_intensity": 170.66666666666666,
       "tile_count_expected": 2048,
-      "sim_wall_clock_s": 1.241,
+      "sim_wall_clock_s": 2.164,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 104867178.97599977,
@@ -1507,7 +1507,7 @@
       "bytes_hbm": 1572864,
       "arith_intensity": 170.66666666666666,
       "tile_count_expected": 2048,
-      "sim_wall_clock_s": 0.742,
+      "sim_wall_clock_s": 1.398,
       "engines": {
         "pe_dma": {
           "occupancy_ns": 37760375.97599977,
@@ -1520,7 +1520,7 @@
           "record_count": 2304
         },
         "pe_gemm": {
-          "occupancy_ns": 36417.40799998891,
+          "occupancy_ns": 36417.407999988936,
           "wall_ns": 33554.43199999761,
           "record_count": 2048
         },
@@ -1547,12 +1547,12 @@
           "record_count": 2048
         },
         "STORE": {
-          "occupancy_ns": 130151.58400000146,
+          "occupancy_ns": 130151.58400000144,
           "wall_ns": 31914.784000000003,
           "record_count": 256
         },
         "GEMM": {
-          "occupancy_ns": 36417.40799998891,
+          "occupancy_ns": 36417.407999988936,
           "wall_ns": 33554.43199999761,
           "record_count": 2048
         },
@@ -1574,10 +1574,10 @@
       "bytes_hbm": 1572864,
       "arith_intensity": 170.66666666666666,
       "tile_count_expected": 2048,
-      "sim_wall_clock_s": 0.406,
+      "sim_wall_clock_s": 0.477,
       "engines": {
         "pe_dma": {
-          "occupancy_ns": 142065.0,
+          "occupancy_ns": 142097.0,
           "wall_ns": 6170.0,
           "record_count": 258
         },
@@ -1587,8 +1587,8 @@
           "record_count": 2304
         },
         "pe_gemm": {
-          "occupancy_ns": 838467.5839981049,
+          "occupancy_ns": 838467.5839981119,
-          "wall_ns": 33554.43199999738,
+          "wall_ns": 33554.431999997396,
           "record_count": 2048
         },
         "pe_math": {
@@ -1619,8 +1619,8 @@
           "record_count": 256
         },
         "GEMM": {
-          "occupancy_ns": 838467.5839981049,
+          "occupancy_ns": 838467.5839981119,
-          "wall_ns": 33554.43199999738,
+          "wall_ns": 33554.431999997396,
           "record_count": 2048
         },
         "MATH": {

src/kernbench/benches/1H_milestone_output/gqa/long_ctx/gqa_long_ctx_6cases_measured_comm.json

@@ -0,0 +1,57 @@
+{
+  "S_kv_measured": 65536,
+  "S_kv_headline": 1048576,
+  "n_layers": 80,
+  "num_pes": 64,
+  "wo_ffn_per_token_bytes": 1310720,
+  "cases": {
+    "1": {
+      "label": "Case 1 (Cube-Repl x PE-repl)",
+      "total_ipcq_bytes_one_layer": 0,
+      "per_pe_partial_score_bytes_one_layer": 0,
+      "per_pe_mlo_bytes_one_layer": 0,
+      "per_pe_attn_bytes_per_token_at_1M": 0,
+      "per_pe_total_bytes_per_token_at_1M": 1310720
+    },
+    "2": {
+      "label": "Case 2 (Cube-SP   x PE-repl)",
+      "total_ipcq_bytes_one_layer": 14560,
+      "per_pe_partial_score_bytes_one_layer": 0,
+      "per_pe_mlo_bytes_one_layer": 227,
+      "per_pe_attn_bytes_per_token_at_1M": 18160,
+      "per_pe_total_bytes_per_token_at_1M": 1328880
+    },
+    "3": {
+      "label": "Case 3 (Cube-Repl x PE-SP)",
+      "total_ipcq_bytes_one_layer": 116480,
+      "per_pe_partial_score_bytes_one_layer": 0,
+      "per_pe_mlo_bytes_one_layer": 1820,
+      "per_pe_attn_bytes_per_token_at_1M": 145600,
+      "per_pe_total_bytes_per_token_at_1M": 1456320
+    },
+    "4": {
+      "label": "Case 4 (Cube-SP   x PE-TP d_head)",
+      "total_ipcq_bytes_one_layer": 14696192,
+      "per_pe_partial_score_bytes_one_layer": 131072,
+      "per_pe_mlo_bytes_one_layer": 98556,
+      "per_pe_attn_bytes_per_token_at_1M": 175656640,
+      "per_pe_total_bytes_per_token_at_1M": 176967360
+    },
+    "5": {
+      "label": "Case 5 (Cube-TP d_head x PE-SP)",
+      "total_ipcq_bytes_one_layer": 14696192,
+      "per_pe_partial_score_bytes_one_layer": 131072,
+      "per_pe_mlo_bytes_one_layer": 98556,
+      "per_pe_attn_bytes_per_token_at_1M": 175656640,
+      "per_pe_total_bytes_per_token_at_1M": 176967360
+    },
+    "6": {
+      "label": "Case 6 (Cube-SP   x PE-SP)  [*]",
+      "total_ipcq_bytes_one_layer": 131040,
+      "per_pe_partial_score_bytes_one_layer": 0,
+      "per_pe_mlo_bytes_one_layer": 2047,
+      "per_pe_attn_bytes_per_token_at_1M": 163760,
+      "per_pe_total_bytes_per_token_at_1M": 1474480
+    }
+  }
+}

src/kernbench/benches/1H_milestone_output/gqa/long_ctx/sweep_decode.json

@@ -4,7 +4,9 @@
     "single_kv_group_decode_long_ctx_gqa_cube_sp_pe_sp",
     "single_kv_group_decode_long_ctx_gqa_cube_repl_pe_tp",
     "single_kv_group_decode_long_ctx_gqa_cube_repl_pe_sp",
-    "single_kv_group_decode_long_ctx_gqa_cube_sp_pe_tp"
+    "single_kv_group_decode_long_ctx_gqa_cube_sp_pe_tp",
+    "single_kv_group_decode_long_ctx_gqa_cube_sp_pe_tp_dhead",
+    "single_kv_group_decode_long_ctx_gqa_cube_tp_dhead_pe_sp"
   ],
   "rows": [
     {
@@ -23,11 +25,11 @@
         "dma_read_count": 192,
         "dma_write_count": 1
       },
-      "latency_ns": 34030.00300000079,
+      "latency_ns": 30646.00300000079,
       "engine_occupancy_ns": {
         "pe_gemm": 4194.303999999538,
         "pe_math": 2282.0,
-        "pe_dma": 1138201.8199999991,
+        "pe_dma": 1814643.714999999,
         "pe_fetch_store": 0,
         "pe_ipcq": 0,
         "pe_cpu": 0
@@ -49,11 +51,11 @@
         "dma_read_count": 17,
         "dma_write_count": 1
       },
-      "latency_ns": 25428.38200085424,
+      "latency_ns": 21332.36200087145,
       "engine_occupancy_ns": {
         "pe_gemm": 4194.303999997675,
         "pe_math": 714.0,
-        "pe_dma": 18217.080000881106,
+        "pe_dma": 26829.080001339316,
         "pe_fetch_store": 0,
         "pe_ipcq": 0,
         "pe_cpu": 0
@@ -75,11 +77,11 @@
         "dma_read_count": 192,
         "dma_write_count": 1
       },
-      "latency_ns": 20152.09400000231,
+      "latency_ns": 17826.79400000139,
       "engine_occupancy_ns": {
         "pe_gemm": 33554.431999996305,
         "pe_math": 5684.0,
-        "pe_dma": 853218.7100000716,
+        "pe_dma": 1347469.350000088,
         "pe_fetch_store": 0,
         "pe_ipcq": 0,
         "pe_cpu": 0
@@ -101,11 +103,63 @@
         "dma_read_count": 24,
         "dma_write_count": 1
       },
-      "latency_ns": 26987.291500001098,
+      "latency_ns": 25750.79150000075,
       "engine_occupancy_ns": {
         "pe_gemm": 4194.303999999538,
         "pe_math": 714.0,
-        "pe_dma": 126624.99000000354,
+        "pe_dma": 200429.48000000394,
+        "pe_fetch_store": 0,
+        "pe_ipcq": 0,
+        "pe_cpu": 0
+      }
+    },
+    {
+      "panel": "single_kv_group_decode_long_ctx_gqa_cube_sp_pe_tp_dhead",
+      "kind": "decode_long_ctx_cube_sp_pe_tp_dhead",
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "S_kv": 8192,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "op_log_summary": {
+        "gemm_count": 128,
+        "ipcq_copy_count": 280,
+        "dma_read_count": 192,
+        "dma_write_count": 8
+      },
+      "latency_ns": 31385.502000000277,
+      "engine_occupancy_ns": {
+        "pe_gemm": 4194.303999999545,
+        "pe_math": 10392.0,
+        "pe_dma": 1815478.0114999914,
+        "pe_fetch_store": 0,
+        "pe_ipcq": 0,
+        "pe_cpu": 0
+      }
+    },
+    {
+      "panel": "single_kv_group_decode_long_ctx_gqa_cube_tp_dhead_pe_sp",
+      "kind": "decode_long_ctx_cube_tp_dhead_pe_sp",
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "S_kv": 8192,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "op_log_summary": {
+        "gemm_count": 128,
+        "ipcq_copy_count": 280,
+        "dma_read_count": 192,
+        "dma_write_count": 8
+      },
+      "latency_ns": 34480.97100000052,
+      "engine_occupancy_ns": {
+        "pe_gemm": 4194.303999999538,
+        "pe_math": 10392.0,
+        "pe_dma": 2093451.153999999,
         "pe_fetch_store": 0,
         "pe_ipcq": 0,
         "pe_cpu": 0

src/kernbench/benches/1H_milestone_output/gqa/long_ctx/sweep_decode_composite.json

@@ -0,0 +1,258 @@
+{
+  "version": 2,
+  "variants": [
+    "primitive",
+    "composite",
+    "composite_extended"
+  ],
+  "s_kv_opcount": [
+    8192,
+    65536,
+    131072,
+    262144,
+    524288,
+    1048576
+  ],
+  "s_kv_latency": [
+    8192,
+    32768,
+    65536,
+    131072
+  ],
+  "rows": [
+    {
+      "variant": "primitive",
+      "S_kv": 8192,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 96,
+      "pe_cpu_dispatch_cycles": 930,
+      "latency_ns": 30646.00300000079
+    },
+    {
+      "variant": "composite",
+      "S_kv": 8192,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 94,
+      "pe_cpu_dispatch_cycles": 978,
+      "latency_ns": 30566.1840000007
+    },
+    {
+      "variant": "composite_extended",
+      "S_kv": 8192,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 98,
+      "pe_cpu_dispatch_cycles": 1032,
+      "latency_ns": 30483.359500000362
+    },
+    {
+      "variant": "primitive",
+      "S_kv": 65536,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 96,
+      "pe_cpu_dispatch_cycles": 930,
+      "latency_ns": 231579.37900000165
+    },
+    {
+      "variant": "composite",
+      "S_kv": 65536,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 94,
+      "pe_cpu_dispatch_cycles": 978,
+      "latency_ns": 231270.18400000152
+    },
+    {
+      "variant": "composite_extended",
+      "S_kv": 65536,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 98,
+      "pe_cpu_dispatch_cycles": 1032,
+      "latency_ns": 231211.1740000015
+    },
+    {
+      "variant": "primitive",
+      "S_kv": 131072,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 118,
+      "pe_cpu_dispatch_cycles": 1145,
+      "latency_ns": 461119.51900000183
+    },
+    {
+      "variant": "composite",
+      "S_kv": 131072,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 94,
+      "pe_cpu_dispatch_cycles": 978,
+      "latency_ns": 460651.3170000027
+    },
+    {
+      "variant": "composite_extended",
+      "S_kv": 131072,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 98,
+      "pe_cpu_dispatch_cycles": 1032,
+      "latency_ns": 460552.9805000025
+    },
+    {
+      "variant": "primitive",
+      "S_kv": 262144,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 162,
+      "pe_cpu_dispatch_cycles": 1575,
+      "latency_ns": null
+    },
+    {
+      "variant": "composite",
+      "S_kv": 262144,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 94,
+      "pe_cpu_dispatch_cycles": 978,
+      "latency_ns": null
+    },
+    {
+      "variant": "composite_extended",
+      "S_kv": 262144,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 98,
+      "pe_cpu_dispatch_cycles": 1032,
+      "latency_ns": null
+    },
+    {
+      "variant": "primitive",
+      "S_kv": 524288,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 250,
+      "pe_cpu_dispatch_cycles": 2435,
+      "latency_ns": null
+    },
+    {
+      "variant": "composite",
+      "S_kv": 524288,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 94,
+      "pe_cpu_dispatch_cycles": 978,
+      "latency_ns": null
+    },
+    {
+      "variant": "composite_extended",
+      "S_kv": 524288,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 98,
+      "pe_cpu_dispatch_cycles": 1032,
+      "latency_ns": null
+    },
+    {
+      "variant": "primitive",
+      "S_kv": 1048576,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 426,
+      "pe_cpu_dispatch_cycles": 4155,
+      "latency_ns": null
+    },
+    {
+      "variant": "composite",
+      "S_kv": 1048576,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 94,
+      "pe_cpu_dispatch_cycles": 978,
+      "latency_ns": null
+    },
+    {
+      "variant": "composite_extended",
+      "S_kv": 1048576,
+      "C": 8,
+      "P": 8,
+      "T_q": 1,
+      "d_head": 128,
+      "h_q": 8,
+      "h_kv": 1,
+      "pe_cpu_cmd_count": 98,
+      "pe_cpu_dispatch_cycles": 1032,
+      "latency_ns": null
+    }
+  ]
+}

src/kernbench/benches/1H_milestone_output/gqa/long_ctx/sweep_decode_streaming.json

@@ -0,0 +1,172 @@
+{
+  "version": 1,
+  "variants": [
+    "primitive",
+    "composite",
+    "composite_extended"
+  ],
+  "s_kv_points": [
+    2048,
+    4096,
+    8192,
+    16384
+  ],
+  "rows": [
+    {
+      "variant": "primitive",
+      "s_kv": 2048,
+      "M": 8,
+      "latency_ns": 6188.437999999816,
+      "gemm_busy_ns": 1048.576000000001,
+      "dma_busy_ns": 6322.0,
+      "kv_bytes": 1048576.0,
+      "achieved_bw_gbs": 169.44114169036374,
+      "achieved_tflops": 1.3555291335229098,
+      "mac_util": 0.16944114169036373,
+      "dma_occupancy": 1.021582505957107
+    },
+    {
+      "variant": "composite",
+      "s_kv": 2048,
+      "M": 8,
+      "latency_ns": 4836.115999999989,
+      "gemm_busy_ns": 6629.632000000123,
+      "dma_busy_ns": 267480.720000013,
+      "kv_bytes": 1048576.0,
+      "achieved_bw_gbs": 216.82192900253062,
+      "achieved_tflops": 1.734575432020245,
+      "mac_util": 0.2168219290025306,
+      "dma_occupancy": 55.30899589671001
+    },
+    {
+      "variant": "composite_extended",
+      "s_kv": 2048,
+      "M": 8,
+      "latency_ns": 4737.751999999986,
+      "gemm_busy_ns": 8481.408000000116,
+      "dma_busy_ns": 251398.7400000122,
+      "kv_bytes": 1048576.0,
+      "achieved_bw_gbs": 221.32353065335693,
+      "achieved_tflops": 1.7705882452268553,
+      "mac_util": 0.22132353065335691,
+      "dma_occupancy": 53.06287454472352
+    },
+    {
+      "variant": "primitive",
+      "s_kv": 4096,
+      "M": 8,
+      "latency_ns": 12510.006000000496,
+      "gemm_busy_ns": 2097.152000000002,
+      "dma_busy_ns": 12602.0,
+      "kv_bytes": 2097152.0,
+      "achieved_bw_gbs": 167.6379691584414,
+      "achieved_tflops": 1.3411037532675312,
+      "mac_util": 0.1676379691584414,
+      "dma_occupancy": 1.0073536335633653
+    },
+    {
+      "variant": "composite",
+      "s_kv": 4096,
+      "M": 8,
+      "latency_ns": 9360.883999999554,
+      "gemm_busy_ns": 19520.799999939867,
+      "dma_busy_ns": 1059043.5999999335,
+      "kv_bytes": 2097152.0,
+      "achieved_bw_gbs": 224.03354213128802,
+      "achieved_tflops": 1.792268337050304,
+      "mac_util": 0.224033542131288,
+      "dma_occupancy": 113.13499878857424
+    },
+    {
+      "variant": "composite_extended",
+      "s_kv": 4096,
+      "M": 8,
+      "latency_ns": 9198.135999999562,
+      "gemm_busy_ns": 39531.583999941715,
+      "dma_busy_ns": 1026582.7399999356,
+      "kv_bytes": 2097152.0,
+      "achieved_bw_gbs": 227.99749862364504,
+      "achieved_tflops": 1.8239799889891604,
+      "mac_util": 0.22799749862364505,
+      "dma_occupancy": 111.60769312390951
+    },
+    {
+      "variant": "primitive",
+      "s_kv": 8192,
+      "M": 8,
+      "latency_ns": 25153.141999997388,
+      "gemm_busy_ns": 4194.304000000004,
+      "dma_busy_ns": 25162.0,
+      "kv_bytes": 4194304.0,
+      "achieved_bw_gbs": 166.7506985807354,
+      "achieved_tflops": 1.334005588645883,
+      "mac_util": 0.16675069858073538,
+      "dma_occupancy": 1.0003521627637062
+    },
+    {
+      "variant": "composite",
+      "s_kv": 8192,
+      "M": 8,
+      "latency_ns": 18419.187999999034,
+      "gemm_busy_ns": 64177.11999976149,
+      "dma_busy_ns": 4214541.840000685,
+      "kv_bytes": 4194304.0,
+      "achieved_bw_gbs": 227.713838416776,
+      "achieved_tflops": 1.821710707334208,
+      "mac_util": 0.227713838416776,
+      "dma_occupancy": 228.81257523409317
+    },
+    {
+      "variant": "composite_extended",
+      "s_kv": 8192,
+      "M": 8,
+      "latency_ns": 18128.439999999027,
+      "gemm_busy_ns": 169650.68799976518,
+      "dma_busy_ns": 4149323.2200006745,
+      "kv_bytes": 4194304.0,
+      "achieved_bw_gbs": 231.365964197704,
+      "achieved_tflops": 1.850927713581632,
+      "mac_util": 0.231365964197704,
+      "dma_occupancy": 228.88473691067168
+    },
+    {
+      "variant": "primitive",
+      "s_kv": 16384,
+      "M": 8,
+      "latency_ns": 50439.414000008954,
+      "gemm_busy_ns": 8388.607999999076,
+      "dma_busy_ns": 50282.0,
+      "kv_bytes": 8388608.0,
+      "achieved_bw_gbs": 166.31057609032712,
+      "achieved_tflops": 1.330484608722617,
+      "mac_util": 0.16631057609032712,
+      "dma_occupancy": 0.9968791469304357
+    },
+    {
+      "variant": "composite",
+      "s_kv": 16384,
+      "M": 8,
+      "latency_ns": 36535.79600000568,
+      "gemm_busy_ns": 228978.46400288073,
+      "dma_busy_ns": 16815028.239995122,
+      "kv_bytes": 8388608.0,
+      "achieved_bw_gbs": 229.59970545047648,
+      "achieved_tflops": 1.8367976436038118,
+      "mac_util": 0.22959970545047648,
+      "dma_occupancy": 460.2343477063565
+    },
+    {
+      "variant": "composite_extended",
+      "s_kv": 16384,
+      "M": 8,
+      "latency_ns": 35989.048000005685,
+      "gemm_busy_ns": 701990.3680028584,
+      "dma_busy_ns": 16684294.09999516,
+      "kv_bytes": 8388608.0,
+      "achieved_bw_gbs": 233.0877993771515,
+      "achieved_tflops": 1.864702395017212,
+      "mac_util": 0.2330877993771515,
+      "dma_occupancy": 463.5936493789034
+    }
+  ]
+}

src/kernbench/benches/1H_milestone_output/gqa/long_ctx/sweep_prefill.json

@@ -23,11 +23,11 @@
         "dma_read_count": 192,
         "dma_write_count": 8
       },
-      "latency_ns": 86702.76799999712,
+      "latency_ns": 75610.43899999752,
       "engine_occupancy_ns": {
-        "pe_gemm": 134217.72799998685,
+        "pe_gemm": 134217.72799999084,
         "pe_math": 546656.0,
-        "pe_dma": 2388650.3709999938,
+        "pe_dma": 4511934.237999992,
         "pe_fetch_store": 0,
         "pe_ipcq": 0,
         "pe_cpu": 0
@@ -49,11 +49,11 @@
         "dma_read_count": 1088,
         "dma_write_count": 64
       },
-      "latency_ns": 93564.33200003332,
+      "latency_ns": 67203.7500000147,
       "engine_occupancy_ns": {
         "pe_gemm": 134217.72799998525,
-        "pe_math": 81279.99999999997,
+        "pe_math": 81280.0,
-        "pe_dma": 177951.73600022023,
+        "pe_dma": 446811.4240000146,
         "pe_fetch_store": 0,
         "pe_ipcq": 0,
         "pe_cpu": 0
@@ -75,11 +75,11 @@
         "dma_read_count": 12288,
         "dma_write_count": 64
       },
-      "latency_ns": 190579.57100032456,
+      "latency_ns": 97218.17750020314,
       "engine_occupancy_ns": {
         "pe_gemm": 1073741.8240003586,
         "pe_math": 635904.0,
-        "pe_dma": 1756854.05751659,
+        "pe_dma": 3291993.537524393,
         "pe_fetch_store": 0,
         "pe_ipcq": 0,
         "pe_cpu": 0
@@ -101,11 +101,11 @@
         "dma_read_count": 192,
         "dma_write_count": 64
       },
-      "latency_ns": 38876.32300000038,
+      "latency_ns": 39045.51900000019,
       "engine_occupancy_ns": {
         "pe_gemm": 134217.7280000001,
         "pe_math": 81280.0,
-        "pe_dma": 916284.0330000015,
+        "pe_dma": 1860586.0079999897,
         "pe_fetch_store": 0,
         "pe_ipcq": 0,
         "pe_cpu": 0

src/kernbench/benches/1H_milestone_output/gqa/long_ctx/sweep_prefill_compute_bound.json

@@ -0,0 +1,105 @@
+{
+  "version": 1,
+  "variants": [
+    "primitive",
+    "composite",
+    "composite_extended"
+  ],
+  "ctx_points": [
+    256,
+    512,
+    1024
+  ],
+  "rows": [
+    {
+      "variant": "primitive",
+      "ctx_len": 256,
+      "M": 2048,
+      "latency_ns": 48857.361999999725,
+      "gemm_busy_ns": 33554.43200000003,
+      "dma_busy_ns": 20880.000000000004,
+      "achieved_tflops": 5.4942683151825005,
+      "mac_util": 0.6867835393978126
+    },
+    {
+      "variant": "composite",
+      "ctx_len": 256,
+      "M": 2048,
+      "latency_ns": 49887.505999999594,
+      "gemm_busy_ns": 34531.32799999314,
+      "dma_busy_ns": 1095773.440000072,
+      "achieved_tflops": 5.380815308746887,
+      "mac_util": 0.6726019135933609
+    },
+    {
+      "variant": "composite_extended",
+      "ctx_len": 256,
+      "M": 2048,
+      "latency_ns": 50280.91399999841,
+      "gemm_busy_ns": 129096.19199996963,
+      "dma_busy_ns": 633563.5200000411,
+      "achieved_tflops": 5.338714725830331,
+      "mac_util": 0.6673393407287914
+    },
+    {
+      "variant": "primitive",
+      "ctx_len": 512,
+      "M": 4096,
+      "latency_ns": 197496.81800000605,
+      "gemm_busy_ns": 134217.72800000012,
+      "dma_busy_ns": 66880.0,
+      "achieved_tflops": 5.436755056985105,
+      "mac_util": 0.6795943821231382
+    },
+    {
+      "variant": "composite",
+      "ctx_len": 512,
+      "M": 4096,
+      "latency_ns": 177341.23400000148,
+      "gemm_busy_ns": 141168.63999995415,
+      "dma_busy_ns": 8567063.039998509,
+      "achieved_tflops": 6.054665346469796,
+      "mac_util": 0.7568331683087245
+    },
+    {
+      "variant": "composite_extended",
+      "ctx_len": 512,
+      "M": 4096,
+      "latency_ns": 174702.7699999963,
+      "gemm_busy_ns": 1076198.3999996716,
+      "dma_busy_ns": 6594394.87999886,
+      "achieved_tflops": 6.146106464139193,
+      "mac_util": 0.7682633080173992
+    },
+    {
+      "variant": "primitive",
+      "ctx_len": 1024,
+      "M": 8192,
+      "latency_ns": 794090.4820000405,
+      "gemm_busy_ns": 536870.9120000004,
+      "dma_busy_ns": 234240.0,
+      "achieved_tflops": 5.4086623544239645,
+      "mac_util": 0.6760827943029956
+    },
+    {
+      "variant": "composite",
+      "ctx_len": 1024,
+      "M": 8192,
+      "latency_ns": 668110.7059999453,
+      "gemm_busy_ns": 873011.1999923651,
+      "dma_busy_ns": 67794150.3999821,
+      "achieved_tflops": 6.428526376585786,
+      "mac_util": 0.8035657970732233
+    },
+    {
+      "variant": "composite_extended",
+      "ctx_len": 1024,
+      "M": 8192,
+      "latency_ns": 646937.2019999702,
+      "gemm_busy_ns": 8870907.903995967,
+      "dma_busy_ns": 59655820.79998447,
+      "achieved_tflops": 6.638924586068552,
+      "mac_util": 0.829865573258569
+    }
+  ]
+}

src/kernbench/benches/gqa_helpers/long_ctx/_gqa_attention_decode_long_ctx_cube_sp_pe_sp.py

@@ -36,26 +36,14 @@ Topology / SFR:
 """
 from __future__ import annotations
 
+from kernbench.benches.gqa_helpers.long_ctx._gqa_mlo_reduce import (
+    _ROOT_CUBE,
+    _merge_running,
+    reduce_mlo,
+)
 
 TILE_S_KV = 1024  # ADR-0063 §A.2 S_kv-axis tile sweep (per-tile width).
 
-# lrab geometry for the C=8 single-KV-head group (4×2 cube sub-mesh).
-_SUB_W = 4
-_SUB_H = 2
-_ROOT_COL = _SUB_W // 2   # 2
-_ROOT_ROW = _SUB_H // 2   # 1
-_ROOT_CUBE = _ROOT_ROW * _SUB_W + _ROOT_COL  # 6
-
-
-def _merge_running(m_local, l_local, O_local, m_other, l_other, O_other, *, tl):
-    """Online-softmax merge of two partial ``(m, ℓ, O)`` triples."""
-    m_new = tl.maximum(m_local, m_other)
-    scale_old = tl.exp(m_local - m_new)
-    scale_new = tl.exp(m_other - m_new)
-    l_new = l_local * scale_old + l_other * scale_new
-    O_new = O_local * scale_old + O_other * scale_new
-    return m_new, l_new, O_new
-
 
 def gqa_attention_decode_long_ctx_cube_sp_pe_sp_kernel(
     q_ptr: int,
@@ -119,175 +107,8 @@ def gqa_attention_decode_long_ctx_cube_sp_pe_sp_kernel(
             tl.copy_to(l_local, l_new)
             tl.copy_to(O_local, O_new)
 
-    # ── Intra-CUBE reduce: row chain (intra_W) + col bridge (intra_N) ──
+    # ── Two-level (m, ℓ, O) reduce-to-root (shared helper) ──
-    PE_GRID_COLS = 4
+    reduce_mlo(pe_id, cube_id, m_local, l_local, O_local, P, tl=tl)
-    pe_col = pe_id % PE_GRID_COLS
-    pe_row = pe_id // PE_GRID_COLS
-    pe_cols_used = min(PE_GRID_COLS, P)
-    pe_rows_used = (P + PE_GRID_COLS - 1) // PE_GRID_COLS
-
-    if pe_cols_used > 1:
-        if pe_col < pe_cols_used - 1:
-            with tl.scratch_scope():
-                m_other = tl.recv(dir="intra_E", shape=m_local.shape, dtype="f16")
-                l_other = tl.recv(dir="intra_E", shape=l_local.shape, dtype="f16")
-                O_other = tl.recv(dir="intra_E", shape=O_local.shape, dtype="f16")
-                m_new, l_new, O_new = _merge_running(
-                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
-                )
-                tl.copy_to(m_local, m_new)
-                tl.copy_to(l_local, l_new)
-                tl.copy_to(O_local, O_new)
-        if pe_col > 0:
-            tl.send(dir="intra_W", src=m_local)
-            tl.send(dir="intra_W", src=l_local)
-            tl.send(dir="intra_W", src=O_local)
-
-    if pe_col == 0 and pe_rows_used > 1:
-        if pe_row < pe_rows_used - 1:
-            with tl.scratch_scope():
-                m_other = tl.recv(dir="intra_S", shape=m_local.shape, dtype="f16")
-                l_other = tl.recv(dir="intra_S", shape=l_local.shape, dtype="f16")
-                O_other = tl.recv(dir="intra_S", shape=O_local.shape, dtype="f16")
-                m_new, l_new, O_new = _merge_running(
-                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
-                )
-                tl.copy_to(m_local, m_new)
-                tl.copy_to(l_local, l_new)
-                tl.copy_to(O_local, O_new)
-        if pe_row > 0:
-            tl.send(dir="intra_N", src=m_local)
-            tl.send(dir="intra_N", src=l_local)
-            tl.send(dir="intra_N", src=O_local)
-
-    # ── Inter-CUBE lrab-adapted center-root reduce (ADR-0060 §4.2) ──
-    # Only PE 0 of each CUBE participates. Adapts Phases 1-2 of
-    # lrab_hierarchical_allreduce.py: bidirectional row reduce converges
-    # at root_col; bidirectional col reduce on root_col converges at
-    # root_row. Plain ``+`` replaced by log-sum-exp ``_merge_running``.
-    if pe_id == 0:
-        row = cube_id // _SUB_W
-        col = cube_id % _SUB_W
-
-        # Phase 1: row reduce — converge at col == _ROOT_COL.
-        if col == 0:
-            tl.send(dir="E", src=m_local)
-            tl.send(dir="E", src=l_local)
-            tl.send(dir="E", src=O_local)
-        elif 0 < col < _ROOT_COL:
-            with tl.scratch_scope():
-                m_other = tl.recv(dir="W", shape=m_local.shape, dtype="f16")
-                l_other = tl.recv(dir="W", shape=l_local.shape, dtype="f16")
-                O_other = tl.recv(dir="W", shape=O_local.shape, dtype="f16")
-                m_new, l_new, O_new = _merge_running(
-                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
-                )
-                tl.copy_to(m_local, m_new)
-                tl.copy_to(l_local, l_new)
-                tl.copy_to(O_local, O_new)
-            tl.send(dir="E", src=m_local)
-            tl.send(dir="E", src=l_local)
-            tl.send(dir="E", src=O_local)
-        elif col == _ROOT_COL:
-            with tl.scratch_scope():
-                m_other = tl.recv(dir="W", shape=m_local.shape, dtype="f16")
-                l_other = tl.recv(dir="W", shape=l_local.shape, dtype="f16")
-                O_other = tl.recv(dir="W", shape=O_local.shape, dtype="f16")
-                m_new, l_new, O_new = _merge_running(
-                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
-                )
-                tl.copy_to(m_local, m_new)
-                tl.copy_to(l_local, l_new)
-                tl.copy_to(O_local, O_new)
-            with tl.scratch_scope():
-                m_other = tl.recv(dir="E", shape=m_local.shape, dtype="f16")
-                l_other = tl.recv(dir="E", shape=l_local.shape, dtype="f16")
-                O_other = tl.recv(dir="E", shape=O_local.shape, dtype="f16")
-                m_new, l_new, O_new = _merge_running(
-                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
-                )
-                tl.copy_to(m_local, m_new)
-                tl.copy_to(l_local, l_new)
-                tl.copy_to(O_local, O_new)
-        elif _ROOT_COL < col < _SUB_W - 1:
-            with tl.scratch_scope():
-                m_other = tl.recv(dir="E", shape=m_local.shape, dtype="f16")
-                l_other = tl.recv(dir="E", shape=l_local.shape, dtype="f16")
-                O_other = tl.recv(dir="E", shape=O_local.shape, dtype="f16")
-                m_new, l_new, O_new = _merge_running(
-                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
-                )
-                tl.copy_to(m_local, m_new)
-                tl.copy_to(l_local, l_new)
-                tl.copy_to(O_local, O_new)
-            tl.send(dir="W", src=m_local)
-            tl.send(dir="W", src=l_local)
-            tl.send(dir="W", src=O_local)
-        elif col == _SUB_W - 1:
-            tl.send(dir="W", src=m_local)
-            tl.send(dir="W", src=l_local)
-            tl.send(dir="W", src=O_local)
-
-        # Phase 2: col reduce on col == _ROOT_COL — converge at row == _ROOT_ROW.
-        if col == _ROOT_COL:
-            if row == 0:
-                tl.send(dir="S", src=m_local)
-                tl.send(dir="S", src=l_local)
-                tl.send(dir="S", src=O_local)
-            elif 0 < row < _ROOT_ROW:
-                with tl.scratch_scope():
-                    m_other = tl.recv(dir="N", shape=m_local.shape, dtype="f16")
-                    l_other = tl.recv(dir="N", shape=l_local.shape, dtype="f16")
-                    O_other = tl.recv(dir="N", shape=O_local.shape, dtype="f16")
-                    m_new, l_new, O_new = _merge_running(
-                        m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
-                    )
-                    tl.copy_to(m_local, m_new)
-                    tl.copy_to(l_local, l_new)
-                    tl.copy_to(O_local, O_new)
-                tl.send(dir="S", src=m_local)
-                tl.send(dir="S", src=l_local)
-                tl.send(dir="S", src=O_local)
-            elif row == _ROOT_ROW:
-                with tl.scratch_scope():
-                    m_other = tl.recv(dir="N", shape=m_local.shape, dtype="f16")
-                    l_other = tl.recv(dir="N", shape=l_local.shape, dtype="f16")
-                    O_other = tl.recv(dir="N", shape=O_local.shape, dtype="f16")
-                    m_new, l_new, O_new = _merge_running(
-                        m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
-                    )
-                    tl.copy_to(m_local, m_new)
-                    tl.copy_to(l_local, l_new)
-                    tl.copy_to(O_local, O_new)
-                if _SUB_H - 1 > _ROOT_ROW:
-                    with tl.scratch_scope():
-                        m_other = tl.recv(dir="S", shape=m_local.shape, dtype="f16")
-                        l_other = tl.recv(dir="S", shape=l_local.shape, dtype="f16")
-                        O_other = tl.recv(dir="S", shape=O_local.shape, dtype="f16")
-                        m_new, l_new, O_new = _merge_running(
-                            m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
-                        )
-                        tl.copy_to(m_local, m_new)
-                        tl.copy_to(l_local, l_new)
-                        tl.copy_to(O_local, O_new)
-            elif _ROOT_ROW < row < _SUB_H - 1:
-                with tl.scratch_scope():
-                    m_other = tl.recv(dir="S", shape=m_local.shape, dtype="f16")
-                    l_other = tl.recv(dir="S", shape=l_local.shape, dtype="f16")
-                    O_other = tl.recv(dir="S", shape=O_local.shape, dtype="f16")
-                    m_new, l_new, O_new = _merge_running(
-                        m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
-                    )
-                    tl.copy_to(m_local, m_new)
-                    tl.copy_to(l_local, l_new)
-                    tl.copy_to(O_local, O_new)
-                tl.send(dir="N", src=m_local)
-                tl.send(dir="N", src=l_local)
-                tl.send(dir="N", src=O_local)
-            elif row == _SUB_H - 1 and _SUB_H - 1 > _ROOT_ROW:
-                tl.send(dir="N", src=m_local)
-                tl.send(dir="N", src=l_local)
-                tl.send(dir="N", src=O_local)
 
     # ── Final normalise + store (root only: PE 0 of CUBE 6) ──
     if pe_id == 0 and cube_id == _ROOT_CUBE:

src/kernbench/benches/gqa_helpers/long_ctx/_gqa_attention_decode_long_ctx_cube_sp_pe_sp_composite.py

@@ -0,0 +1,71 @@
+"""GQA decode kernel — Case 6, **composite-GEMM** command form.
+
+Identical placement and (m, ℓ, O) reduce as the primitive Case-6 kernel
+(``_gqa_attention_decode_long_ctx_cube_sp_pe_sp``); the difference is the
+command *granularity* of the local attention. The primitive kernel walks
+its KV slice in ``TILE_S_KV``-wide tiles, issuing per-tile loads + GEMMs
++ a manual online-softmax merge — O(n_tiles) PE_CPU commands. This kernel
+instead issues **one coarse composite GEMM over the whole ``S_local``**
+for each matrix product, passing K and V as HBM refs so PE_SCHEDULER
+streams and tiles them (the DMA fan-out moves off PE_CPU). The
+online-softmax stays primitive but now runs once over the full score row.
+
+So the kernel issues O(1) coarse commands; the scheduler expands each
+composite into the same per-tile DMA + MAC work the primitive kernel
+issued by hand. Fewer, coarser PE_CPU commands ⇒ lower dispatch cost
+under the ADR-0064 Rev2 structural model (the CPU-offload win).
+"""
+from __future__ import annotations
+
+from kernbench.benches.gqa_helpers.long_ctx._gqa_mlo_reduce import (
+    _ROOT_CUBE,
+    reduce_mlo,
+)
+
+
+def gqa_attention_decode_long_ctx_cube_sp_pe_sp_composite_kernel(
+    q_ptr: int,
+    k_ptr: int,
+    v_ptr: int,
+    o_ptr: int,
+    T_q: int,
+    S_kv: int,
+    h_q: int,
+    h_kv: int,
+    d_head: int,
+    C: int,
+    P: int,
+    *,
+    tl,
+) -> None:
+    """Case-6 decode (Cube-SP × PE-SP) — coarse composite-GEMM command form."""
+    G = h_q // h_kv
+    n_ranks = C * P
+    S_local = S_kv // n_ranks
+    pe_id = tl.program_id(axis=0)
+    cube_id = tl.program_id(axis=1)
+
+    # ── Local attention as two coarse composite GEMMs ──
+    # K, V are HBM refs: PE_SCHEDULER streams + tiles them over S_local
+    # (the per-tile DMA fan-out the primitive kernel issued by hand).
+    Q = tl.load(q_ptr, shape=(G * T_q, d_head), dtype="f16")
+    K_T = tl.ref(k_ptr, shape=(d_head, S_local), dtype="f16")
+    V = tl.ref(v_ptr, shape=(S_local, d_head), dtype="f16")
+
+    scores = tl.composite(op="gemm", a=Q, b=K_T)          # Q·Kᵀ, one command
+    m_local = tl.max(scores, axis=-1)
+    centered = scores - m_local
+    exp_scores = tl.exp(centered)
+    l_local = tl.sum(exp_scores, axis=-1)
+    # P·V into a pre-allocated TCM output (explicit out — the composite
+    # output must be a real TCM handle, not auto-allocated scratch).
+    O_local = tl.zeros((G * T_q, d_head), dtype="f16")
+    tl.composite(op="gemm", a=exp_scores, b=V, out=O_local)  # P·V, one command
+
+    # ── Two-level (m, ℓ, O) reduce-to-root (shared helper) ──
+    reduce_mlo(pe_id, cube_id, m_local, l_local, O_local, P, tl=tl)
+
+    # ── Final normalise + store (root only: PE 0 of CUBE 6) ──
+    if pe_id == 0 and cube_id == _ROOT_CUBE:
+        O_final = O_local / l_local
+        tl.store(o_ptr, O_final)

src/kernbench/benches/gqa_helpers/long_ctx/_gqa_attention_decode_long_ctx_cube_sp_pe_sp_composite_ext.py

@@ -0,0 +1,94 @@
+"""GQA decode kernel — Case 6, **composite_extended** command form.
+
+Same placement and (m, ℓ, O) reduce as the primitive / composite-GEMM
+Case-6 kernels; the local attention is the opt2 **two-composite** form
+(ADR-0060 §8 item 4 / ADR-0065), issued coarsely:
+
+  establish  a small first KV slice computes the running (m, ℓ, O) with
+             primitives (kernbench has no scratch-backed ``-inf``
+             initializer, so the recipe — which *merges* into an existing
+             accumulator — needs a seed).
+  #1  Q·Kᵀ   one coarse composite GEMM over the remaining S_local
+             (K passed as an HBM ref → PE_SCHEDULER streams + tiles it).
+  #2  softmax → one ``softmax_merge`` recipe composite (online-softmax
+      + P·V        merge of (m, ℓ, O)) whose head GEMM is P·V over the same
+                   remaining slice (V as an HBM ref), with an ``add``
+                   epilogue folding the P·V contribution into ``O``.
+
+So the bulk of the KV slice is two coarse PE_CPU commands, and the recipe
+additionally folds the per-tile online merge + P·V into one descriptor —
+the fewest / cheapest commands of the three forms (the headline
+CPU-offload win, ADR-0064 Rev2).
+
+Scope: runnable in op_log mode (latency / dispatch only). Full data-mode
+numeric parity of the recipe's 8 MATH ops is a separate follow-up
+(DDD-0065 / the P5-numerics note).
+"""
+from __future__ import annotations
+
+from kernbench.benches.gqa_helpers.long_ctx._gqa_mlo_reduce import (
+    _ROOT_CUBE,
+    reduce_mlo,
+)
+
+# Small primitive seed slice that establishes the running (m, ℓ, O) the
+# recipe then merges into. One scheduler K-tile wide (ADR-0064 TILE_K).
+_SEED_S_KV = 64
+
+
+def gqa_attention_decode_long_ctx_cube_sp_pe_sp_composite_ext_kernel(
+    q_ptr: int,
+    k_ptr: int,
+    v_ptr: int,
+    o_ptr: int,
+    T_q: int,
+    S_kv: int,
+    h_q: int,
+    h_kv: int,
+    d_head: int,
+    C: int,
+    P: int,
+    *,
+    tl,
+) -> None:
+    """Case-6 decode (Cube-SP × PE-SP) — softmax_merge recipe command form."""
+    G = h_q // h_kv
+    n_ranks = C * P
+    S_local = S_kv // n_ranks
+    pe_id = tl.program_id(axis=0)
+    cube_id = tl.program_id(axis=1)
+    KV_ROW_BYTES = d_head * 2  # f16
+
+    # ── Establish running (m, ℓ, O) on a small seed slice (primitive) ──
+    Q = tl.load(q_ptr, shape=(G * T_q, d_head), dtype="f16")
+    seed = min(_SEED_S_KV, S_local)
+    K_T0 = tl.load(k_ptr, shape=(d_head, seed), dtype="f16")
+    V0 = tl.load(v_ptr, shape=(seed, d_head), dtype="f16")
+    scores0 = tl.dot(Q, K_T0)
+    m_local = tl.max(scores0, axis=-1)
+    exp0 = tl.exp(scores0 - m_local)
+    l_local = tl.sum(exp0, axis=-1)
+    O_local = tl.dot(exp0, V0)
+
+    # ── Remaining slice: one Q·Kᵀ composite + one softmax_merge recipe ──
+    rest = S_local - seed
+    if rest > 0:
+        K_T1 = tl.ref(k_ptr + seed * KV_ROW_BYTES,
+                      shape=(d_head, rest), dtype="f16")
+        V1 = tl.ref(v_ptr + seed * KV_ROW_BYTES,
+                    shape=(rest, d_head), dtype="f16")
+        scores1 = tl.composite(op="gemm", a=Q, b=K_T1)
+        tl.composite(
+            prologue=[{"op": "softmax_merge", "s": scores1,
+                       "m": m_local, "l": l_local, "O": O_local}],
+            op="gemm", b=V1, out=O_local,
+            epilogue=[{"op": "add", "other": O_local}],
+        )
+
+    # ── Two-level (m, ℓ, O) reduce-to-root (shared helper) ──
+    reduce_mlo(pe_id, cube_id, m_local, l_local, O_local, P, tl=tl)
+
+    # ── Final normalise + store (root only: PE 0 of CUBE 6) ──
+    if pe_id == 0 and cube_id == _ROOT_CUBE:
+        O_final = O_local / l_local
+        tl.store(o_ptr, O_final)

src/kernbench/benches/gqa_helpers/long_ctx/_gqa_attention_decode_long_ctx_cube_sp_pe_sp_tiled.py

@@ -0,0 +1,129 @@
+"""GQA decode kernel — Case 6, **primitive-TILED** (16×16×16 MAC blocking).
+
+Same Case-6 placement and (m, ℓ, O) reduce as the primitive baseline
+(``_gqa_attention_decode_long_ctx_cube_sp_pe_sp``); the only difference is
+that each local-attention matmul is *hand-blocked into 16×16×16 GemmCmds*
+(mac=16) instead of one coarse ``tl.dot`` per tile. This models a kernel
+that issues the MAC-array fan-out from PE_CPU itself: the per-block GemmCmd
+count is ``ceil(M/16)·ceil(K/16)·ceil(N/16)`` per matmul, charging the full
+ADR-0064 dispatch cost for every block — the "dispatch explosion" the
+composite form offloads to PE_SCHEDULER.
+
+FLOPs are conserved (each 16³ GemmCmd carries the TFLOPS-model compute of
+its block; the blocks sum to the full matmul), so end-to-end compute time
+is unchanged vs the coarse primitive — only the PE_CPU command count and
+its dispatch cycles grow. Inputs are zero (decode bench convention), so the
+blocked accumulation is identically zero; the kernel returns a single
+zeroed ``(M, N)`` output handle that the downstream softmax consumes — no
+per-block accumulation handle needed for this zero-input study.
+"""
+from __future__ import annotations
+
+from math import ceil
+
+from kernbench.common.pe_commands import GemmCmd
+from kernbench.benches.gqa_helpers.long_ctx._gqa_mlo_reduce import (
+    _ROOT_CUBE,
+    _merge_running,
+    reduce_mlo,
+)
+
+TILE_S_KV = 1024  # ADR-0063 §A.2 S_kv-axis tile sweep (per-tile width).
+MAC = 16          # 16×16×16 MAC-array blocking granularity.
+
+
+def _blocked_dot(A, B, *, tl):
+    """``A @ B`` issued as one GemmCmd per 16×16×16 block.
+
+    ``A``:(M, K), ``B``:(K, N) → out:(M, N). Emits
+    ``ceil(M/16)·ceil(K/16)·ceil(N/16)`` GemmCmds (each charged the full
+    ADR-0064 dispatch cost via ``tl.dot``'s emit path). **All blocks write
+    the same ``(M, N)`` output handle** (``out``), and that handle is
+    returned — so downstream softmax ops depend on it exactly like the
+    coarse ``tl.dot`` path (the engine tracks the producer by output handle
+    id, and ``GemmCmd`` is a blocking PE_CPU command, so the block GEMMs
+    serialize on the critical path ahead of the consuming ``tl.max``/
+    ``tl.exp``). K is innermost so each (mi, ni) output tile accumulates
+    across the K blocks into the same handle.
+    """
+    M, K = A.shape[-2], A.shape[-1]
+    K2, N = B.shape[-2], B.shape[-1]
+    assert K == K2, f"blocked_dot shape mismatch K={K} != {K2}"
+    out = tl._make_compute_out(shape=(M, N), dtype="f16")
+    # One GemmCmd per 16³ block, all writing the shared `out` handle. A/B
+    # reuse the full operand handles at block dims (m,k,n = 16, or the
+    # ragged tail) — operands are TCM-resident (pinned), so this charges
+    # dispatch + the block's TFLOPS-model compute. K innermost = accumulate
+    # the K blocks into each (mi, ni) output tile of the shared handle.
+    for _mi in range(ceil(M / MAC)):
+        bm = min(MAC, M - _mi * MAC)
+        for _ni in range(ceil(N / MAC)):
+            bn = min(MAC, N - _ni * MAC)
+            for _ki in range(ceil(K / MAC)):
+                bk = min(MAC, K - _ki * MAC)
+                tl._emit(GemmCmd(a=A, b=B, out=out, m=bm, k=bk, n=bn))
+    return out
+
+
+def gqa_attention_decode_long_ctx_cube_sp_pe_sp_tiled_kernel(
+    q_ptr: int,
+    k_ptr: int,
+    v_ptr: int,
+    o_ptr: int,
+    T_q: int,
+    S_kv: int,
+    h_q: int,
+    h_kv: int,
+    d_head: int,
+    C: int,
+    P: int,
+    *,
+    tl,
+) -> None:
+    """Case-6 decode, primitive-TILED (16×16×16 GemmCmd blocking)."""
+    G = h_q // h_kv
+    n_ranks = C * P
+    S_local = S_kv // n_ranks
+    pe_id = tl.program_id(axis=0)
+    cube_id = tl.program_id(axis=1)
+
+    Q = tl.load(q_ptr, shape=(G * T_q, d_head), dtype="f16")
+    n_tiles = (S_local + TILE_S_KV - 1) // TILE_S_KV
+    KV_ROW_BYTES = d_head * 2  # f16
+
+    tile_s0 = min(TILE_S_KV, S_local)
+    K_T = tl.load(k_ptr, shape=(d_head, tile_s0), dtype="f16")
+    V = tl.load(v_ptr, shape=(tile_s0, d_head), dtype="f16")
+    scores = _blocked_dot(Q, K_T, tl=tl)
+    m_local = tl.max(scores, axis=-1)
+    centered = scores - m_local
+    exp_scores = tl.exp(centered)
+    l_local = tl.sum(exp_scores, axis=-1)
+    O_local = _blocked_dot(exp_scores, V, tl=tl)
+
+    for tile_idx in range(1, n_tiles):
+        tile_start = tile_idx * TILE_S_KV
+        tile_s = min(TILE_S_KV, S_local - tile_start)
+        with tl.scratch_scope():
+            K_T_t = tl.load(k_ptr + tile_start * KV_ROW_BYTES,
+                            shape=(d_head, tile_s), dtype="f16")
+            V_t = tl.load(v_ptr + tile_start * KV_ROW_BYTES,
+                          shape=(tile_s, d_head), dtype="f16")
+            scores_t = _blocked_dot(Q, K_T_t, tl=tl)
+            m_tile = tl.max(scores_t, axis=-1)
+            centered_t = scores_t - m_tile
+            exp_scores_t = tl.exp(centered_t)
+            l_tile = tl.sum(exp_scores_t, axis=-1)
+            O_tile = _blocked_dot(exp_scores_t, V_t, tl=tl)
+            m_new, l_new, O_new = _merge_running(
+                m_local, l_local, O_local, m_tile, l_tile, O_tile, tl=tl,
+            )
+            tl.copy_to(m_local, m_new)
+            tl.copy_to(l_local, l_new)
+            tl.copy_to(O_local, O_new)
+
+    reduce_mlo(pe_id, cube_id, m_local, l_local, O_local, P, tl=tl)
+
+    if pe_id == 0 and cube_id == _ROOT_CUBE:
+        O_final = O_local / l_local
+        tl.store(o_ptr, O_final)

src/kernbench/benches/gqa_helpers/long_ctx/_gqa_attention_decode_long_ctx_cube_sp_pe_tp_dhead.py

@@ -0,0 +1,363 @@
+"""GQA decode kernel — Case 4 (Cube-SP × PE-TP on d_head).
+
+Per the per-PE memory/comm summary chart  (paper_plot_gqa_4cases_summary.py):
+this is the d_head-TP variant — KV is sharded 64-way (cube=row_wise on
+S_kv, pe=column_wise on d_head), so per-PE KV memory matches Case 6 ★
+(640 MB / PE at 1M context across 80 layers), but the attention
+algorithm needs an extra mid-layer partial-score AllReduce because
+each PE only owns d_head/P dimensions of K/V.
+
+Algorithm (single layer, decode T_q=1, B=1):
+  1. Each PE loads its own d_head/P slice of Q, K, V (Q replicated
+     in HBM; PE-load via column-wise DPPolicy carves out the slice).
+  2. Per S_kv tile, each PE computes a PARTIAL score tile:
+        partial_score = Q_shard · K_shard^T  shape: (G·T_q, S_local)
+  3. AR-SUM these partial scores INTRA-CUBE (across the P PEs of a
+     cube) so every PE has the full Q · K^T for the cube's S_local
+     tokens. Online-softmax merge folds tiles into a running state.
+  4. After all tiles, each PE has (m_i, ℓ_i, O_i_shard) for cube i,
+     where O_i_shard is the cube's partial attention output on this
+     PE's d_head/P slice.
+  5. INTER-CUBE online-softmax merge of (m, ℓ, O_shard) across C
+     cubes (lrab center-root reduce). After this, every PE has the
+     full-sequence attention output for its d_head/P slice.
+  6. Final normalise + store: the P PEs of the root cube each write
+     their (G·T_q, d_head/P) slice of the output.
+
+Tensor layout:
+  Q : (T_q, h_q · d_head)  cube=replicate, pe=column_wise on d_head
+      → per-PE shape (T_q, h_q · d_head/P), byte-reshaped to
+        (G·T_q, d_head/P) for the per-PE GEMM input.
+  K : (S_kv, h_kv · d_head) cube=row_wise on S_kv, pe=column_wise
+      on d_head → per-PE shape (S_local, d_head/P), byte-reshaped to
+      (d_head/P, S_local) for the K^T-form GEMM.
+  V : same layout as K.
+  O : (T_q, h_q · d_head) cube=replicate, pe=column_wise on d_head
+      → each writer PE owns its d_head/P slice.
+
+Topology / SFR:
+  - Requires ``configure_sfr_intercube_multisip`` (provides disjoint
+    ``intra_*`` and ``E/W/N/S`` namespaces).
+  - Intra-CUBE PEs in a 2×4 grid; AR-SUM of partial scores uses a
+    reduce-to-PE0-then-broadcast pattern over intra_E/W + intra_N/S.
+  - Inter-CUBE lrab center-root reduce on (m, ℓ, O_shard) over the
+    4×2 cube sub-mesh (root cube = 6).
+
+Naming note: this kernel exemplifies the chart's "PE-TP" label
+meaning **d_head TP** (NOT the batch-axis TP of the original slide-17
+Case 1). The chart-side Case label is "Case 4  Cube-SP × PE-TP".
+
+ADR-0065 P5 scope note: this kernel is meant for op_log / latency
+measurement (matches the milestone-1h-gqa sweep mode). Numeric
+data-mode parity is a separate follow-up.
+"""
+from __future__ import annotations
+
+
+TILE_S_KV = 1024  # ADR-0063 §A.2 S_kv-axis tile sweep (per-tile width).
+
+# lrab geometry for the C=8 single-KV-head group (4×2 cube sub-mesh).
+_SUB_W = 4
+_SUB_H = 2
+_ROOT_COL = _SUB_W // 2   # 2
+_ROOT_ROW = _SUB_H // 2   # 1
+_ROOT_CUBE = _ROOT_ROW * _SUB_W + _ROOT_COL  # 6
+
+
+def _merge_running(m_local, l_local, O_local, m_other, l_other, O_other, *, tl):
+    """Online-softmax merge of two partial ``(m, ℓ, O)`` triples.
+
+    O is the per-PE d_head-slice partial; merge is element-wise so it
+    composes naturally even with O being a d_head shard.
+    """
+    m_new = tl.maximum(m_local, m_other)
+    scale_old = tl.exp(m_local - m_new)
+    scale_new = tl.exp(m_other - m_new)
+    l_new = l_local * scale_old + l_other * scale_new
+    O_new = O_local * scale_old + O_other * scale_new
+    return m_new, l_new, O_new
+
+
+def _intra_cube_allreduce_sum(tensor, *, tl,
+                              pe_col: int, pe_row: int,
+                              pe_cols_used: int, pe_rows_used: int):
+    """Sum-AllReduce a tensor across the 8 PEs of one cube (2×4 grid).
+
+    Reduce-to-PE0 (row chain west + col bridge north) followed by
+    broadcast-from-PE0 (col bridge south + row chain east). Each phase
+    operates in place on ``tensor`` via ``tl.copy_to``.
+
+    Used here to combine PARTIAL Q·K^T score tiles across the P=8 PEs
+    of a cube — each PE has only d_head/P dims, so each contributes a
+    partial dot product; the full score = elementwise sum across PEs.
+    """
+    # ── Phase 1a: row reduce (west) — accumulate at pe_col=0 ────────
+    if pe_cols_used > 1:
+        if pe_col == pe_cols_used - 1:
+            tl.send(dir="intra_W", src=tensor)
+        elif 0 < pe_col < pe_cols_used - 1:
+            with tl.scratch_scope():
+                other = tl.recv(dir="intra_E", shape=tensor.shape, dtype="f16")
+                tl.copy_to(tensor, tensor + other)
+            tl.send(dir="intra_W", src=tensor)
+        elif pe_col == 0:
+            with tl.scratch_scope():
+                other = tl.recv(dir="intra_E", shape=tensor.shape, dtype="f16")
+                tl.copy_to(tensor, tensor + other)
+
+    # ── Phase 1b: col reduce on pe_col=0 (north) — accumulate at row=0 ─
+    if pe_col == 0 and pe_rows_used > 1:
+        if pe_row == pe_rows_used - 1:
+            tl.send(dir="intra_N", src=tensor)
+        elif pe_row == 0:
+            with tl.scratch_scope():
+                other = tl.recv(dir="intra_S", shape=tensor.shape, dtype="f16")
+                tl.copy_to(tensor, tensor + other)
+
+    # ── Phase 2a: col broadcast (south) — row=0 → row=1 on pe_col=0 ─
+    if pe_col == 0 and pe_rows_used > 1:
+        if pe_row == 0:
+            tl.send(dir="intra_S", src=tensor)
+        elif pe_row > 0:
+            with tl.scratch_scope():
+                received = tl.recv(dir="intra_N", shape=tensor.shape, dtype="f16")
+                tl.copy_to(tensor, received)
+
+    # ── Phase 2b: row broadcast (east) — pe_col=0 → all on each row ─
+    if pe_cols_used > 1:
+        if pe_col == 0:
+            tl.send(dir="intra_E", src=tensor)
+        elif 0 < pe_col < pe_cols_used - 1:
+            with tl.scratch_scope():
+                received = tl.recv(dir="intra_W", shape=tensor.shape, dtype="f16")
+                tl.copy_to(tensor, received)
+            tl.send(dir="intra_E", src=tensor)
+        elif pe_col == pe_cols_used - 1:
+            with tl.scratch_scope():
+                received = tl.recv(dir="intra_W", shape=tensor.shape, dtype="f16")
+                tl.copy_to(tensor, received)
+
+
+def gqa_attention_decode_long_ctx_cube_sp_pe_tp_dhead_kernel(
+    q_ptr: int,
+    k_ptr: int,
+    v_ptr: int,
+    o_ptr: int,
+    T_q: int,
+    S_kv: int,
+    h_q: int,
+    h_kv: int,
+    d_head: int,
+    C: int,
+    P: int,
+    *,
+    tl,
+) -> None:
+    """Case-4 decode: Cube-SP × PE-TP-on-d_head, lrab center-root at cube 6."""
+    G = h_q // h_kv
+    if d_head % P != 0:
+        raise ValueError(
+            f"Case 4 requires d_head={d_head} divisible by P={P}"
+        )
+    if S_kv % C != 0:
+        raise ValueError(
+            f"Case 4 requires S_kv={S_kv} divisible by C={C}"
+        )
+
+    d_head_local = d_head // P   # 128 / 8 = 16 in the headline setup
+    S_local = S_kv // C
+    pe_id = tl.program_id(axis=0)
+    cube_id = tl.program_id(axis=1)
+
+    PE_GRID_COLS = 4
+    pe_col = pe_id % PE_GRID_COLS
+    pe_row = pe_id // PE_GRID_COLS
+    pe_cols_used = min(PE_GRID_COLS, P)
+    pe_rows_used = (P + PE_GRID_COLS - 1) // PE_GRID_COLS
+
+    KV_ROW_BYTES = d_head_local * 2  # bytes per K/V row at this PE's d_head slice
+
+    # ── Load this PE's d_head slice of Q (byte-conserving reshape) ──
+    # Q in HBM is (T_q, h_q · d_head); the column_wise pe shard gives
+    # each PE (T_q, h_q · d_head/P). Reshape to (G·T_q, d_head/P) for the
+    # GEMM: byte-conserving since G = h_q (h_kv = 1).
+    Q = tl.load(q_ptr, shape=(G * T_q, d_head_local), dtype="f16")
+
+    # ── Local attention (S_kv-axis tile sweep) with intra-cube AR sum ─
+    n_tiles = (S_local + TILE_S_KV - 1) // TILE_S_KV
+
+    # Tile 0: establish persistent (m, ℓ, O_shard). O_shard is the
+    # per-PE d_head/P slice of the partial attention output.
+    tile_s0 = min(TILE_S_KV, S_local)
+    K_T = tl.load(k_ptr, shape=(d_head_local, tile_s0), dtype="f16")
+    V = tl.load(v_ptr, shape=(tile_s0, d_head_local), dtype="f16")
+    partial_scores0 = tl.dot(Q, K_T)
+    # AR-SUM partial scores intra-cube → every PE in the cube has the
+    # full Q · K^T_tile0 scores.
+    _intra_cube_allreduce_sum(
+        partial_scores0, tl=tl,
+        pe_col=pe_col, pe_row=pe_row,
+        pe_cols_used=pe_cols_used, pe_rows_used=pe_rows_used,
+    )
+    m_local = tl.max(partial_scores0, axis=-1)
+    exp_scores = tl.exp(partial_scores0 - m_local)
+    l_local = tl.sum(exp_scores, axis=-1)
+    # Partial output sharded on d_head: (G·T_q, d_head/P) per PE.
+    O_local = tl.dot(exp_scores, V)
+
+    for tile_idx in range(1, n_tiles):
+        tile_start = tile_idx * TILE_S_KV
+        tile_s = min(TILE_S_KV, S_local - tile_start)
+        with tl.scratch_scope():
+            K_T_t = tl.load(k_ptr + tile_start * KV_ROW_BYTES,
+                            shape=(d_head_local, tile_s), dtype="f16")
+            V_t = tl.load(v_ptr + tile_start * KV_ROW_BYTES,
+                          shape=(tile_s, d_head_local), dtype="f16")
+            partial_scores_t = tl.dot(Q, K_T_t)
+            _intra_cube_allreduce_sum(
+                partial_scores_t, tl=tl,
+                pe_col=pe_col, pe_row=pe_row,
+                pe_cols_used=pe_cols_used, pe_rows_used=pe_rows_used,
+            )
+            m_tile = tl.max(partial_scores_t, axis=-1)
+            exp_scores_t = tl.exp(partial_scores_t - m_tile)
+            l_tile = tl.sum(exp_scores_t, axis=-1)
+            O_tile = tl.dot(exp_scores_t, V_t)
+            m_new, l_new, O_new = _merge_running(
+                m_local, l_local, O_local, m_tile, l_tile, O_tile, tl=tl,
+            )
+            tl.copy_to(m_local, m_new)
+            tl.copy_to(l_local, l_new)
+            tl.copy_to(O_local, O_new)
+
+    # ── Inter-CUBE lrab-adapted center-root reduce on (m, ℓ, O_shard) ─
+    # Every PE participates (each contributes its own d_head/P slice
+    # of O_shard); the merge is element-wise so d_head-sharding is
+    # preserved through the AR.
+    row = cube_id // _SUB_W
+    col = cube_id % _SUB_W
+
+    # Phase 1: row reduce — converge at col == _ROOT_COL.
+    if col == 0:
+        tl.send(dir="E", src=m_local)
+        tl.send(dir="E", src=l_local)
+        tl.send(dir="E", src=O_local)
+    elif 0 < col < _ROOT_COL:
+        with tl.scratch_scope():
+            m_other = tl.recv(dir="W", shape=m_local.shape, dtype="f16")
+            l_other = tl.recv(dir="W", shape=l_local.shape, dtype="f16")
+            O_other = tl.recv(dir="W", shape=O_local.shape, dtype="f16")
+            m_new, l_new, O_new = _merge_running(
+                m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+            )
+            tl.copy_to(m_local, m_new)
+            tl.copy_to(l_local, l_new)
+            tl.copy_to(O_local, O_new)
+        tl.send(dir="E", src=m_local)
+        tl.send(dir="E", src=l_local)
+        tl.send(dir="E", src=O_local)
+    elif col == _ROOT_COL:
+        with tl.scratch_scope():
+            m_other = tl.recv(dir="W", shape=m_local.shape, dtype="f16")
+            l_other = tl.recv(dir="W", shape=l_local.shape, dtype="f16")
+            O_other = tl.recv(dir="W", shape=O_local.shape, dtype="f16")
+            m_new, l_new, O_new = _merge_running(
+                m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+            )
+            tl.copy_to(m_local, m_new)
+            tl.copy_to(l_local, l_new)
+            tl.copy_to(O_local, O_new)
+        with tl.scratch_scope():
+            m_other = tl.recv(dir="E", shape=m_local.shape, dtype="f16")
+            l_other = tl.recv(dir="E", shape=l_local.shape, dtype="f16")
+            O_other = tl.recv(dir="E", shape=O_local.shape, dtype="f16")
+            m_new, l_new, O_new = _merge_running(
+                m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+            )
+            tl.copy_to(m_local, m_new)
+            tl.copy_to(l_local, l_new)
+            tl.copy_to(O_local, O_new)
+    elif _ROOT_COL < col < _SUB_W - 1:
+        with tl.scratch_scope():
+            m_other = tl.recv(dir="E", shape=m_local.shape, dtype="f16")
+            l_other = tl.recv(dir="E", shape=l_local.shape, dtype="f16")
+            O_other = tl.recv(dir="E", shape=O_local.shape, dtype="f16")
+            m_new, l_new, O_new = _merge_running(
+                m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+            )
+            tl.copy_to(m_local, m_new)
+            tl.copy_to(l_local, l_new)
+            tl.copy_to(O_local, O_new)
+        tl.send(dir="W", src=m_local)
+        tl.send(dir="W", src=l_local)
+        tl.send(dir="W", src=O_local)
+    elif col == _SUB_W - 1:
+        tl.send(dir="W", src=m_local)
+        tl.send(dir="W", src=l_local)
+        tl.send(dir="W", src=O_local)
+
+    # Phase 2: col reduce on col == _ROOT_COL — converge at row == _ROOT_ROW.
+    if col == _ROOT_COL:
+        if row == 0:
+            tl.send(dir="S", src=m_local)
+            tl.send(dir="S", src=l_local)
+            tl.send(dir="S", src=O_local)
+        elif 0 < row < _ROOT_ROW:
+            with tl.scratch_scope():
+                m_other = tl.recv(dir="N", shape=m_local.shape, dtype="f16")
+                l_other = tl.recv(dir="N", shape=l_local.shape, dtype="f16")
+                O_other = tl.recv(dir="N", shape=O_local.shape, dtype="f16")
+                m_new, l_new, O_new = _merge_running(
+                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+                )
+                tl.copy_to(m_local, m_new)
+                tl.copy_to(l_local, l_new)
+                tl.copy_to(O_local, O_new)
+            tl.send(dir="S", src=m_local)
+            tl.send(dir="S", src=l_local)
+            tl.send(dir="S", src=O_local)
+        elif row == _ROOT_ROW:
+            with tl.scratch_scope():
+                m_other = tl.recv(dir="N", shape=m_local.shape, dtype="f16")
+                l_other = tl.recv(dir="N", shape=l_local.shape, dtype="f16")
+                O_other = tl.recv(dir="N", shape=O_local.shape, dtype="f16")
+                m_new, l_new, O_new = _merge_running(
+                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+                )
+                tl.copy_to(m_local, m_new)
+                tl.copy_to(l_local, l_new)
+                tl.copy_to(O_local, O_new)
+            if _SUB_H - 1 > _ROOT_ROW:
+                with tl.scratch_scope():
+                    m_other = tl.recv(dir="S", shape=m_local.shape, dtype="f16")
+                    l_other = tl.recv(dir="S", shape=l_local.shape, dtype="f16")
+                    O_other = tl.recv(dir="S", shape=O_local.shape, dtype="f16")
+                    m_new, l_new, O_new = _merge_running(
+                        m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+                    )
+                    tl.copy_to(m_local, m_new)
+                    tl.copy_to(l_local, l_new)
+                    tl.copy_to(O_local, O_new)
+        elif _ROOT_ROW < row < _SUB_H - 1:
+            with tl.scratch_scope():
+                m_other = tl.recv(dir="S", shape=m_local.shape, dtype="f16")
+                l_other = tl.recv(dir="S", shape=l_local.shape, dtype="f16")
+                O_other = tl.recv(dir="S", shape=O_local.shape, dtype="f16")
+                m_new, l_new, O_new = _merge_running(
+                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+                )
+                tl.copy_to(m_local, m_new)
+                tl.copy_to(l_local, l_new)
+                tl.copy_to(O_local, O_new)
+            tl.send(dir="N", src=m_local)
+            tl.send(dir="N", src=l_local)
+            tl.send(dir="N", src=O_local)
+        elif row == _SUB_H - 1 and _SUB_H - 1 > _ROOT_ROW:
+            tl.send(dir="N", src=m_local)
+            tl.send(dir="N", src=l_local)
+            tl.send(dir="N", src=O_local)
+
+    # ── Final normalise + store (root cube: each P-PE writes its d_head slice) ─
+    if cube_id == _ROOT_CUBE:
+        O_final = O_local / l_local
+        tl.store(o_ptr, O_final)

src/kernbench/benches/gqa_helpers/long_ctx/_gqa_attention_decode_long_ctx_cube_tp_dhead_pe_sp.py

@@ -0,0 +1,265 @@
+"""GQA decode kernel — Case 5 (Cube-TP on d_head × PE-SP on S_kv).
+
+Per the per-PE memory/comm summary chart  (paper_plot_gqa_4cases_summary.py):
+this is the second d_head-TP variant — KV is sharded 64-way (cube=
+column_wise on d_head, pe=row_wise on S_kv). Per-PE KV memory matches
+Cases 4 and 6 ★ (640 MB / PE at 1M context across 80 layers), but the
+attention algorithm pays the partial-score AR over INTER-cube fabric
+(UCIe, slower) instead of Case 4's intra-cube fabric (NoC).
+
+Symmetric mirror of Case 4 — intra and inter-cube AR roles SWAPPED:
+
+  Step             Case 4 (Cube-SP × PE-TP/d_head)   Case 5 (THIS — Cube-TP/d_head × PE-SP)
+  ───────────────  ────────────────────────────────  ──────────────────────────────────────
+  partial Q·K^T    AR-SUM intra-cube (NoC)           AR-SUM inter-cube (UCIe)
+  online-softmax   inter-cube lrab on (m, ℓ, O)      intra-cube merge on (m, ℓ, O)
+                   (every PE participates; each       (online-softmax reduce-to-PE0)
+                    PE owns its d_head/P slice)
+  writer           root cube's P PEs (each writes    each cube's PE 0 writes its
+                   its d_head/P slice)                d_head/C slice (C writers total)
+
+Algorithm:
+  1. Each PE loads its (d_head/C × S_kv/P) slice of K and V; full-d_head
+     Q from HBM but only its d_head/C slice via column_wise DPPolicy.
+  2. Per S_kv tile (per PE — already a strict subset of the cube's
+     S_kv), compute partial_score = Q_shard · K_shard^T.  Shape:
+     (G·T_q, S_kv/P).  AR-SUM these PARTIAL scores INTER-CUBE across
+     the C cubes (same PE rank, different d_head slices) so every PE
+     in cube c, rank p has the full Q · K^T_p scores for THIS PE's
+     S_kv/P token range.
+  3. Local softmax + P · V_shard → partial output (G·T_q, d_head/C)
+     for this PE's S_kv/P slice and this cube's d_head/C slice.
+  4. INTRA-CUBE online-softmax merge of (m, ℓ, O) reduce-to-PE0
+     across the cube's P PEs (different S_kv slices, same d_head slice).
+  5. PE 0 of each cube normalises and writes the cube's d_head/C slice.
+
+Tensor layout:
+  Q : (T_q, h_q · d_head)  cube=column_wise on d_head, pe=replicate.
+      Per PE shape (T_q, h_q · d_head/C); reshape to (G·T_q, d_head/C).
+  K : (S_kv, h_kv · d_head) cube=column_wise on d_head, pe=row_wise on S_kv.
+      Per PE shape (S_kv/P, d_head/C); reshape to (d_head/C, S_kv/P) for K^T.
+  V : same layout as K.
+  O : (T_q, h_q · d_head) cube=column_wise on d_head, pe=replicate.
+      Per-cube writer (PE 0) owns its d_head/C slice.
+
+Topology / SFR:
+  - Requires ``configure_sfr_intercube_multisip`` (disjoint intra_*
+    and E/W/N/S namespaces).
+  - INTRA-CUBE PEs arranged 2×4 (used for online-softmax merge).
+  - INTER-CUBE 4×2 cube sub-mesh (used for partial-score AR sum).
+
+ADR-0065 P5 scope note: op_log / latency measurement only; numeric
+data-mode parity is a separate follow-up (matches the milestone-1h-gqa
+sweep mode).
+"""
+from __future__ import annotations
+
+
+TILE_S_KV = 1024  # per-tile S_kv width within a PE's own S_kv/P shard.
+
+# Cube sub-mesh geometry (4×2). For Case 5 we pick cube 0 as the AR-sum
+# root (simpler than lrab; partial-score AR doesn't need geometric centering
+# because the reduction is element-wise sum, not online-softmax merge).
+_SUB_W = 4
+_SUB_H = 2
+_AR_ROOT_CUBE = 0
+
+
+def _merge_running(m_local, l_local, O_local, m_other, l_other, O_other, *, tl):
+    """Online-softmax merge of two partial ``(m, ℓ, O)`` triples."""
+    m_new = tl.maximum(m_local, m_other)
+    scale_old = tl.exp(m_local - m_new)
+    scale_new = tl.exp(m_other - m_new)
+    l_new = l_local * scale_old + l_other * scale_new
+    O_new = O_local * scale_old + O_other * scale_new
+    return m_new, l_new, O_new
+
+
+def _inter_cube_allreduce_sum(tensor, *, tl, cube_id: int,
+                              sub_w: int = _SUB_W, sub_h: int = _SUB_H):
+    """Sum-AllReduce a tensor INTER-CUBE across the 4×2 cube sub-mesh.
+
+    Reduce-to-cube-0 (row chain west + col bridge north) followed by
+    broadcast-from-cube-0 (col bridge south + row chain east). The
+    payload is the partial-score tile — each cube has a different
+    d_head slice's contribution; element-wise sum recovers the full
+    Q·K^T scores for this PE rank's S_kv/P tokens.
+
+    All P PEs of every cube participate independently (per-PE-rank
+    inter-cube fabric lanes).
+    """
+    row = cube_id // sub_w
+    col = cube_id % sub_w
+
+    # ── Phase 1a: row reduce (west) — accumulate at col=0 ────────────
+    if sub_w > 1:
+        if col == sub_w - 1:
+            tl.send(dir="W", src=tensor)
+        elif 0 < col < sub_w - 1:
+            with tl.scratch_scope():
+                other = tl.recv(dir="E", shape=tensor.shape, dtype="f16")
+                tl.copy_to(tensor, tensor + other)
+            tl.send(dir="W", src=tensor)
+        elif col == 0:
+            with tl.scratch_scope():
+                other = tl.recv(dir="E", shape=tensor.shape, dtype="f16")
+                tl.copy_to(tensor, tensor + other)
+
+    # ── Phase 1b: col reduce on col=0 (north) — accumulate at row=0 ──
+    if col == 0 and sub_h > 1:
+        if row == sub_h - 1:
+            tl.send(dir="N", src=tensor)
+        elif row == 0:
+            with tl.scratch_scope():
+                other = tl.recv(dir="S", shape=tensor.shape, dtype="f16")
+                tl.copy_to(tensor, tensor + other)
+
+    # ── Phase 2a: col broadcast (south) — row=0 → row=1 on col=0 ─────
+    if col == 0 and sub_h > 1:
+        if row == 0:
+            tl.send(dir="S", src=tensor)
+        elif row > 0:
+            with tl.scratch_scope():
+                received = tl.recv(dir="N", shape=tensor.shape, dtype="f16")
+                tl.copy_to(tensor, received)
+
+    # ── Phase 2b: row broadcast (east) — col=0 → all on each row ─────
+    if sub_w > 1:
+        if col == 0:
+            tl.send(dir="E", src=tensor)
+        elif 0 < col < sub_w - 1:
+            with tl.scratch_scope():
+                received = tl.recv(dir="W", shape=tensor.shape, dtype="f16")
+                tl.copy_to(tensor, received)
+            tl.send(dir="E", src=tensor)
+        elif col == sub_w - 1:
+            with tl.scratch_scope():
+                received = tl.recv(dir="W", shape=tensor.shape, dtype="f16")
+                tl.copy_to(tensor, received)
+
+
+def gqa_attention_decode_long_ctx_cube_tp_dhead_pe_sp_kernel(
+    q_ptr: int,
+    k_ptr: int,
+    v_ptr: int,
+    o_ptr: int,
+    T_q: int,
+    S_kv: int,
+    h_q: int,
+    h_kv: int,
+    d_head: int,
+    C: int,
+    P: int,
+    *,
+    tl,
+) -> None:
+    """Case-5 decode: Cube-TP-on-d_head × PE-SP-on-S_kv."""
+    G = h_q // h_kv
+    if d_head % C != 0:
+        raise ValueError(
+            f"Case 5 requires d_head={d_head} divisible by C={C}"
+        )
+    if S_kv % P != 0:
+        raise ValueError(
+            f"Case 5 requires S_kv={S_kv} divisible by P={P}"
+        )
+
+    d_head_local = d_head // C       # 128 / 8 = 16 in the headline setup
+    S_local_pe = S_kv // P           # each PE owns S_kv/P tokens
+    pe_id = tl.program_id(axis=0)
+    cube_id = tl.program_id(axis=1)
+
+    PE_GRID_COLS = 4
+    pe_col = pe_id % PE_GRID_COLS
+    pe_row = pe_id // PE_GRID_COLS
+    pe_cols_used = min(PE_GRID_COLS, P)
+    pe_rows_used = (P + PE_GRID_COLS - 1) // PE_GRID_COLS
+
+    KV_ROW_BYTES = d_head_local * 2  # bytes per K/V row at this PE's d_head slice
+
+    # ── Load this PE's d_head slice of Q ────────────────────────────
+    # Q in HBM is (T_q, h_q · d_head); column_wise cube shard gives each
+    # cube (T_q, h_q · d_head/C); pe=replicate within cube → every PE
+    # in the cube has the same data. Reshape to (G·T_q, d_head/C).
+    Q = tl.load(q_ptr, shape=(G * T_q, d_head_local), dtype="f16")
+
+    # ── Local attention (tile sweep on this PE's S_kv/P shard) ──────
+    n_tiles = (S_local_pe + TILE_S_KV - 1) // TILE_S_KV
+
+    # Tile 0: bootstrap (m, ℓ, O_shard). The partial Q·K^T is across
+    # d_head — must AR inter-cube before softmax.
+    tile_s0 = min(TILE_S_KV, S_local_pe)
+    K_T = tl.load(k_ptr, shape=(d_head_local, tile_s0), dtype="f16")
+    V = tl.load(v_ptr, shape=(tile_s0, d_head_local), dtype="f16")
+    partial_scores0 = tl.dot(Q, K_T)
+    _inter_cube_allreduce_sum(partial_scores0, tl=tl, cube_id=cube_id)
+    m_local = tl.max(partial_scores0, axis=-1)
+    exp_scores = tl.exp(partial_scores0 - m_local)
+    l_local = tl.sum(exp_scores, axis=-1)
+    # Partial output (G·T_q, d_head/C) — sharded on d_head only; PE
+    # still holds the full d_head/C slice for its S_kv/P tokens.
+    O_local = tl.dot(exp_scores, V)
+
+    for tile_idx in range(1, n_tiles):
+        tile_start = tile_idx * TILE_S_KV
+        tile_s = min(TILE_S_KV, S_local_pe - tile_start)
+        with tl.scratch_scope():
+            K_T_t = tl.load(k_ptr + tile_start * KV_ROW_BYTES,
+                            shape=(d_head_local, tile_s), dtype="f16")
+            V_t = tl.load(v_ptr + tile_start * KV_ROW_BYTES,
+                          shape=(tile_s, d_head_local), dtype="f16")
+            partial_scores_t = tl.dot(Q, K_T_t)
+            _inter_cube_allreduce_sum(partial_scores_t, tl=tl,
+                                      cube_id=cube_id)
+            m_tile = tl.max(partial_scores_t, axis=-1)
+            exp_scores_t = tl.exp(partial_scores_t - m_tile)
+            l_tile = tl.sum(exp_scores_t, axis=-1)
+            O_tile = tl.dot(exp_scores_t, V_t)
+            m_new, l_new, O_new = _merge_running(
+                m_local, l_local, O_local, m_tile, l_tile, O_tile, tl=tl,
+            )
+            tl.copy_to(m_local, m_new)
+            tl.copy_to(l_local, l_new)
+            tl.copy_to(O_local, O_new)
+
+    # ── INTRA-CUBE online-softmax reduce-to-PE0 (different S_kv slices) ─
+    # Each cube ends with (m, ℓ, O) for its d_head/C slice at PE 0.
+    if pe_cols_used > 1:
+        if pe_col < pe_cols_used - 1:
+            with tl.scratch_scope():
+                m_other = tl.recv(dir="intra_E", shape=m_local.shape, dtype="f16")
+                l_other = tl.recv(dir="intra_E", shape=l_local.shape, dtype="f16")
+                O_other = tl.recv(dir="intra_E", shape=O_local.shape, dtype="f16")
+                m_new, l_new, O_new = _merge_running(
+                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+                )
+                tl.copy_to(m_local, m_new)
+                tl.copy_to(l_local, l_new)
+                tl.copy_to(O_local, O_new)
+        if pe_col > 0:
+            tl.send(dir="intra_W", src=m_local)
+            tl.send(dir="intra_W", src=l_local)
+            tl.send(dir="intra_W", src=O_local)
+
+    if pe_col == 0 and pe_rows_used > 1:
+        if pe_row < pe_rows_used - 1:
+            with tl.scratch_scope():
+                m_other = tl.recv(dir="intra_S", shape=m_local.shape, dtype="f16")
+                l_other = tl.recv(dir="intra_S", shape=l_local.shape, dtype="f16")
+                O_other = tl.recv(dir="intra_S", shape=O_local.shape, dtype="f16")
+                m_new, l_new, O_new = _merge_running(
+                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+                )
+                tl.copy_to(m_local, m_new)
+                tl.copy_to(l_local, l_new)
+                tl.copy_to(O_local, O_new)
+        if pe_row > 0:
+            tl.send(dir="intra_N", src=m_local)
+            tl.send(dir="intra_N", src=l_local)
+            tl.send(dir="intra_N", src=O_local)
+
+    # ── Final normalise + store (one writer per cube: PE 0; each
+    #     cube owns a disjoint d_head/C slice of the output) ─────────
+    if pe_id == 0:
+        O_final = O_local / l_local
+        tl.store(o_ptr, O_final)

src/kernbench/benches/gqa_helpers/long_ctx/_gqa_mlo_reduce.py

@@ -0,0 +1,214 @@
+"""Shared (m, ℓ, O) reduce for the Case-6 long-context decode kernels.
+
+Extracted from the Cube-SP × PE-SP decode kernel so the three
+command-form variants (primitive / composite / composite_extended)
+share one identical reduce. The local attention differs per variant;
+the reduce does not. Behavior is byte-equal to the pre-extraction inline
+reduce (guarded by tests/attention/test_gqa_decode_long_ctx_composite.py).
+
+The reduce is two-level:
+  • intra-CUBE 8-way (row chain along intra_W + col bridge along
+    intra_N) over the 2×4 PE grid;
+  • inter-CUBE 2-phase lrab over the 4×2 CUBE sub-mesh (ADR-0060 §4.2
+    lrab-adapted center-root reduce): Phase 1 row reduce converges at
+    root_col=2; Phase 2 col reduce on root_col converges at root_row=1;
+    root cube id = 6.
+Plain ``+`` is replaced by log-sum-exp ``_merge_running``.
+"""
+from __future__ import annotations
+
+
+# lrab geometry for the C=8 single-KV-head group (4×2 cube sub-mesh).
+_SUB_W = 4
+_SUB_H = 2
+_ROOT_COL = _SUB_W // 2   # 2
+_ROOT_ROW = _SUB_H // 2   # 1
+_ROOT_CUBE = _ROOT_ROW * _SUB_W + _ROOT_COL  # 6
+
+PE_GRID_COLS = 4
+
+
+def _merge_running(m_local, l_local, O_local, m_other, l_other, O_other, *, tl):
+    """Online-softmax merge of two partial ``(m, ℓ, O)`` triples."""
+    m_new = tl.maximum(m_local, m_other)
+    scale_old = tl.exp(m_local - m_new)
+    scale_new = tl.exp(m_other - m_new)
+    l_new = l_local * scale_old + l_other * scale_new
+    O_new = O_local * scale_old + O_other * scale_new
+    return m_new, l_new, O_new
+
+
+def reduce_mlo(pe_id, cube_id, m_local, l_local, O_local, P, *, tl):
+    """Two-level (m, ℓ, O) reduce-to-root (PE 0 of CUBE 6). In place.
+
+    Updates ``m_local``/``l_local``/``O_local`` in place via ``copy_to``;
+    after this call only PE 0 of CUBE ``_ROOT_CUBE`` holds the full result.
+    """
+    # ── Intra-CUBE reduce: row chain (intra_W) + col bridge (intra_N) ──
+    pe_col = pe_id % PE_GRID_COLS
+    pe_row = pe_id // PE_GRID_COLS
+    pe_cols_used = min(PE_GRID_COLS, P)
+    pe_rows_used = (P + PE_GRID_COLS - 1) // PE_GRID_COLS
+
+    if pe_cols_used > 1:
+        if pe_col < pe_cols_used - 1:
+            with tl.scratch_scope():
+                m_other = tl.recv(dir="intra_E", shape=m_local.shape, dtype="f16")
+                l_other = tl.recv(dir="intra_E", shape=l_local.shape, dtype="f16")
+                O_other = tl.recv(dir="intra_E", shape=O_local.shape, dtype="f16")
+                m_new, l_new, O_new = _merge_running(
+                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+                )
+                tl.copy_to(m_local, m_new)
+                tl.copy_to(l_local, l_new)
+                tl.copy_to(O_local, O_new)
+        if pe_col > 0:
+            tl.send(dir="intra_W", src=m_local)
+            tl.send(dir="intra_W", src=l_local)
+            tl.send(dir="intra_W", src=O_local)
+
+    if pe_col == 0 and pe_rows_used > 1:
+        if pe_row < pe_rows_used - 1:
+            with tl.scratch_scope():
+                m_other = tl.recv(dir="intra_S", shape=m_local.shape, dtype="f16")
+                l_other = tl.recv(dir="intra_S", shape=l_local.shape, dtype="f16")
+                O_other = tl.recv(dir="intra_S", shape=O_local.shape, dtype="f16")
+                m_new, l_new, O_new = _merge_running(
+                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+                )
+                tl.copy_to(m_local, m_new)
+                tl.copy_to(l_local, l_new)
+                tl.copy_to(O_local, O_new)
+        if pe_row > 0:
+            tl.send(dir="intra_N", src=m_local)
+            tl.send(dir="intra_N", src=l_local)
+            tl.send(dir="intra_N", src=O_local)
+
+    # ── Inter-CUBE lrab-adapted center-root reduce (ADR-0060 §4.2) ──
+    # Only PE 0 of each CUBE participates. Adapts Phases 1-2 of
+    # lrab_hierarchical_allreduce.py: bidirectional row reduce converges
+    # at root_col; bidirectional col reduce on root_col converges at
+    # root_row. Plain ``+`` replaced by log-sum-exp ``_merge_running``.
+    if pe_id == 0:
+        row = cube_id // _SUB_W
+        col = cube_id % _SUB_W
+
+        # Phase 1: row reduce — converge at col == _ROOT_COL.
+        if col == 0:
+            tl.send(dir="E", src=m_local)
+            tl.send(dir="E", src=l_local)
+            tl.send(dir="E", src=O_local)
+        elif 0 < col < _ROOT_COL:
+            with tl.scratch_scope():
+                m_other = tl.recv(dir="W", shape=m_local.shape, dtype="f16")
+                l_other = tl.recv(dir="W", shape=l_local.shape, dtype="f16")
+                O_other = tl.recv(dir="W", shape=O_local.shape, dtype="f16")
+                m_new, l_new, O_new = _merge_running(
+                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+                )
+                tl.copy_to(m_local, m_new)
+                tl.copy_to(l_local, l_new)
+                tl.copy_to(O_local, O_new)
+            tl.send(dir="E", src=m_local)
+            tl.send(dir="E", src=l_local)
+            tl.send(dir="E", src=O_local)
+        elif col == _ROOT_COL:
+            with tl.scratch_scope():
+                m_other = tl.recv(dir="W", shape=m_local.shape, dtype="f16")
+                l_other = tl.recv(dir="W", shape=l_local.shape, dtype="f16")
+                O_other = tl.recv(dir="W", shape=O_local.shape, dtype="f16")
+                m_new, l_new, O_new = _merge_running(
+                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+                )
+                tl.copy_to(m_local, m_new)
+                tl.copy_to(l_local, l_new)
+                tl.copy_to(O_local, O_new)
+            with tl.scratch_scope():
+                m_other = tl.recv(dir="E", shape=m_local.shape, dtype="f16")
+                l_other = tl.recv(dir="E", shape=l_local.shape, dtype="f16")
+                O_other = tl.recv(dir="E", shape=O_local.shape, dtype="f16")
+                m_new, l_new, O_new = _merge_running(
+                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+                )
+                tl.copy_to(m_local, m_new)
+                tl.copy_to(l_local, l_new)
+                tl.copy_to(O_local, O_new)
+        elif _ROOT_COL < col < _SUB_W - 1:
+            with tl.scratch_scope():
+                m_other = tl.recv(dir="E", shape=m_local.shape, dtype="f16")
+                l_other = tl.recv(dir="E", shape=l_local.shape, dtype="f16")
+                O_other = tl.recv(dir="E", shape=O_local.shape, dtype="f16")
+                m_new, l_new, O_new = _merge_running(
+                    m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+                )
+                tl.copy_to(m_local, m_new)
+                tl.copy_to(l_local, l_new)
+                tl.copy_to(O_local, O_new)
+            tl.send(dir="W", src=m_local)
+            tl.send(dir="W", src=l_local)
+            tl.send(dir="W", src=O_local)
+        elif col == _SUB_W - 1:
+            tl.send(dir="W", src=m_local)
+            tl.send(dir="W", src=l_local)
+            tl.send(dir="W", src=O_local)
+
+        # Phase 2: col reduce on col == _ROOT_COL — converge at row == _ROOT_ROW.
+        if col == _ROOT_COL:
+            if row == 0:
+                tl.send(dir="S", src=m_local)
+                tl.send(dir="S", src=l_local)
+                tl.send(dir="S", src=O_local)
+            elif 0 < row < _ROOT_ROW:
+                with tl.scratch_scope():
+                    m_other = tl.recv(dir="N", shape=m_local.shape, dtype="f16")
+                    l_other = tl.recv(dir="N", shape=l_local.shape, dtype="f16")
+                    O_other = tl.recv(dir="N", shape=O_local.shape, dtype="f16")
+                    m_new, l_new, O_new = _merge_running(
+                        m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+                    )
+                    tl.copy_to(m_local, m_new)
+                    tl.copy_to(l_local, l_new)
+                    tl.copy_to(O_local, O_new)
+                tl.send(dir="S", src=m_local)
+                tl.send(dir="S", src=l_local)
+                tl.send(dir="S", src=O_local)
+            elif row == _ROOT_ROW:
+                with tl.scratch_scope():
+                    m_other = tl.recv(dir="N", shape=m_local.shape, dtype="f16")
+                    l_other = tl.recv(dir="N", shape=l_local.shape, dtype="f16")
+                    O_other = tl.recv(dir="N", shape=O_local.shape, dtype="f16")
+                    m_new, l_new, O_new = _merge_running(
+                        m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+                    )
+                    tl.copy_to(m_local, m_new)
+                    tl.copy_to(l_local, l_new)
+                    tl.copy_to(O_local, O_new)
+                if _SUB_H - 1 > _ROOT_ROW:
+                    with tl.scratch_scope():
+                        m_other = tl.recv(dir="S", shape=m_local.shape, dtype="f16")
+                        l_other = tl.recv(dir="S", shape=l_local.shape, dtype="f16")
+                        O_other = tl.recv(dir="S", shape=O_local.shape, dtype="f16")
+                        m_new, l_new, O_new = _merge_running(
+                            m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+                        )
+                        tl.copy_to(m_local, m_new)
+                        tl.copy_to(l_local, l_new)
+                        tl.copy_to(O_local, O_new)
+            elif _ROOT_ROW < row < _SUB_H - 1:
+                with tl.scratch_scope():
+                    m_other = tl.recv(dir="S", shape=m_local.shape, dtype="f16")
+                    l_other = tl.recv(dir="S", shape=l_local.shape, dtype="f16")
+                    O_other = tl.recv(dir="S", shape=O_local.shape, dtype="f16")
+                    m_new, l_new, O_new = _merge_running(
+                        m_local, l_local, O_local, m_other, l_other, O_other, tl=tl,
+                    )
+                    tl.copy_to(m_local, m_new)
+                    tl.copy_to(l_local, l_new)
+                    tl.copy_to(O_local, O_new)
+                tl.send(dir="N", src=m_local)
+                tl.send(dir="N", src=l_local)
+                tl.send(dir="N", src=O_local)
+            elif row == _SUB_H - 1 and _SUB_H - 1 > _ROOT_ROW:
+                tl.send(dir="N", src=m_local)
+                tl.send(dir="N", src=l_local)
+                tl.send(dir="N", src=O_local)

src/kernbench/benches/gqa_helpers/long_ctx/_gqa_prefill_compute_bound.py

@@ -0,0 +1,161 @@
+"""Compute-bound prefill attention — 3 command-form variants (single-rank).
+
+Companion to the memory-bound decode study
+(``_gqa_attention_decode_long_ctx_cube_sp_pe_sp*``). Prefill processes a
+block of T_q query positions at once, so the score / context GEMMs have a
+large M = G·T_q and high arithmetic intensity (~M flops/byte) — the
+workload is **compute-bound** (above the roofline ridge), unlike T_q=1
+decode (M=8, memory-bound). This is the regime where the composite
+command's value shows: it streams DMA↔compute per HW tile to keep the MAC
+array fed, while the primitive kernel serializes load→dot and starves it.
+
+Single-rank (C=P=1): no cross-device reduce — the focus is the per-PE
+GEMM-issue mechanism. FlashAttention 2-D tiling (Q-block × S_kv-tile,
+online softmax) bounds the TCM scratch.
+
+Three forms, differing only in how each Q-block's local attention is
+issued (placement/softmax identical):
+  primitive           tl.dot per S_kv-tile + primitive online merge.
+  composite           one coarse composite GEMM per Q-block over the whole
+                      S_kv (K, V as HBM refs → PE_SCHEDULER tiles/streams).
+  composite_extended  Q·Kᵀ composite + softmax_merge recipe composite.
+"""
+from __future__ import annotations
+
+from kernbench.benches.gqa_helpers.long_ctx._gqa_mlo_reduce import _merge_running
+
+M_BLOCK = 128       # query rows per FlashAttention Q-block (tile-filling).
+TILE_S_KV = 256     # primitive per-tile S_kv width.
+_SEED_S_KV = 64     # composite_extended recipe seed slice.
+
+
+def _qblock_bounds(T_q: int, h_q: int, h_kv: int):
+    G = h_q // h_kv
+    M_total = G * T_q
+    n_qblocks = (M_total + M_BLOCK - 1) // M_BLOCK
+    return G, M_total, n_qblocks
+
+
+# ── primitive: hand-tiled monolithic dots + online merge ─────────────
+
+
+def gqa_prefill_primitive_kernel(
+    q_ptr, k_ptr, v_ptr, o_ptr,
+    T_q, S_kv, h_q, h_kv, d_head, C, P, *, tl,
+) -> None:
+    G, M_total, n_qblocks = _qblock_bounds(T_q, h_q, h_kv)
+    ROW = d_head * 2  # f16
+    n_tiles = (S_kv + TILE_S_KV - 1) // TILE_S_KV
+
+    for qb in range(n_qblocks):
+        # Per-Q-block scratch_scope: this block's running (m,ℓ,O) plus its
+        # tile-0 transients are freed before the next block, so scratch
+        # stays O(one block) rather than accumulating across all blocks.
+        with tl.scratch_scope():
+            m_blk = min(M_BLOCK, M_total - qb * M_BLOCK)
+            q_off = qb * M_BLOCK * ROW
+            Q = tl.load(q_ptr + q_off, shape=(m_blk, d_head), dtype="f16")
+
+            tile_s0 = min(TILE_S_KV, S_kv)
+            K_T = tl.load(k_ptr, shape=(d_head, tile_s0), dtype="f16")
+            V = tl.load(v_ptr, shape=(tile_s0, d_head), dtype="f16")
+            scores = tl.dot(Q, K_T)
+            m_local = tl.max(scores, axis=-1)
+            exp_s = tl.exp(scores - m_local)
+            l_local = tl.sum(exp_s, axis=-1)
+            O_local = tl.dot(exp_s, V)
+
+            for ti in range(1, n_tiles):
+                start = ti * TILE_S_KV
+                tile_s = min(TILE_S_KV, S_kv - start)
+                with tl.scratch_scope():
+                    K_T_t = tl.load(k_ptr + start * ROW,
+                                    shape=(d_head, tile_s), dtype="f16")
+                    V_t = tl.load(v_ptr + start * ROW,
+                                  shape=(tile_s, d_head), dtype="f16")
+                    scores_t = tl.dot(Q, K_T_t)
+                    m_tile = tl.max(scores_t, axis=-1)
+                    exp_t = tl.exp(scores_t - m_tile)
+                    l_tile = tl.sum(exp_t, axis=-1)
+                    O_tile = tl.dot(exp_t, V_t)
+                    m_new, l_new, O_new = _merge_running(
+                        m_local, l_local, O_local, m_tile, l_tile, O_tile, tl=tl,
+                    )
+                    tl.copy_to(m_local, m_new)
+                    tl.copy_to(l_local, l_new)
+                    tl.copy_to(O_local, O_new)
+
+            O_final = O_local / l_local
+            tl.store(o_ptr + q_off, O_final)
+
+
+# ── composite: one coarse composite GEMM per Q-block ─────────────────
+
+
+def gqa_prefill_composite_kernel(
+    q_ptr, k_ptr, v_ptr, o_ptr,
+    T_q, S_kv, h_q, h_kv, d_head, C, P, *, tl,
+) -> None:
+    G, M_total, n_qblocks = _qblock_bounds(T_q, h_q, h_kv)
+    ROW = d_head * 2
+
+    for qb in range(n_qblocks):
+        with tl.scratch_scope():
+            m_blk = min(M_BLOCK, M_total - qb * M_BLOCK)
+            q_off = qb * M_BLOCK * ROW
+            Q = tl.load(q_ptr + q_off, shape=(m_blk, d_head), dtype="f16")
+            K_T = tl.ref(k_ptr, shape=(d_head, S_kv), dtype="f16")
+            V = tl.ref(v_ptr, shape=(S_kv, d_head), dtype="f16")
+
+            scores = tl.composite(op="gemm", a=Q, b=K_T)      # Q·Kᵀ, one command
+            m_local = tl.max(scores, axis=-1)
+            exp_s = tl.exp(scores - m_local)
+            l_local = tl.sum(exp_s, axis=-1)
+            O_local = tl.zeros((m_blk, d_head), dtype="f16")
+            tl.composite(op="gemm", a=exp_s, b=V, out=O_local)  # P·V, one command
+
+            O_final = O_local / l_local
+            tl.store(o_ptr + q_off, O_final)
+
+
+# ── composite_extended: Q·Kᵀ composite + softmax_merge recipe ────────
+
+
+def gqa_prefill_composite_ext_kernel(
+    q_ptr, k_ptr, v_ptr, o_ptr,
+    T_q, S_kv, h_q, h_kv, d_head, C, P, *, tl,
+) -> None:
+    G, M_total, n_qblocks = _qblock_bounds(T_q, h_q, h_kv)
+    ROW = d_head * 2
+
+    for qb in range(n_qblocks):
+        with tl.scratch_scope():
+            m_blk = min(M_BLOCK, M_total - qb * M_BLOCK)
+            q_off = qb * M_BLOCK * ROW
+            Q = tl.load(q_ptr + q_off, shape=(m_blk, d_head), dtype="f16")
+
+            seed = min(_SEED_S_KV, S_kv)
+            K_T0 = tl.load(k_ptr, shape=(d_head, seed), dtype="f16")
+            V0 = tl.load(v_ptr, shape=(seed, d_head), dtype="f16")
+            scores0 = tl.dot(Q, K_T0)
+            m_local = tl.max(scores0, axis=-1)
+            exp0 = tl.exp(scores0 - m_local)
+            l_local = tl.sum(exp0, axis=-1)
+            O_local = tl.dot(exp0, V0)
+
+            rest = S_kv - seed
+            if rest > 0:
+                K_T1 = tl.ref(k_ptr + seed * ROW,
+                              shape=(d_head, rest), dtype="f16")
+                V1 = tl.ref(v_ptr + seed * ROW,
+                            shape=(rest, d_head), dtype="f16")
+                scores1 = tl.composite(op="gemm", a=Q, b=K_T1)
+                tl.composite(
+                    prologue=[{"op": "softmax_merge", "s": scores1,
+                               "m": m_local, "l": l_local, "O": O_local}],
+                    op="gemm", b=V1, out=O_local,
+                    epilogue=[{"op": "add", "other": O_local}],
+                )
+
+            O_final = O_local / l_local
+            tl.store(o_ptr + q_off, O_final)

src/kernbench/benches/gqa_helpers/long_ctx/gqa_decode_long_ctx_4cases.py

@@ -38,6 +38,12 @@ from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_
 from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_sp_pe_tp import (
     gqa_attention_decode_long_ctx_cube_sp_pe_tp_kernel,
 )
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_sp_pe_tp_dhead import (
+    gqa_attention_decode_long_ctx_cube_sp_pe_tp_dhead_kernel,
+)
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_tp_dhead_pe_sp import (
+    gqa_attention_decode_long_ctx_cube_tp_dhead_pe_sp_kernel,
+)
 from kernbench.benches.gqa_helpers.shared._gqa_panel_helpers import (
     _ccl_cfg,
     _summarize_op_log,
@@ -58,10 +64,12 @@ _SWEEP_JSON = _OUTPUT_DIR / "sweep_decode.json"
 
 
 _PANELS = (
-    "single_kv_group_decode_long_ctx_gqa_cube_sp_pe_sp",      # Case 4 ★ optimal
+    "single_kv_group_decode_long_ctx_gqa_cube_sp_pe_sp",         # Case 6 ★ optimal (S_kv-SP both axes; was "Case 4" in slide-17 numbering)
-    "single_kv_group_decode_long_ctx_gqa_cube_repl_pe_tp",    # Case 2 (no comm; 8× memory)
+    "single_kv_group_decode_long_ctx_gqa_cube_repl_pe_tp",       # Case 1 (no sharding; 40 GB / PE @ 1M)
-    "single_kv_group_decode_long_ctx_gqa_cube_repl_pe_sp",    # Case 3 (intra-CUBE AR only; 8× memory)
+    "single_kv_group_decode_long_ctx_gqa_cube_repl_pe_sp",       # Case 3 (intra-cube AR on (m,ℓ,O); 5 GB / PE)
-    "single_kv_group_decode_long_ctx_gqa_cube_sp_pe_tp",      # Case 1 (inter-CUBE lrab only; PE-TP B=1 waste)
+    "single_kv_group_decode_long_ctx_gqa_cube_sp_pe_tp",         # Case 2 (inter-cube AR on (m,ℓ,O); 5 GB / PE)
+    "single_kv_group_decode_long_ctx_gqa_cube_sp_pe_tp_dhead",   # Case 4 (Cube-SP × PE-TP-on-d_head — d_head-TP partial-score AR intra-cube)
+    "single_kv_group_decode_long_ctx_gqa_cube_tp_dhead_pe_sp",   # Case 5 (Cube-TP-on-d_head × PE-SP — d_head-TP partial-score AR inter-cube)
 )
 
 # Each entry: (kind, panel-specific params).
@@ -102,6 +110,23 @@ _PANEL_DISPATCH: dict[str, tuple[str, dict]] = {
         "T_q": 1, "S_kv": 8_192,
         "d_head": 128, "h_q": 8, "h_kv": 1,
     }),
+    "single_kv_group_decode_long_ctx_gqa_cube_sp_pe_tp_dhead": ("decode_long_ctx_cube_sp_pe_tp_dhead", {
+        # Case 4 (new numbering): Cube-SP on S_kv × PE-TP on d_head.
+        # Partial-score AR happens intra-cube (NoC); (m,ℓ,O) merge happens
+        # inter-cube (UCIe). Memory-equal to Case 6; comm dominated by
+        # partial-score AR which scales with S_kv.
+        "C": 8, "P": 8,
+        "T_q": 1, "S_kv": 8_192,
+        "d_head": 128, "h_q": 8, "h_kv": 1,
+    }),
+    "single_kv_group_decode_long_ctx_gqa_cube_tp_dhead_pe_sp": ("decode_long_ctx_cube_tp_dhead_pe_sp", {
+        # Case 5 (new numbering): Cube-TP on d_head × PE-SP on S_kv.
+        # Mirror of Case 4: partial-score AR happens INTER-cube (UCIe);
+        # (m,ℓ,O) merge happens INTRA-cube (NoC). Memory-equal to Case 6.
+        "C": 8, "P": 8,
+        "T_q": 1, "S_kv": 8_192,
+        "d_head": 128, "h_q": 8, "h_kv": 1,
+    }),
 }
 
 
@@ -238,6 +263,74 @@ def _run_decode_panel_long_ctx_cube_sp_pe_sp(
     )
 
 
+def _run_decode_panel_long_ctx_cube_sp_pe_tp_dhead(
+    ctx, *, panel: str, C: int, P: int,
+    T_q: int, S_kv: int,
+    d_head: int, h_q: int, h_kv: int,
+) -> None:
+    """Case 4 (new) runner: Cube-SP × PE-TP-on-d_head.
+
+    Q is cube-replicated, PE-sharded column-wise (d_head/P per PE).
+    K, V are cube=row_wise on S_kv, pe=column_wise on d_head — each PE
+    owns (S_kv/C, d_head/P). The kernel does an intra-cube AR-SUM of
+    partial scores then an inter-cube online-softmax merge of
+    (m, ℓ, O_shard).
+    """
+    configure_sfr_intercube_multisip(ctx.engine, ctx.spec, _ccl_cfg())
+    dp_q = DPPolicy(cube="replicate", pe="column_wise",
+                    num_cubes=C, num_pes=P)
+    dp_kv = DPPolicy(cube="row_wise", pe="column_wise",
+                     num_cubes=C, num_pes=P)
+    q = ctx.zeros((T_q, h_q * d_head),
+                  dtype="f16", dp=dp_q, name=f"{panel}_q")
+    k = ctx.zeros((S_kv, h_kv * d_head),
+                  dtype="f16", dp=dp_kv, name=f"{panel}_k")
+    v = ctx.zeros((S_kv, h_kv * d_head),
+                  dtype="f16", dp=dp_kv, name=f"{panel}_v")
+    o = ctx.empty((T_q, h_q * d_head),
+                  dtype="f16", dp=dp_q, name=f"{panel}_o")
+    ctx.launch(
+        panel, gqa_attention_decode_long_ctx_cube_sp_pe_tp_dhead_kernel,
+        q, k, v, o,
+        T_q, S_kv, h_q, h_kv, d_head, C, P,
+        _auto_dim_remap=False,
+    )
+
+
+def _run_decode_panel_long_ctx_cube_tp_dhead_pe_sp(
+    ctx, *, panel: str, C: int, P: int,
+    T_q: int, S_kv: int,
+    d_head: int, h_q: int, h_kv: int,
+) -> None:
+    """Case 5 (new) runner: Cube-TP-on-d_head × PE-SP-on-S_kv.
+
+    Q is cube-sharded column_wise (d_head/C per cube), pe-replicated
+    within cube. K, V are cube=column_wise on d_head, pe=row_wise on
+    S_kv — each PE owns (S_kv/P, d_head/C). The kernel does an
+    inter-cube AR-SUM of partial scores then an intra-cube online-
+    softmax merge of (m, ℓ, O) reduce-to-PE0.
+    """
+    configure_sfr_intercube_multisip(ctx.engine, ctx.spec, _ccl_cfg())
+    dp_q = DPPolicy(cube="column_wise", pe="replicate",
+                    num_cubes=C, num_pes=P)
+    dp_kv = DPPolicy(cube="column_wise", pe="row_wise",
+                     num_cubes=C, num_pes=P)
+    q = ctx.zeros((T_q, h_q * d_head),
+                  dtype="f16", dp=dp_q, name=f"{panel}_q")
+    k = ctx.zeros((S_kv, h_kv * d_head),
+                  dtype="f16", dp=dp_kv, name=f"{panel}_k")
+    v = ctx.zeros((S_kv, h_kv * d_head),
+                  dtype="f16", dp=dp_kv, name=f"{panel}_v")
+    o = ctx.empty((T_q, h_q * d_head),
+                  dtype="f16", dp=dp_q, name=f"{panel}_o")
+    ctx.launch(
+        panel, gqa_attention_decode_long_ctx_cube_tp_dhead_pe_sp_kernel,
+        q, k, v, o,
+        T_q, S_kv, h_q, h_kv, d_head, C, P,
+        _auto_dim_remap=False,
+    )
+
+
 def _make_bench_fn(panel: str):
     kind, params = _PANEL_DISPATCH[panel]
 
@@ -250,6 +343,10 @@ def _make_bench_fn(panel: str):
             _run_decode_panel_long_ctx_cube_repl_pe_sp(ctx, panel=panel, **params)
         elif kind == "decode_long_ctx_cube_sp_pe_tp":
             _run_decode_panel_long_ctx_cube_sp_pe_tp(ctx, panel=panel, **params)
+        elif kind == "decode_long_ctx_cube_sp_pe_tp_dhead":
+            _run_decode_panel_long_ctx_cube_sp_pe_tp_dhead(ctx, panel=panel, **params)
+        elif kind == "decode_long_ctx_cube_tp_dhead_pe_sp":
+            _run_decode_panel_long_ctx_cube_tp_dhead_pe_sp(ctx, panel=panel, **params)
         else:
             raise RuntimeError(
                 f"milestone-gqa-decode-long-ctx-4cases panel {panel!r} has "

src/kernbench/benches/gqa_helpers/long_ctx/gqa_decode_long_ctx_composite.py

@@ -0,0 +1,182 @@
+"""milestone-1h-gqa decode: Case-6 composite-command study.
+
+Three command-form variants of the Case-6 (Cube-SP × PE-SP) long-context
+decode kernel, swept over S_kv so the local-attention tile loop runs
+multiple tiles (S_local = S_kv/(C·P) > TILE_S_KV = 1024):
+
+    primitive           tl.dot + primitive online-softmax merge (baseline).
+    composite           per-tile GEMMs as tl.composite GEMM commands.
+    composite_extended  per-tile attention as a Q·Kᵀ composite + a
+                        softmax_merge recipe composite (ADR-0065).
+
+The placement, DP policy, and (m, ℓ, O) reduce are identical across
+variants; only the per-tile command form differs. The sweep writes
+per-(variant, S_kv) latency, engine occupancy, and PE_CPU command count
+to sweep_decode_composite.json so the comparative plot
+(paper_plot_gqa_decode_long_ctx_composite.py) can read off the
+CPU-offload win.
+
+Run in op_log mode (enable_data=False): latency / dispatch only — recipe
+data-mode numeric parity is a separate follow-up (DDD-0065).
+"""
+from __future__ import annotations
+
+import json
+from pathlib import Path
+
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_sp_pe_sp import (  # noqa: E501
+    gqa_attention_decode_long_ctx_cube_sp_pe_sp_kernel,
+)
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_sp_pe_sp_composite import (  # noqa: E501
+    gqa_attention_decode_long_ctx_cube_sp_pe_sp_composite_kernel,
+)
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_sp_pe_sp_composite_ext import (  # noqa: E501
+    gqa_attention_decode_long_ctx_cube_sp_pe_sp_composite_ext_kernel,
+)
+from kernbench.benches.gqa_helpers.shared._gqa_panel_helpers import _ccl_cfg
+from kernbench.ccl.sfr_config import configure_sfr_intercube_multisip
+from kernbench.policy.placement.dp import DPPolicy
+
+_OUTPUT_DIR = (
+    Path(__file__).resolve().parents[2]
+    / "1H_milestone_output" / "gqa" / "long_ctx"
+)
+_SWEEP_JSON = _OUTPUT_DIR / "sweep_decode_composite.json"
+
+
+# ── Variant + S_kv registry ──────────────────────────────────────────
+
+
+# Each kernel implements the same Case-6 placement; only the per-tile
+# command form differs (see module docstring).
+_VARIANT_KERNELS = {
+    "primitive": gqa_attention_decode_long_ctx_cube_sp_pe_sp_kernel,
+    "composite": gqa_attention_decode_long_ctx_cube_sp_pe_sp_composite_kernel,
+    "composite_extended":
+        gqa_attention_decode_long_ctx_cube_sp_pe_sp_composite_ext_kernel,
+}
+_VARIANTS = ("primitive", "composite", "composite_extended")
+
+# Op-count (PE_CPU dispatch) is computed at emit time (exact, instant), so
+# it spans up to the 1M production point (S_local = 1M/64 = 16384, 16
+# tiles): the per-PE work is Q·Kᵀ (8,128)·(128,16384) and P·V
+# (8,16384)·(16384,128). End-to-end latency needs the data-mode engine,
+# whose cost scales with S_kv, so it is swept only over a tractable range.
+_S_KV_OPCOUNT = (8192, 65_536, 131_072, 262_144, 524_288, 1_048_576)
+_S_KV_LATENCY = (8192, 32_768, 65_536, 131_072)
+
+# LLaMA-3.1-70B single-KV-head group (8 cubes × 8 PEs), one decode step.
+_BASE_PARAMS = dict(C=8, P=8, T_q=1, d_head=128, h_q=8, h_kv=1)
+
+
+# ── Per-(variant, S_kv) runner ───────────────────────────────────────
+
+
+def _run_panel_fn(variant: str, S_kv: int):
+    kernel = _VARIANT_KERNELS[variant]
+    p = _BASE_PARAMS
+    panel = f"decode_long_ctx_composite_{variant}_s{S_kv}"
+
+    def _bench_fn(ctx):
+        configure_sfr_intercube_multisip(ctx.engine, ctx.spec, _ccl_cfg())
+        dp_full = DPPolicy(cube="replicate", pe="replicate",
+                           num_cubes=p["C"], num_pes=p["P"])
+        dp_kv = DPPolicy(cube="row_wise", pe="row_wise",
+                         num_cubes=p["C"], num_pes=p["P"])
+        q = ctx.zeros((p["T_q"], p["h_q"] * p["d_head"]),
+                      dtype="f16", dp=dp_full, name=f"{panel}_q")
+        k = ctx.zeros((S_kv, p["h_kv"] * p["d_head"]),
+                      dtype="f16", dp=dp_kv, name=f"{panel}_k")
+        v = ctx.zeros((S_kv, p["h_kv"] * p["d_head"]),
+                      dtype="f16", dp=dp_kv, name=f"{panel}_v")
+        o = ctx.empty((p["T_q"], p["h_q"] * p["d_head"]),
+                      dtype="f16", dp=dp_full, name=f"{panel}_o")
+        ctx.launch(panel, kernel, q, k, v, o,
+                   p["T_q"], S_kv, p["h_q"], p["h_kv"], p["d_head"],
+                   p["C"], p["P"], _auto_dim_remap=False)
+
+    return _bench_fn
+
+
+def _end_to_end_ns(op_log) -> float:
+    if not op_log:
+        return 0.0
+    return max(r.t_end for r in op_log) - min(r.t_start for r in op_log)
+
+
+def _emit_dispatch(variant: str, S_kv: int) -> tuple[int, float]:
+    """PE_CPU dispatch (# commands, summed cycles) the kernel emits at the
+    lrab center rank (cube 6, pe 0) — computed at command-emit time, so it
+    is exact and S_kv-independent for the composite forms (no engine)."""
+    from kernbench.common.pe_commands import PeCpuOverheadCmd
+    from kernbench.common.pe_cost_model import DEFAULT_PE_COST_MODEL
+    from kernbench.triton_emu.tl_context import TLContext, run_kernel
+
+    p = _BASE_PARAMS
+    tl = TLContext(
+        pe_id=0, num_programs=p["P"], cost_model=DEFAULT_PE_COST_MODEL,
+        cube_id=6, num_cubes=p["C"], scratch_base=1 << 61, scratch_size=1 << 20,
+    )
+    run_kernel(
+        _VARIANT_KERNELS[variant], tl,
+        0x1000, 0x2000, 0x3000, 0x4000,
+        p["T_q"], S_kv, p["h_q"], p["h_kv"], p["d_head"], p["C"], p["P"],
+    )
+    cmds = [c for c in tl.commands if isinstance(c, PeCpuOverheadCmd)]
+    return len(cmds), sum(c.cycles for c in cmds)
+
+
+def _engine_latency_ns(variant: str, S_kv: int, topology: str) -> float:
+    from kernbench.runtime_api.bench_runner import run_bench
+    from kernbench.runtime_api.types import resolve_device
+    from kernbench.sim_engine.engine import GraphEngine
+    from kernbench.topology.builder import resolve_topology
+
+    topo = resolve_topology(topology)
+    result = run_bench(
+        topology=topo, bench_fn=_run_panel_fn(variant, S_kv),
+        device=resolve_device(None),
+        engine_factory=lambda t, d: GraphEngine(
+            getattr(t, "topology_obj", t), enable_data=True,
+        ),
+    )
+    if not result.completion.ok:
+        raise RuntimeError(
+            f"gqa-decode-composite {variant}@{S_kv} failed: {result.completion}"
+        )
+    return _end_to_end_ns(result.engine.op_log)
+
+
+# ── Sweep entry (called by the umbrella milestone_1h_gqa) ────────────
+
+
+def run_sweep(topology: str = "topology.yaml") -> int:
+    """Emit-level dispatch over the full S_kv range (to 1M) + engine
+    latency over the tractable range; write sweep.json. Returns row count."""
+    _OUTPUT_DIR.mkdir(parents=True, exist_ok=True)
+    rows = []
+    for S_kv in _S_KV_OPCOUNT:
+        for variant in _VARIANTS:
+            n_cmds, cycles = _emit_dispatch(variant, S_kv)
+            latency = (
+                _engine_latency_ns(variant, S_kv, topology)
+                if S_kv in _S_KV_LATENCY else None
+            )
+            rows.append({
+                "variant": variant,
+                "S_kv": S_kv,
+                **_BASE_PARAMS,
+                "pe_cpu_cmd_count": n_cmds,
+                "pe_cpu_dispatch_cycles": cycles,
+                "latency_ns": latency,
+            })
+    sweep = {
+        "version": 2,
+        "variants": list(_VARIANTS),
+        "s_kv_opcount": list(_S_KV_OPCOUNT),
+        "s_kv_latency": list(_S_KV_LATENCY),
+        "rows": rows,
+    }
+    _SWEEP_JSON.write_text(json.dumps(sweep, indent=2))
+    print(f"  gqa-decode-composite: {len(rows)} rows -> {_SWEEP_JSON}")
+    return len(rows)

src/kernbench/benches/gqa_helpers/long_ctx/gqa_decode_streaming.py

@@ -0,0 +1,168 @@
+"""milestone-1h-gqa: memory-bound decode streaming composite-command study.
+
+Single-rank companion to the compute-bound prefill study
+(``gqa_prefill_compute_bound``). Same three command-form kernels
+(``_gqa_prefill_compute_bound``), same single-rank (C=P=1) harness, but
+driven with **T_q=1** (decode) and swept over a large S_kv. With T_q=1 the
+score / context GEMMs are skinny (M = G·T_q = 8), so the MAC array is
+barely fed and the kernel is **memory-bound** — streaming the KV cache out
+of HBM dominates. This is the regime mirror of prefill: command form is
+*latency-neutral* here, because the bottleneck is data movement, not
+issue.
+
+Running single-rank (no cross-CUBE (m,ℓ,O) reduce) isolates the
+local-attention streaming vs. issue trade-off cleanly — unlike the 64-way
+Case-6 decode sweep, whose small-S_kv end-to-end latency is masked by the
+inter-CUBE reduce tail. S_kv here is the *per-rank* context, so S_kv=64K
+single-rank is the per-PE load of a 4M-token, 64-way-sharded decode.
+
+Records per (variant, S_kv) the end-to-end latency, GEMM/DMA engine busy
+time, MAC utilization (achieved ÷ peak), and DMA occupancy (dma_busy ÷
+e2e), so the comparative plot can show all three command forms landing on
+the same latency curve while the MAC array stays floored and the DMA
+channel stays saturated.
+
+Runs in data mode (engine latency). Gated via the umbrella
+``GQA_1H_SWEEPS=decode_streaming``.
+"""
+from __future__ import annotations
+
+import json
+from pathlib import Path
+
+from kernbench.benches.gqa_helpers.long_ctx._gqa_prefill_compute_bound import (
+    gqa_prefill_composite_ext_kernel,
+    gqa_prefill_composite_kernel,
+    gqa_prefill_primitive_kernel,
+)
+from kernbench.policy.placement.dp import DPPolicy
+
+_OUTPUT_DIR = (
+    Path(__file__).resolve().parents[2]
+    / "1H_milestone_output" / "gqa" / "long_ctx"
+)
+_SWEEP_JSON = _OUTPUT_DIR / "sweep_decode_streaming.json"
+
+# Same kernels as the prefill study — they are general attention kernels;
+# the regime is set by T_q (=1 here → memory-bound).
+_VARIANT_KERNELS = {
+    "primitive": gqa_prefill_primitive_kernel,
+    "composite": gqa_prefill_composite_kernel,
+    "composite_extended": gqa_prefill_composite_ext_kernel,
+}
+_VARIANTS = ("primitive", "composite", "composite_extended")
+
+# Per-rank context length. T_q=1, so M = G·T_q = 8 (skinny, memory-bound)
+# at every point. Capped at 16K: the *plain* composite materializes the
+# full (M, S_kv) scores in TCM scratch (~48·S_kv B incl. the exp transients),
+# which overflows the 1 MB kernel scratch beyond ~21K — itself a sign that
+# the recipe (composite_extended), which tiles the softmax, is what long
+# context actually needs. S_kv is per-rank, so 16K single-rank already
+# equals the per-PE load of a 1M-token, 64-way-sharded decode.
+_S_KV_POINTS = (2048, 4096, 8192, 16384)
+
+_T_Q = 1
+_H_Q, _H_KV, _D_HEAD = 8, 1, 128
+_PEAK_TFLOPS = 8.0  # per-PE f16 GEMM peak (topology.yaml pe_gemm.peak_tflops_f16)
+
+
+def _run_panel_fn(variant: str, s_kv: int):
+    kernel = _VARIANT_KERNELS[variant]
+    panel = f"decode_stream_{variant}_s{s_kv}"
+
+    def _bench_fn(ctx):
+        dp = DPPolicy(cube="replicate", pe="replicate",
+                      num_cubes=1, num_pes=1)
+        q = ctx.zeros((_T_Q, _H_Q * _D_HEAD),
+                      dtype="f16", dp=dp, name=f"{panel}_q")
+        k = ctx.zeros((s_kv, _H_KV * _D_HEAD),
+                      dtype="f16", dp=dp, name=f"{panel}_k")
+        v = ctx.zeros((s_kv, _H_KV * _D_HEAD),
+                      dtype="f16", dp=dp, name=f"{panel}_v")
+        o = ctx.empty((_T_Q, _H_Q * _D_HEAD),
+                      dtype="f16", dp=dp, name=f"{panel}_o")
+        ctx.launch(panel, kernel, q, k, v, o,
+                   _T_Q, s_kv, _H_Q, _H_KV, _D_HEAD, 1, 1,
+                   _auto_dim_remap=False)
+
+    return _bench_fn
+
+
+def _end_to_end_ns(op_log) -> float:
+    if not op_log:
+        return 0.0
+    return max(r.t_end for r in op_log) - min(r.t_start for r in op_log)
+
+
+def _engine_busy_ns(op_log, suffix: str) -> float:
+    return sum(r.t_end - r.t_start
+               for r in op_log if r.component_id.endswith("." + suffix))
+
+
+def _run_panel(variant: str, s_kv: int, topology: str) -> dict:
+    from kernbench.runtime_api.bench_runner import run_bench
+    from kernbench.runtime_api.types import resolve_device
+    from kernbench.sim_engine.engine import GraphEngine
+    from kernbench.topology.builder import resolve_topology
+
+    topo = resolve_topology(topology)
+    result = run_bench(
+        topology=topo, bench_fn=_run_panel_fn(variant, s_kv),
+        device=resolve_device(None),
+        engine_factory=lambda t, d: GraphEngine(
+            getattr(t, "topology_obj", t), enable_data=True,
+        ),
+    )
+    if not result.completion.ok:
+        raise RuntimeError(
+            f"gqa-decode-streaming {variant}@{s_kv} failed: {result.completion}"
+        )
+    op_log = result.engine.op_log
+    e2e = _end_to_end_ns(op_log)
+    gemm = _engine_busy_ns(op_log, "pe_gemm")
+    dma = _engine_busy_ns(op_log, "pe_dma")
+    G = _H_Q // _H_KV
+    M = G * _T_Q
+    # Useful attention flops (Q·Kᵀ + P·V), single rank.
+    useful_flops = 4.0 * M * _D_HEAD * s_kv
+    # KV-cache bytes streamed from HBM (K + V, f16) — the memory-bound
+    # denominator. achieved_bw = bytes / e2e (GB/s) measures how close the
+    # command form gets to the HBM roofline (the memory-bound mirror of
+    # MAC utilization in the compute-bound prefill study).
+    kv_bytes = 2.0 * s_kv * _D_HEAD * 2.0
+    return {
+        "variant": variant,
+        "s_kv": s_kv,
+        "M": M,
+        "latency_ns": e2e,
+        "gemm_busy_ns": gemm,
+        "dma_busy_ns": dma,
+        "kv_bytes": kv_bytes,
+        "achieved_bw_gbs": (kv_bytes / e2e) if e2e > 0 else 0.0,
+        "achieved_tflops": (useful_flops / e2e / 1e3) if e2e > 0 else 0.0,
+        "mac_util": (useful_flops / e2e / 1e3 / _PEAK_TFLOPS) if e2e > 0 else 0.0,
+        "dma_occupancy": (dma / e2e) if e2e > 0 else 0.0,
+    }
+
+
+def run_sweep(topology: str = "topology.yaml") -> int:
+    """Drive all (variant, S_kv) decode-streaming panels; write sweep.json."""
+    _OUTPUT_DIR.mkdir(parents=True, exist_ok=True)
+    rows = [
+        _run_panel(variant, s_kv, topology)
+        for s_kv in _S_KV_POINTS
+        for variant in _VARIANTS
+    ]
+    sweep = {
+        "version": 1,
+        "variants": list(_VARIANTS),
+        "s_kv_points": list(_S_KV_POINTS),
+        "rows": rows,
+    }
+    _SWEEP_JSON.write_text(json.dumps(sweep, indent=2))
+    print(f"  gqa-decode-streaming: {len(rows)} rows -> {_SWEEP_JSON}")
+    return len(rows)
+
+
+if __name__ == "__main__":
+    run_sweep()

src/kernbench/benches/gqa_helpers/long_ctx/gqa_prefill_compute_bound.py

@@ -0,0 +1,142 @@
+"""milestone-1h-gqa: compute-bound prefill composite-command study.
+
+Three command-form variants of a single-rank compute-bound prefill
+attention kernel (``_gqa_prefill_compute_bound``), swept over context
+length S_kv = T_q. Unlike the memory-bound decode study, prefill has a
+large M = G·T_q, so the score / context GEMMs are compute-bound — the
+regime where the composite command keeps the MAC array fed (DMA↔compute
+pipelining) and the hand-tiled primitive starves it on the serial
+load→dot path.
+
+Records per (variant, context) the end-to-end latency, the GEMM-engine
+busy time, and the MAC occupancy (gemm_busy / e2e) so the comparative
+plot can show composite winning on both latency and utilization.
+
+Runs in data mode (engine latency). Gated via the umbrella
+``GQA_1H_SWEEPS=prefill_cb``.
+"""
+from __future__ import annotations
+
+import json
+from pathlib import Path
+
+from kernbench.benches.gqa_helpers.long_ctx._gqa_prefill_compute_bound import (
+    gqa_prefill_composite_ext_kernel,
+    gqa_prefill_composite_kernel,
+    gqa_prefill_primitive_kernel,
+)
+from kernbench.policy.placement.dp import DPPolicy
+
+_OUTPUT_DIR = (
+    Path(__file__).resolve().parents[2]
+    / "1H_milestone_output" / "gqa" / "long_ctx"
+)
+_SWEEP_JSON = _OUTPUT_DIR / "sweep_prefill_compute_bound.json"
+
+_VARIANT_KERNELS = {
+    "primitive": gqa_prefill_primitive_kernel,
+    "composite": gqa_prefill_composite_kernel,
+    "composite_extended": gqa_prefill_composite_ext_kernel,
+}
+_VARIANTS = ("primitive", "composite", "composite_extended")
+
+# Context length S_kv = T_q (prefill processes T_q tokens against S_kv=T_q
+# keys). M = G·T_q = 8·T_q is tile-filling/compute-bound at every point.
+_CTX_POINTS = (256, 512, 1024)
+
+_H_Q, _H_KV, _D_HEAD = 8, 1, 128
+_PEAK_TFLOPS = 8.0  # per-PE f16 GEMM peak (topology.yaml pe_gemm.peak_tflops_f16)
+
+
+def _run_panel_fn(variant: str, ctx_len: int):
+    kernel = _VARIANT_KERNELS[variant]
+    panel = f"prefill_cb_{variant}_c{ctx_len}"
+
+    def _bench_fn(ctx):
+        dp = DPPolicy(cube="replicate", pe="replicate",
+                      num_cubes=1, num_pes=1)
+        q = ctx.zeros((ctx_len, _H_Q * _D_HEAD),
+                      dtype="f16", dp=dp, name=f"{panel}_q")
+        k = ctx.zeros((ctx_len, _H_KV * _D_HEAD),
+                      dtype="f16", dp=dp, name=f"{panel}_k")
+        v = ctx.zeros((ctx_len, _H_KV * _D_HEAD),
+                      dtype="f16", dp=dp, name=f"{panel}_v")
+        o = ctx.empty((ctx_len, _H_Q * _D_HEAD),
+                      dtype="f16", dp=dp, name=f"{panel}_o")
+        ctx.launch(panel, kernel, q, k, v, o,
+                   ctx_len, ctx_len, _H_Q, _H_KV, _D_HEAD, 1, 1,
+                   _auto_dim_remap=False)
+
+    return _bench_fn
+
+
+def _end_to_end_ns(op_log) -> float:
+    if not op_log:
+        return 0.0
+    return max(r.t_end for r in op_log) - min(r.t_start for r in op_log)
+
+
+def _engine_busy_ns(op_log, suffix: str) -> float:
+    return sum(r.t_end - r.t_start
+               for r in op_log if r.component_id.endswith("." + suffix))
+
+
+def _run_panel(variant: str, ctx_len: int, topology: str) -> dict:
+    from kernbench.runtime_api.bench_runner import run_bench
+    from kernbench.runtime_api.types import resolve_device
+    from kernbench.sim_engine.engine import GraphEngine
+    from kernbench.topology.builder import resolve_topology
+
+    topo = resolve_topology(topology)
+    result = run_bench(
+        topology=topo, bench_fn=_run_panel_fn(variant, ctx_len),
+        device=resolve_device(None),
+        engine_factory=lambda t, d: GraphEngine(
+            getattr(t, "topology_obj", t), enable_data=True,
+        ),
+    )
+    if not result.completion.ok:
+        raise RuntimeError(
+            f"gqa-prefill-cb {variant}@{ctx_len} failed: {result.completion}"
+        )
+    op_log = result.engine.op_log
+    e2e = _end_to_end_ns(op_log)
+    gemm = _engine_busy_ns(op_log, "pe_gemm")
+    dma = _engine_busy_ns(op_log, "pe_dma")
+    # Useful attention flops (Q·Kᵀ + P·V), single rank.
+    G = _H_Q // _H_KV
+    M = G * ctx_len
+    useful_flops = 4.0 * M * _D_HEAD * ctx_len
+    # MAC utilization = achieved / peak. ``achieved_tflops`` uses wall-clock
+    # (useful_flops / e2e) so it is bounded by peak even when the composite
+    # path overlaps many tile GEMMs (gemm_busy is a sum over overlapping ops
+    # and is kept only for diagnostics).
+    return {
+        "variant": variant,
+        "ctx_len": ctx_len,
+        "M": M,
+        "latency_ns": e2e,
+        "gemm_busy_ns": gemm,
+        "dma_busy_ns": dma,
+        "achieved_tflops": (useful_flops / e2e / 1e3) if e2e > 0 else 0.0,
+        "mac_util": (useful_flops / e2e / 1e3 / _PEAK_TFLOPS) if e2e > 0 else 0.0,
+    }
+
+
+def run_sweep(topology: str = "topology.yaml") -> int:
+    """Drive all (variant, context) prefill panels; write sweep.json."""
+    _OUTPUT_DIR.mkdir(parents=True, exist_ok=True)
+    rows = [
+        _run_panel(variant, ctx_len, topology)
+        for ctx_len in _CTX_POINTS
+        for variant in _VARIANTS
+    ]
+    sweep = {
+        "version": 1,
+        "variants": list(_VARIANTS),
+        "ctx_points": list(_CTX_POINTS),
+        "rows": rows,
+    }
+    _SWEEP_JSON.write_text(json.dumps(sweep, indent=2))
+    print(f"  gqa-prefill-cb: {len(rows)} rows -> {_SWEEP_JSON}")
+    return len(rows)

src/kernbench/benches/milestone_1h_gqa.py

@@ -8,12 +8,13 @@ Currently exercises (long-context only — short-context panels are
 future work):
   - 4-cases prefill comparative study (gqa_helpers.long_ctx.gqa_prefill_long_ctx_4cases)
   - 4-cases decode comparative study  (gqa_helpers.long_ctx.gqa_decode_long_ctx_4cases)
+  - Case-6 composite-command study    (gqa_helpers.long_ctx.gqa_decode_long_ctx_composite)
 
-Each sub-sweep writes its own ``sweep_{prefill,decode}.json`` into the
+Each sub-sweep writes its own ``sweep_*.json`` into the shared output
-shared output dir ``benches/1H_milestone_output/gqa/gqa_long_ctx/``.
+dir ``benches/1H_milestone_output/gqa/gqa_long_ctx/``.
 Selection via the env var ``GQA_1H_SWEEPS=prefill,decode`` (default
-runs both). Toggle individual sweeps with ``GQA_1H_SWEEPS=prefill``
+runs prefill+decode). The composite study is opt-in (it sweeps the
-or ``GQA_1H_SWEEPS=decode``.
+data-mode engine over several S_kv points): ``GQA_1H_SWEEPS=composite``.
 
 Gated by ``GQA_1H_RUN=1`` to keep CI fast.
 """
@@ -24,6 +25,15 @@ import os
 from kernbench.benches.gqa_helpers.long_ctx.gqa_decode_long_ctx_4cases import (
     run_sweep as _run_decode_sweep,
 )
+from kernbench.benches.gqa_helpers.long_ctx.gqa_decode_long_ctx_composite import (
+    run_sweep as _run_composite_sweep,
+)
+from kernbench.benches.gqa_helpers.long_ctx.gqa_decode_streaming import (
+    run_sweep as _run_decode_streaming_sweep,
+)
+from kernbench.benches.gqa_helpers.long_ctx.gqa_prefill_compute_bound import (
+    run_sweep as _run_prefill_cb_sweep,
+)
 from kernbench.benches.gqa_helpers.long_ctx.gqa_prefill_long_ctx_4cases import (
     run_sweep as _run_prefill_sweep,
 )
@@ -57,6 +67,9 @@ def run(torch) -> None:
     runners = {
         "prefill": _run_prefill_sweep,
         "decode": _run_decode_sweep,
+        "composite": _run_composite_sweep,
+        "prefill_cb": _run_prefill_cb_sweep,
+        "decode_streaming": _run_decode_streaming_sweep,
     }
     unknown = [s for s in sweeps if s not in runners]
     if unknown:

src/kernbench/triton_emu/tl_context.py

@@ -264,6 +264,10 @@ class TLContext:
             id=self._next_handle_id(),
             addr=addr, shape=shape, dtype=dtype,
             nbytes=nbytes, space="tcm",
+            # TCM-resident: a downstream composite operand reads it on-chip,
+            # not via a DMA_READ of its (bit-61) scratch address — same as
+            # the composite auto-output and recipe scratch handles below.
+            pinned=True,
         )
 
     # ── Reference (no DMA, metadata only) ────────────────────────

tests/attention/test_gqa_decode_long_ctx_composite.py

@@ -0,0 +1,191 @@
+"""Phase 1 tests for the Case-6 composite-command decode variants.
+
+Three command-form variants of the Case-6 (Cube-SP × PE-SP) long-context
+decode kernel are compared in the "Use of Composite Commands" study:
+
+  async              the current primitive kernel (``tl.dot`` + primitive
+                     online-softmax merge) — the measured baseline.
+  composite          per-tile GEMMs (Q·Kᵀ, P·V) issued as ``tl.composite``
+                     GEMM commands; softmax merge stays primitive.
+  composite_extended per-tile attention as two composites — Q·Kᵀ GEMM +
+                     a ``softmax_merge`` recipe composite folding the
+                     online merge and P·V (ADR-0065).
+
+This module currently holds the **refactor guard** only (Phase 1): it
+freezes the byte-level command stream of the current baseline kernel so
+the Phase-2 extraction of the shared (m,ℓ,O) reduce into a helper is
+provably behavior-preserving. The variant kernels, the dispatch-ordering
+test, and the e2e op_log test land in Phase 2 alongside the production
+kernels.
+"""
+from __future__ import annotations
+
+import hashlib
+from pathlib import Path
+
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_sp_pe_sp import (  # noqa: E501
+    gqa_attention_decode_long_ctx_cube_sp_pe_sp_kernel as _baseline,
+)
+from kernbench.triton_emu.tl_context import TLContext, run_kernel
+
+_TOPOLOGY = Path(__file__).resolve().parents[2] / "topology.yaml"
+
+# C=8 CUBEs × P=8 PEs single-KV-head group; S_kv chosen so
+# S_local = S_kv/(C·P) = 2048 > TILE_S_KV (1024) → the local-attention
+# tile loop runs 2 tiles, exercising the per-tile merge path that the
+# composite variants restructure.
+_C, _P = 8, 8
+_S_KV = 131_072
+_D_HEAD, _H_Q, _H_KV, _T_Q = 128, 8, 1, 1
+
+# Volatile per-handle identifiers (assigned by an internal counter); the
+# refactor must not change *structure*, but raw ids are not part of the
+# behavioral contract, so they are normalized out of the signature.
+_VOLATILE_FIELDS = frozenset({"id", "completion", "cid"})
+
+
+def _norm_value(v):
+    """Normalize a command field to an id-free, hashable form."""
+    # TensorHandle / CompletionHandle-like: keep shape/addr/space/dir, drop id.
+    if hasattr(v, "shape") and hasattr(v, "addr"):
+        return ("H", getattr(v, "shape", None), getattr(v, "addr", None),
+                getattr(v, "space", None), getattr(v, "dtype", None))
+    if isinstance(v, (list, tuple)):
+        return tuple(_norm_value(x) for x in v)
+    if isinstance(v, dict):
+        return tuple(sorted((k, _norm_value(x)) for k, x in v.items()))
+    return v
+
+
+def _cmd_signature(cmd) -> tuple:
+    fields = getattr(cmd, "__dict__", {})
+    return (
+        type(cmd).__name__,
+        tuple(sorted(
+            (name, _norm_value(val))
+            for name, val in fields.items()
+            if name not in _VOLATILE_FIELDS
+        )),
+    )
+
+
+def _rank_stream(pe_id: int, cube_id: int) -> tuple:
+    tl = TLContext(
+        pe_id=pe_id, num_programs=_P, cube_id=cube_id, num_cubes=_C,
+        scratch_base=0x200000, scratch_size=1 << 20,
+    )
+    run_kernel(
+        _baseline, tl,
+        0x1000, 0x2000, 0x3000, 0x4000,
+        _T_Q, _S_KV, _H_Q, _H_KV, _D_HEAD, _C, _P,
+    )
+    return tuple(_cmd_signature(c) for c in tl.commands)
+
+
+def _all_ranks_digest() -> str:
+    h = hashlib.sha256()
+    for cube_id in range(_C):
+        for pe_id in range(_P):
+            h.update(repr(_rank_stream(pe_id, cube_id)).encode())
+    return h.hexdigest()
+
+
+# Golden digest captured from the current (pre-refactor) baseline kernel.
+# Phase 2 extracts the intra-/inter-CUBE (m,ℓ,O) reduce into a shared
+# helper; the refactored kernel MUST reproduce this exact stream.
+_GOLDEN_DIGEST = "28069b4b1b19f427b33ee594fb71dc3987f3c92bf38b855058a2261e643047a9"
+
+
+def test_case6_reduce_refactor_byte_equal():
+    """Frozen command stream of the Case-6 baseline across all 64 ranks.
+
+    Guards the Phase-2 extraction of the shared reduce helper: the
+    refactored kernel must emit a byte-identical command stream."""
+    assert _all_ranks_digest() == _GOLDEN_DIGEST
+
+
+# ── Variant command-form behaviour (ADR-0064 Rev2 / ADR-0065) ────────
+
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_sp_pe_sp_composite import (  # noqa: E402,E501
+    gqa_attention_decode_long_ctx_cube_sp_pe_sp_composite_kernel as _composite,
+)
+from kernbench.benches.gqa_helpers.long_ctx._gqa_attention_decode_long_ctx_cube_sp_pe_sp_composite_ext import (  # noqa: E402,E501
+    gqa_attention_decode_long_ctx_cube_sp_pe_sp_composite_ext_kernel as _composite_ext,  # noqa: E501
+)
+from kernbench.common.pe_commands import CompositeCmd, PeCpuOverheadCmd  # noqa: E402,E501
+from kernbench.common.pe_cost_model import DEFAULT_PE_COST_MODEL  # noqa: E402
+
+
+def _dispatch(kernel, S_kv: int) -> tuple[int, float, int]:
+    """(# PE_CPU dispatch commands, summed dispatch cycles, # composites)
+    emitted by ``kernel`` at the lrab center rank (cube 6, pe 0)."""
+    tl = TLContext(
+        pe_id=0, num_programs=_P, cost_model=DEFAULT_PE_COST_MODEL,
+        cube_id=6, num_cubes=_C, scratch_base=0x200000, scratch_size=1 << 20,
+    )
+    run_kernel(
+        kernel, tl, 0x1000, 0x2000, 0x3000, 0x4000,
+        _T_Q, S_kv, _H_Q, _H_KV, _D_HEAD, _C, _P,
+    )
+    n_disp = sum(1 for c in tl.commands if isinstance(c, PeCpuOverheadCmd))
+    cycles = sum(c.cycles for c in tl.commands if isinstance(c, PeCpuOverheadCmd))
+    n_comp = sum(1 for c in tl.commands if isinstance(c, CompositeCmd))
+    return n_disp, cycles, n_comp
+
+
+# Two multi-tile S_kv points (S_local = 4096 / 8192 → 4 / 8 primitive tiles).
+_S_KV_A, _S_KV_B = 262_144, 524_288
+
+
+def test_variant_composite_counts():
+    """The primitive kernel issues no composites; each composite variant
+    issues exactly two (Q·Kᵀ and P·V / the recipe head GEMM)."""
+    assert _dispatch(_baseline, _S_KV_A)[2] == 0
+    assert _dispatch(_composite, _S_KV_A)[2] == 2
+    assert _dispatch(_composite_ext, _S_KV_A)[2] == 2
+
+
+def test_composite_dispatch_flat_in_skv():
+    """Coarse composite issue is S_kv-independent: the kernel emits O(1)
+    commands and PE_SCHEDULER absorbs the per-tile fan-out, so dispatch
+    cost is identical at 4-tile and 8-tile S_kv (it *saturates*)."""
+    for kernel in (_composite, _composite_ext):
+        n_a, cyc_a, _ = _dispatch(kernel, _S_KV_A)
+        n_b, cyc_b, _ = _dispatch(kernel, _S_KV_B)
+        assert n_a == n_b, f"{kernel.__name__}: {n_a} != {n_b}"
+        assert cyc_a == cyc_b
+
+
+def test_primitive_dispatch_grows_in_skv():
+    """The hand-tiled primitive kernel issues O(n_tiles) commands, so its
+    dispatch cost grows with S_kv — the cost the composite forms remove."""
+    n_a, cyc_a, _ = _dispatch(_baseline, _S_KV_A)
+    n_b, cyc_b, _ = _dispatch(_baseline, _S_KV_B)
+    assert n_b > n_a
+    assert cyc_b > cyc_a
+
+
+def test_composite_cheaper_than_primitive_at_long_ctx():
+    """At a long-context (8-tile) S_kv, both composite forms issue far
+    fewer / cheaper PE_CPU commands than the primitive kernel — the
+    CPU-offload win (ADR-0064 Rev2). The shared (m,ℓ,O) reduce is a fixed
+    common term, so the margin is the local-attention command form alone."""
+    prim_cyc = _dispatch(_baseline, _S_KV_B)[1]
+    comp_cyc = _dispatch(_composite, _S_KV_B)[1]
+    ext_cyc = _dispatch(_composite_ext, _S_KV_B)[1]
+    assert prim_cyc > 1.5 * comp_cyc, (prim_cyc, comp_cyc)
+    assert prim_cyc > 1.5 * ext_cyc, (prim_cyc, ext_cyc)
+
+
+def test_three_variants_complete_in_data_mode():
+    """All three command forms run end-to-end through the 64-PE engine in
+    data mode and report a positive decode latency. Guards the composite
+    P·V path + the on-chip-operand engine fix (TCM operand not DMA-streamed)
+    at a multi-rank S_kv where S_local (=128) exceeds the recipe seed."""
+    from kernbench.benches.gqa_helpers.long_ctx.gqa_decode_long_ctx_composite import (  # noqa: E501
+        _engine_latency_ns,
+    )
+
+    for variant in ("primitive", "composite", "composite_extended"):
+        lat = _engine_latency_ns(variant, 8192, str(_TOPOLOGY))
+        assert lat > 0, f"{variant}: latency={lat}"

tests/attention/test_gqa_prefill_compute_bound.py

@@ -0,0 +1,61 @@
+"""Tests for the compute-bound prefill composite-command study.
+
+Unlike memory-bound decode (where command form is latency-neutral), in
+compute-bound prefill the composite command keeps the MAC array fed by
+pipelining DMA↔compute per HW tile, so it wins on wall-clock latency. The
+rigorous claim here is therefore the *opposite* of the decode study:
+``composite_latency < primitive_latency`` at a compute-bound context.
+"""
+from __future__ import annotations
+
+from pathlib import Path
+
+from kernbench.common.pe_commands import CompositeCmd
+from kernbench.triton_emu.tl_context import TLContext, run_kernel
+
+from kernbench.benches.gqa_helpers.long_ctx._gqa_prefill_compute_bound import (
+    gqa_prefill_composite_ext_kernel,
+    gqa_prefill_composite_kernel,
+    gqa_prefill_primitive_kernel,
+)
+
+_TOPOLOGY = Path(__file__).resolve().parents[2] / "topology.yaml"
+_H_Q, _H_KV, _D_HEAD = 8, 1, 128
+
+
+def _n_composites(kernel, T_q: int, S_kv: int) -> int:
+    tl = TLContext(pe_id=0, num_programs=1, scratch_base=1 << 61,
+                   scratch_size=1 << 20)
+    run_kernel(kernel, tl, 0x1000, 0x2000, 0x3000, 0x4000,
+               T_q, S_kv, _H_Q, _H_KV, _D_HEAD, 1, 1)
+    return sum(1 for c in tl.commands if isinstance(c, CompositeCmd))
+
+
+def test_command_form_composite_counts():
+    """One Q-block (T_q=16 → M=128): the primitive issues no composites;
+    each composite form issues exactly two (Q·Kᵀ and P·V / recipe)."""
+    assert _n_composites(gqa_prefill_primitive_kernel, 16, 256) == 0
+    assert _n_composites(gqa_prefill_composite_kernel, 16, 256) == 2
+    assert _n_composites(gqa_prefill_composite_ext_kernel, 16, 256) == 2
+
+
+def _latency_ns(variant: str, ctx_len: int) -> float:
+    from kernbench.benches.gqa_helpers.long_ctx.gqa_prefill_compute_bound import (
+        _run_panel,
+    )
+    return _run_panel(variant, ctx_len, str(_TOPOLOGY))["latency_ns"]
+
+
+def test_three_variants_complete_in_data_mode():
+    """All three prefill forms run end-to-end through the engine."""
+    for variant in ("primitive", "composite", "composite_extended"):
+        assert _latency_ns(variant, 256) > 0
+
+
+def test_composite_faster_in_compute_bound_prefill():
+    """At a compute-bound context (512), the composite command's DMA↔compute
+    pipelining beats the primitive's serial load→dot — the opposite of the
+    memory-bound decode study, where the two tie."""
+    prim = _latency_ns("primitive", 512)
+    comp = _latency_ns("composite", 512)
+    assert comp < prim, f"composite {comp} not < primitive {prim}"

tests/test_composite_tcm_operand_dma.py

@@ -0,0 +1,59 @@
+"""Engine-fix spec test — composite GEMM with an on-chip (TCM) operand.
+
+Root cause: ``_make_compute_out`` (tl_context) builds compute-result
+handles with ``space="tcm"`` but omits ``pinned=True``. The PE_SCHEDULER
+tiling gate (tiling.py:306) skips the operand DMA_READ only when
+``pinned`` is set, so a compute result fed as a composite head-GEMM
+operand ``a`` (e.g. the attention probabilities into P·V) is wrongly
+scheduled as an HBM ``DMA_READ`` of its bit-61 scratch address →
+``PhysAddrError`` in data mode.
+
+Every other TCM-resident handle in tl_context is already marked
+``pinned=True`` for exactly this reason (composite auto-output, recipe
+scratch / primary-out). ``_make_compute_out`` is the one site that
+forgot. ``.pinned`` is read ONLY by the tiling DMA gate, so setting it on
+compute outputs is side-effect-free and byte-equal for existing benches
+(which never feed an unpinned compute output into a composite operand).
+
+Phase 1: this test FAILS until the fix (a DMA_READ for operand A is
+emitted for the on-chip ``a``).
+"""
+from __future__ import annotations
+
+from kernbench.common.pe_commands import CompositeCmd
+from kernbench.components.builtin.pe_types import StageType
+from kernbench.components.builtin.tiling import generate_plan_from_ops
+from kernbench.triton_emu.tl_context import TLContext, run_kernel
+
+
+def _pv_composite_kernel(*, tl) -> None:
+    """P·V composite whose ``a`` is a real on-chip compute result."""
+    x = tl.load(0x1000, shape=(8, 1024), dtype="f16")
+    a = tl.exp(x)                                   # compute output → TCM
+    V = tl.ref(0x3000, shape=(1024, 128), dtype="f16")   # HBM, streamed
+    O = tl.zeros((8, 128), dtype="f16")
+    tl.composite(op="gemm", a=a, b=V, out=O)
+
+
+def _pv_plan():
+    tl = TLContext(pe_id=0, num_programs=1,
+                   scratch_base=0x200000, scratch_size=1 << 20)
+    run_kernel(_pv_composite_kernel, tl)
+    comp = next(c for c in tl.commands if isinstance(c, CompositeCmd))
+    return generate_plan_from_ops(
+        ops=comp.ops, tile_m=32, tile_k=64, tile_n=32,
+        bytes_per_element=2, pe_prefix="sip0.cube0.pe0",
+    )
+
+
+def test_onchip_operand_a_not_dma_read():
+    """The on-chip ``a`` (a compute result) must NOT be streamed via a
+    DMA_READ; only the HBM ref ``b`` (V) is."""
+    plan = _pv_plan()
+    read_operands = [
+        s.params.get("operand")
+        for t in plan.tiles for s in t.stages
+        if s.stage_type == StageType.DMA_READ
+    ]
+    assert "A" not in read_operands, read_operands
+    assert "B" in read_operands, read_operands