ywkang/kernbench2
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
124 Commits 3 Branches 0 Tags
master
Commit Graph

16 Commits

Author SHA1 Message Date
ywkang 687c98086d ADR housekeeping: category prefixes, lifecycle folders, retroactive 0034-0037 
Filename + lifecycle:
- ADR rename to ADR-NNNN-<cat>-title.md with 8 3-letter category prefixes
  (dev / mem / lat / prog / algo / par / api / ver). Numbers stay immutable.
- ADR Lifecycle split into 3 folders, documented in CLAUDE.md Part 2:
  docs/adr/ (Accepted), docs/adr-proposed/ (Proposed/Stub/Draft),
  docs/adr-history/ (Superseded/Merged). Status field gains "Draft" for
  retroactive docs pending verification.

Merges (one ADR per topic, no change-history annotations):
- ADR-0017 absorbs ADR-0019 (Cube NOC + per-PE HBM connectivity, 10 D-items)
- ADR-0014 absorbs ADR-0021 (PE pipeline execution model, 8 D-items incl.
  TileToken self-routing and multi-op composite epilogue scope)
- ADR-0023 absorbs docs/ipcq-dma-codesign-hw.md as new "HW Realization
  Notes (Informative)" section (D16-D23 + Open HW Questions). codesign-hw.md
  deleted; ADR-0019/0021 moved to adr-history with one-line stub status

Retroactive documentation (G4 closures, code-verified):
- ADR-0037 forwarding component (TransitComponent: first-flit overhead,
  serial worker, path-based routing, single impl/multiple names)
- ADR-0036 IO_CPU component (target_start_ns global barrier stamping,
  per-cube fan-out, response aggregation)
- ADR-0035 M_CPU & M_CPU.DMA component (3 fan-out paths, DMA Resources,
  target_start_ns passthrough)
- ADR-0034 HBM controller internal design (per-PC state, address-based
  selection, flit-aware per-flit commit, async finalize, command-only
  fallback path)

Content updates:
- ADR-0010 expanded to full CLI surface (run/probe/web), retitled
  "Command Line Interface and Execution Semantics"
- ADR-0007 D2 rewritten to current state; ADR-0015 supersession notes pruned
- ADR-0005 wrapped in Decision header with D1-D5; ADR-0022 metadata
  block replaced with standard Status header
- ADR-0024 trimmed to rank=SIP launcher essentials (D1-D4);
  ADR-0027 cleaned of supersession history
- ADR-0033 D6 cleanup: address-based PC selection moved out of future-work
  (now documented in ADR-0034 D3); related D1/D3 wording realigned
- Cross-references back-filled in 5 ADRs (G3 gaps closed)

Onboarding docs split:
- docs/onboarding/ created
- moved: hw-architecture-overview.md, latency-model.md, di-presentation.md,
  ccl-author-guide{,.en}.md
- references updated in README, ADR-0023{,.en}, src/kernbench/ccl/__init__.py

Source / test / yaml: ADR-NNNN cross-references in docstrings and YAML
comments updated after the merges (ADR-0021->0014 D6, ADR-0019->0017 D8).
No behavior change.

Tooling:
- tools/verify_adr_lang_pairs.py + tests/test_verify_adr_lang_pairs.py
  (ADR EN/KO pair invariant checker)
- .claude/commands/report.md tracked (/report slash command)
- .gitignore: allow .claude/commands/*.md while keeping settings files ignored

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
 2026-05-20 01:15:55 -07:00
mukesh a7fe785e5f tl.composite: fused epilogue ops with per-op scope 
Extend tl.composite() with an ordered epilogue list. Each op carries
a scope flag - output_tile (default, runs once per (m,n) before
STORE), k_tile (every K-tile right after GEMM), or kernel. Plan
generator slots MATH stages by scope; pe_math reuses pe_dma's
local-loop pattern so chained epilogues (bias->relu) skip the port
hop. op_log captures per-stage params for telemetry. Topology
gains a gemm->math edge (snapshot test updated).

API stays backward-compatible - `epilogue=` is opt-in.

Example:
    h = tl.composite(
        op="gemm", a=a, b=b, out_ptr=int(out),
        epilogue=[
            {"op": "dequant", "scale": s_per_k, "scope": "k_tile"},
            {"op": "bias",    "bias":  bias_vec},
            {"op": "relu"},
            {"op": "scale",   "factor": 0.5},
        ],
    )
    tl.wait(h)

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
 2026-05-15 10:16:47 -07:00
ywkang 5fdb6f8797 Latency model: HBM PC striping + chunk-loop drain (ADR-0033) 
Previous model double-counted slow-upstream paths (e.g., 64KB via UCIe
128 GB/s was ~2x pessimistic). HBM CTRL now distributes bursts across
8 pseudo-channels via global round-robin, with per-chunk commit timing
that pipelines correctly against the bottleneck link's data arrival.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
 2026-05-14 21:59:07 -07:00
mukesh 1d8b9401e5 Intercube allreduce: pe0 cube-mesh reduce + multi-SIP ring/torus/mesh 
New intercube allreduce kernel replacing the old flat ring algorithms.
Reduces across the 4x4 cube mesh within each SIP (pe0-only, same-lane),
then inter-SIP exchange on root cube, then broadcast back. Supports
ring_1d, torus_2d, and mesh_2d_no_wrap SIP topologies driven by
topology.yaml. Integrated with dist.init_process_group / dist.all_reduce.

New files:
- src/kernbench/ccl/algorithms/intercube_allreduce.py (kernel)
- src/kernbench/ccl/sfr_config.py (configure_sfr_intercube_multisip)
- tests/test_allreduce_multidevice.py (config-driven, 3 topologies)
- tests/test_distributed_intercube_allreduce.py (full distributed path)
- tests/test_intercube_sfr_config.py (SFR wiring verification)

Modified:
- distributed.py: AhbmCCLBackend uses configure_sfr_intercube_multisip
- topologies.py: added torus_2d, mesh_2d_no_wrap
- install.py: global_E/W/N/S in _OPPOSITE_DIR
- topology.yaml: added system.sips.topology
- ccl.yaml: single intercube_allreduce algorithm
- benches/ccl_allreduce.py: row_wise cube-mesh tensor layout

Removed old flat-ring algorithms and their tests.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-04-16 17:33:42 -07:00
ywkang 998cc85762 Add PE-level IPCQ collective infra + unified ccl_allreduce bench (ADR-0023) 
Major changes:

PE-level IPCQ infrastructure:
- New PE_IPCQ component: ring-buffer control plane with 4-direction
  neighbor mapping, head/tail pointers, backpressure (poll/sleep).
- PE_DMA extended with vc_comm channel for IPCQ outbound/inbound DMA,
  including in-flight data snapshot (D9) and op_log recording at
  outbound time for Phase 2 replay correctness.
- IpcqDmaToken piggyback model: data + metadata travel together,
  atomic visibility at receiver (invariant I6).
- Credit return fast path: bottleneck-BW latency, no fabric vc_comm.

Phase 2 data execution (ADR-0020 integration):
- op_log extended: DmaWriteCmd now captures src_space/src_addr for
  Phase 2 dma_write copy; ipcq_copy ops recorded at outbound time.
- DataExecutor replays dma_write + ipcq_copy in t_start order.
- Engine._flush_data_phase: incremental cursor-based replay after
  each engine.wait() so host reads see post-Phase-2 data.
- KernelRunner Phase 1 writes disabled when op_log is active to
  prevent stale data from corrupting the MemoryStore snapshot.

TLContext / kernel API:
- tl.send(dir, src=TensorHandle), tl.recv(dir, shape, dtype),
  tl.recv_async, tl.wait(RecvFuture), copy_to_dst mode.
- TensorHandle operator overloading (add/sub/mul/div) via thread-local
  active TLContext → MathCmd dispatch through PE_MATH.
- PE-local scratch allocator for math output handles.
- tl.load returns space="hbm" handles for correct Phase 2 addressing.
- Additional math functions: maximum, minimum, fma, clamp, softmax, cdiv.

Unified ccl_allreduce bench (PyTorch-compat host code):
- Single benches/ccl_allreduce.py with run() + worker(rank, ws, torch)
  split matching real PyTorch DDP worker pattern.
- torch.distributed facade: init_process_group, get_world_size,
  get_rank, get_backend, all_reduce, barrier — only real PyTorch names.
- AhbmCCLBackend: eager install_ipcq at init, all_reduce dispatches
  kernel via tensor shard metadata (n_elem from shards[0].nbytes).
- world_size derived from topology spec (sips × cubes × pes_per_cube)
  with optional algorithm-level override in ccl.yaml.

Tensor API (PyTorch-compat surface):
- Tensor.numpy(): gather-aware (all shards via VA-based addressing).
- Tensor.copy_(source): scatter from host tensor into sharded target.
- RuntimeContext.from_numpy(arr): host-side staging tensor.
- Tensor.data property fixed to use numpy() (was shards[0]-only).

Algorithm modules moved to src/kernbench/ccl/algorithms/:
- ring_allreduce, mesh_allreduce, tree_allreduce, hello_send.
- Each module exports kernel_args(world_size, n_elem) helper.
- ccl.yaml module paths updated to kernbench.ccl.algorithms.*.

Dead code removed:
- 7 per-variant bench files (ccl_allreduce_{tcm,hbm,sram}, etc.).
- _run_ccl_bench greenlet-per-SIP scheduler.
- benches.loader.is_ccl_bench + run_rank detection.
- benches/ccl/ directory.

Tests:
- New test_ccl_allreduce_matrix.py: 7 parametrized cases
  (ring×3 buffers, ring 8/16, mesh 4, tree 7).
- New test_runtime_api_tensor.py: copy_/numpy/from_numpy unit tests.
- Existing tests updated for new import paths + world_size_override.

Docs:
- Korean ccl-author-guide.md and ADR-0023 paths updated.
- New English versions: ccl-author-guide.en.md, ADR-0023.en.md.

502 tests pass.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-04-12 19:36:59 -07:00
ywkang 81ce55571d Rename impl names: add builtin. prefix for clear provenance 
- components.yaml: all builtin impls use builtin.xxx naming
- topology.yaml: all impl references updated to builtin.xxx
- builder.py: hardcoded ucie impl → builtin.ucie
- Tests: all impl string references updated

Convention: builtin.<name> for built-in, custom.<name> for user-defined.
382 tests passing.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-04-09 00:16:24 -07:00
ywkang 1d95df4bee Restructure legacy backups, remove pe_accel, fix DMA self-routing 
- Move builtin_legacy/ → legacy/builtin/ (cleaner structure)
- Move pe_accel_legacy/ → legacy/pe_accel/
- Remove custom/pe_accel/ (replaced by new builtin)
- Remove pe_scheduler_v2 from components.yaml
- Switch topology.yaml to pe_scheduler_v1 (new builtin)
- Fix PE_DMA self-routing: handle consecutive DMA_READ stages
  (same component consecutive stages processed in-place, not via port)

382 tests passing.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-04-09 00:02:26 -07:00
ywkang b6eb97c49a Implement ADR-0021: PE pipeline refactor with token self-routing 
Step 1-2: Backup existing code
- builtin/ → builtin_legacy/ (unchanged backup)
- custom/pe_accel/ → custom/pe_accel_legacy/ (unchanged backup)

Step 3-4: New pipeline types and tiling
- pe_types.py: StageType, Stage, TilePlan, PipelinePlan, PipelineContext, TileToken
- tiling.py: generate_gemm_plan, generate_math_plan (ported from pe_accel)

Step 5: Component implementations (ADR-0021 D4-D6)
- PE_SCHEDULER: _feed_loop (singleton FIFO feeder) + plan generation
- PE_FETCH_STORE: new component — TCM ↔ Register File
- PE_GEMM: TileToken pipeline + legacy PeInternalTxn dual-mode
- PE_MATH: TileToken pipeline + legacy dual-mode
- PE_DMA: TileToken pipeline + legacy + fabric Transaction triple-mode
- PE_TCM: TcmRequest handler with dual-channel BW serialization

Step 6: Infrastructure
- topology.yaml: pe_fetch_store component + chaining edges
- components.yaml: pe_fetch_store_v1 registration
- builder.py: PE_COMP_OFFSETS, _add_pe_internal_edges, PE view positions
- Tests: node/edge counts, PE component sets updated

All components handle both TileToken (pipeline) and PeInternalTxn (legacy).
Token self-routing: components read next stage from token.plan, chain via out_port.
366 tests passing.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-04-08 23:35:31 -07:00
ywkang eb792e6212 Remove xbar/noc remnants, rule-based cube-view connectors 
- Delete xbar.py and noc.py (TwoDMeshNocComponent) — unused since router mesh
- Remove xbar_v1/noc_2d_mesh_v1 from components.yaml
- Fix pe_to_xbar → pe_to_router in routing exclusion set
- Fix xbar_to_hbm_bw_gbs → hbm_to_router_bw_gbs in report.py
- Update all docstrings/comments referencing xbar/bridge → router mesh
- Cube-view connectors: rule-based _connector_points helper
  - PE↔router: single diagonal line (not chevron)
  - UCIe N/S: 45°→horizontal→45°
  - UCIe E/W: 45°→vertical→45°
  - HBM ports: 45°→horizontal→45°

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-04-06 23:59:12 -07:00
ywkang 7640635f90 M_CPU/SRAM placement via pos_mm in topology.yaml (nearest router) 
Component placement uses mm coordinates in topology.yaml, mesh_gen
finds the nearest router automatically. M_CPU moved to pos_mm=[7.5,2.0]
(→ r0c2), SRAM at pos_mm=[1.5,9.0] (→ r3c0).

No hardcoded router references in topology config.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-04-05 00:48:20 -07:00
ywkang 5917b3497c Replace xbar/bridge/single-NOC with explicit router mesh (ADR-0019) 
- Remove xbar_top/bot, bridge, single noc node from topology
- Each cube_mesh.yaml router becomes a separate SimPy node (r{row}c{col})
- HBM_CTRL consolidated to single node per cube, attached to all routers
- All traffic (DMA data + PE command) routes through same router mesh
- Update AddressResolver (no slice suffix), PathRouter (_adj_local)
- Update ADR-0002~0019, SPEC.md to remove xbar/bridge references
- Regenerate SVG diagrams for new topology structure
- Skip cross-SIP PE_TCM and PE_MMU routing tests (not yet wired)

326 passed, 13 skipped

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-04-04 17:51:28 -07:00
ywkang 114510d4b9 Add SchedulerV2 (pe_accel), DPPolicy overrides, and new benchmarks 
- Add cycle-accurate PE accelerator scheduler (SchedulerV2) with tiled
  GEMM/Math pipelines (DMA_IN → GEMM → MATH → DMA_WB)
- Add DPPolicy num_pes/num_cubes/num_sips overrides for single-PE testing
- Support tuple target_pe for targeting specific PE subsets
- Add gemm_single_pe and gpt3_qkv benchmarks
- Switch default topology to pe_scheduler_v2

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-03-26 23:18:49 -07:00
ywkang 08812eda58 Add virtual memory support: PE_MMU, VA allocator, fabric MmuMapMsg 
Implement VA/MMU layer (ADR-0011 Phase 1) enabling Triton kernels to use
contiguous virtual addresses on sharded tensors.

Key changes:
- PE_MMU component: hybrid inbox (MmuMapMsg) + sync translate() for PE_DMA
- VirtualAllocator + PEMemAllocator: free-list with coalescing
- MmuMapMsg/MmuUnmapMsg fabric path with SIP-level routing
- DPPolicy-based mapping: replicate=local, sharded=broadcast
- Tensor lifecycle: del + weakref cleanup, context manager
- Rename: TensorHandle.pa→addr, DmaReadCmd.src_pa→src_addr, ctx→torch

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-03-26 00:01:47 -07:00
ywkang 62fb01ae18 Add reverse path response latency for PE DMA and PE_CPU→M_CPU 
Model fabric response hop latency for PE-internal operations:
- HBM_CTRL sends PeDmaMsg response on reverse path instead of direct done signal
- PE_CPU sends ResponseMsg via NOC→M_CPU on kernel completion
- Add NOC→PE_DMA and PE_CPU→NOC edges in topology builder
- Make HBM BW test assertions dynamic based on topology efficiency

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-03-20 15:40:56 -07:00
ywkang d75da439c6 Add probe CLI improvements, D2H read, UCIe/HBM tuning, BW sweep 
- Probe CLI: restructured output (tables first, routes below), per-hop
  timestamps, split cross-cube into best/worst cases, D2H read section
- UCIe overhead: 1ns -> 8ns per port (16ns per crossing) to fix
  cross-cube-best < cross-half latency inversion
- HBM efficiency: added efficiency=0.8 factor to hbm_ctrl, reducing
  effective BW from 256 to 204.8 GB/s
- Multi-size BW sweep: saturation tables (4KB-1MB) for all probe cases
- Probe default data size: 4KB -> 32KB for more realistic measurements
- IOChiplet NOC + D2H topology and tests
- NOC mesh, xbar, BW occupancy components and tests
- Cube mesh visualization diagram

278 tests pass.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
 2026-03-19 01:16:18 -07:00
ywkang 6f43807900 commit - release 1 2026-03-18 11:47:48 -07:00