Documents four cross-cutting surfaces that previously had no ADR backing, each surfaced as a G4 candidate by /report: - 0046 prog-tl-context-contract: the kernel-side tl.* API. Enumerates all primitives (ref/load/store/dot/composite/math/reduction/IPCQ/...), the two execution modes (command-list vs greenlet runner), scratch allocator semantics, dispatch-overhead model, and the kernel registry. - 0047 par-ahbm-ccl-backend: torch.distributed.init_process_group (backend="ahbm") install path. world_size priority (algorithm > defaults > topology), the 4-step init sequence (load ccl.yaml, import algorithm module, derive world_size, install SFR + IPCQ), greenlet- local rank registry, all_reduce dispatch via _defer_wait, barrier no-op rationale, and the explicit list of unsupported dist.* APIs. - 0048 mem-allocator-algorithms: VirtualAllocator + PEMemAllocator free-list semantics. Offset-keyed first-fit with coalescing, the no-validation trust model for free(), HBM/TCM channel separation, page-aligned VA allocation, the page_size dual-default (VirtualAllocator 2 MiB / _ensure_allocators 4 KiB fallback), and one-allocator-per-sub-unit rule. - 0049 ver-probe-subcommand: kernbench probe traffic-pattern catalog. H2D / D2H / PE DMA categories with their exact cube-index choices, the 32 KiB reference size, the 5-point utilization sweep, the formula vs actual column meanings, automatic invariant checks (monotonicity, D2H >= H2D, best < worst), per-case GraphEngine isolation, and the human-readable (not machine-parsable) output contract. Bilingual pair verifier passes for all four EN/KO pairs. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
11 KiB
ADR-0048: Memory Allocator Algorithms — VirtualAllocator + PEMemAllocator
Status
Accepted (2026-05-22).
Pins down the free-list algorithm, page alignment, and coalescing rules
used by policy/address/allocator.py's _FreeList / PEMemAllocator
and va_allocator.py's VirtualAllocator. ADR-0001 (PhysAddr layout)
and ADR-0011 (PA/VA/LA models) define the address schemes; the
allocation algorithms had no ADR-level coverage.
First action
_FreeList(capacity)
On construction: self._capacity = capacity, self._used = 0,
self._free = [(0, capacity)]. The first act is establishing the
entire region as one free block — the tuple (offset=0, size=capacity) is the sole entry in the free list.
PEMemAllocator(sip_id, die_id, pe_id, cfg)
On construction, builds two _FreeLists:
self._hbm = _FreeList(cfg.hbm_slice_bytes)— the size of this PE's HBM slice (hbm_bytes_per_cube // hbm_slices_per_cube).self._tcm = _FreeList(cfg.tcm_allocatable_bytes)— equalstcm_bytes_per_pe - tcm_scheduler_reserved_bytes(the scheduler reservation is pre-deducted).
So PEMemAllocator's first act is constructing single-free-block HBM-slice and TCM regions for this PE.
VirtualAllocator(va_base, va_size, page_size=2*1024*1024)
On construction: self._va_base = va_base, self._va_size = va_size,
self._page_size = page_size, self._used = 0, self._free = [(va_base, va_size)]. The first act is establishing one block from
va_base to va_size and stashing page_size.
Context
runtime_api/context.py::_ensure_allocators builds the allocator set
in these stages:
- Read
hbm_total_gb_per_cube,hbm_slices_per_cube,tcm_size_mb, per-target_device SIP range, etc. fromspec. - Pack everything into a frozen
AddressConfig. - For every combination in the target SIP range × cubes × PEs,
construct one
PEMemAllocator(sip, cube, pe, cfg)instance. - Construct one
VirtualAllocator(va_base=0x1_0000_0000, va_size=64 GiB, page_size=pe_mmu.page_size).
Allocator responsibilities:
- PEMemAllocator: PA-space allocation in the PE-local HBM slice / TCM (including PhysAddr encoding).
- VirtualAllocator: device-wide VA allocation, page-aligned.
RuntimeContext._create_tensorthen pushes VA → PA mappings to components viaMmuMapMsg.
These algorithms are:
- First-fit, kept simple.
- The free-block list is sorted by start offset.
- On
free(), adjacent blocks coalesce.
The rationale was not documented anywhere, so when someone asks "why not best-fit?", "why not a buddy allocator?", "why does partial-overlap free pass silently?", there was no anchor to answer from. This ADR provides it.
Decision
D1. _FreeList — offset-keyed first-fit + coalescing
policy/address/allocator.py::_FreeList:
- Internal representation:
list[tuple[int, int]] = [(start_offset, size), ...]— sorted by start offset. alloc(nbytes):- Iterate the free list from the front (first-fit).
- Take from the first block with
size >= nbytes. - Exact match → drop the block; otherwise shrink it to
(start + nbytes, size - nbytes). _used += nbytes; return the takenstart.- If no block fits,
AllocationError("overflow ... largest free block ...").
free(offset, nbytes):_used -= nbytes.bisect_left(self._free, (offset,))finds the insertion index.- If adjacent to the previous block (
prev_start + prev_size == offset), merge. - If adjacent to the next block (
offset + nbytes == next_start), merge. - Insert the coalesced range at the right sorted position.
This algorithm is weaker than best-fit / buddy on fragmentation, but the simulator's workload (mostly stack-like deploy/free) tolerates it. If the workload shape changes, D1 is a supersession candidate.
D2. Partial-overlap free is not validated
_FreeList.free(offset, nbytes) trusts the caller to pass the exact
(offset, nbytes). It does not verify:
- That the range was actually allocated.
- That the range does not overlap another allocated region.
Reason: in a simulator context, callers always store the return value
of alloc() and pass it back to free() — there is no external user
input. Adding a safety check would cost O(N) per free and impact
simulation wall-clock.
If this trust model breaks (e.g., a code path lets two tensors point at the same PA), this ADR must be revisited.
D3. PEMemAllocator — two channels for HBM/TCM
PEMemAllocator(sip_id, die_id, pe_id, cfg) holds two _FreeLists:
_hbm: sizecfg.hbm_slice_bytes._tcm: sizecfg.tcm_allocatable_bytes(=tcm_bytes_per_pe - tcm_scheduler_reserved_bytes).
alloc_hbm(nbytes) -> PhysAddr:
_hbm.alloc(nbytes)→ offset.PhysAddr.pe_hbm_addr(sip_id, die_id, pe_id, pe_local_hbm_offset=offset, slice_size_bytes=cfg.hbm_slice_bytes).- Failure raises
AllocationError("HBM overflow ...").
free_hbm(pa, nbytes):
- Recover PE-local offset via
pa.hbm_offset - pe_id * cfg.hbm_slice_bytes. _hbm.free(offset, nbytes).
alloc_tcm(nbytes) -> PhysAddr: similar; uses PhysAddr.pe_tcm_addr.
free_tcm(pa, nbytes): uses pa.sub_offset directly (TCM's PE-local
offset equals its sub_offset).
The allocator does not see the scheduler-reserved TCM region
(cfg.tcm_scheduler_reserved_bytes) — it is pre-subtracted from the
_tcm capacity. This is consistent with ADR-0014's PE_SCHEDULER
internal-buffer reservation.
D4. VirtualAllocator — page-aligned first-fit + coalescing
policy/address/va_allocator.py::VirtualAllocator:
- Internal representation: same sorted
list[tuple[int, int]]as_FreeList. Initially[(va_base, va_size)]. _align_up(nbytes) = ceil(nbytes / page_size) * page_size.alloc(nbytes) -> int:aligned = _align_up(nbytes).- First-fit a block with
size >= aligned. - Take
alignedfrom the block's front; remove if exact. _used += aligned. Return the block'sstart(which is page- aligned).- Failure →
VaAllocationError.
free(va, nbytes): free_align_up(nbytes)worth. Coalesces with the same algorithm as_FreeList.
page_size has different defaults in two places:
VirtualAllocator.__init__'s parameter default:2 MiB. Direct-call tests receive this.RuntimeContext._ensure_allocatorswhen constructing the instance:pe_mmu.attrs.get("page_size", 4096)— usestopology.yaml'spe_mmu.attrs.page_sizeif set, else falls back to4 KiB.
The two defaults differ on purpose: VirtualAllocator's standalone
default (2 MiB) aligns with ADR-0039's PE_MMU stopgap default for
direct-test ergonomics; the context fallback (4 KiB) is the safe
minimum when topology.yaml doesn't specify a page size. The
production path is always the latter (via _ensure_allocators), and
when topology.yaml sets page_size, that value flows consistently
into both the MMU and the VA allocator.
If consistency breaks (e.g., VirtualAllocator instantiated with a
page_size different from PE_MMU's), MMU map() falls into the
sub-page region mode (ADR-0039 D3).
VA range defaults: va_base = 0x1_0000_0000 (= 4 GiB), va_size = 64 GiB. These are hardcoded in _ensure_allocators and have no
semantic meaning in ADR-0011's VA model — they simply reserve enough
device-wide space without colliding with host code.
D5. Lifecycle of allocator instances
RuntimeContext._ensure_allocatorsis lazy — called on the first_create_tensor.- The allocator dict (
self._allocators) lives for the RuntimeContext's lifetime. A second deploy in the same process does not construct new objects. RuntimeContext.cleanup()walks living tensors and calls_free_tensor(), which issues MMU unmaps +va_allocator.free+pemem_allocator.free_hbm— restoring the free lists. A subsequent RuntimeContext starts fresh.
This per-RuntimeContext isolation guarantees deterministic deploy → cleanup → deploy sequences within a single process.
D6. Allocator failure raises (no silent OOM)
Both _FreeList.alloc and VirtualAllocator.alloc raise
AllocationError / VaAllocationError when no block fits. The message
includes "required size + largest available block" to distinguish
fragmentation from true OOM.
A silent fallback (e.g., allocating only as much as the largest free block) is strictly forbidden — a partially-allocated tensor reaching SimPy would cause routing / DMA to see incorrect PAs and silently corrupt simulation results.
D7. One allocator per address space
Physical address spaces are separated by PhysAddr sub-units (ADR-0001 D2.3); each sub-unit gets its own allocator instance:
- HBM slice →
PEMemAllocator._hbm. - PE TCM →
PEMemAllocator._tcm. - (Currently unused) M_CPU local memory, CUBE SRAM → would need their own allocators. Today these are handled as IPCQ-only slots (ADR-0023 D9.7) and do not share PA space, so no free-list exists for them.
When a cube-level SRAM allocator is needed,
_FreeList(cfg.sram_bytes_per_cube) is added per-cube
(cfg.sram_bytes_per_cube is already defined in AddressConfig —
the data model is ready).
Alternatives Considered
A1. Best-fit / buddy allocator
Rejected (currently). The workload's alloc/free pattern is stack-like (deploy order ≈ free order), so first-fit + coalescing controls fragmentation well enough. If long-running fragmentation appears in LLM kernel sweeps, a buddy-allocator ADR will replace D1.
A2. Add partial-overlap free validation
Rejected. D2's trust model plus the O(N) per-free cost makes this
unattractive. A debug mode (e.g., KERNBENCH_DEBUG env var) that
enables the check could be added later.
A3. A unified allocator for VA and PA
Rejected. VA space (64 GiB device-wide) and PA space (per-slice ~6 GiB) have different semantic dimensions — VA is the kernel's view, PA is the device sub-unit's view. ADR-0011's VA model (MMU maps between the two) calls for separated allocators.
A4. Multi-tier page sizes (large pages + small pages)
Rejected (currently). A single page size (2 MiB) matches LLM kernel tensor sizes (a few MiB to GiB); smaller mappings are absorbed by ADR-0039 D3's sub-page region mode. Multi-tier paging would require extending the MMU model itself — a separate ADR candidate.
Consequences
- The allocator algorithm is pinned at ADR level (D1, D3, D4), so any future simulation scenario hitting fragmentation has a clear "we're using first-fit + coalescing" anchor to inspect.
- D2's trust model is explicit, so any future code path that exposes alloc/free to direct user input will trigger this ADR's supersession early.
- D7's one-allocator-per-sub-unit mapping is on record, so when M_CPU or SRAM need their own free-list, the addition point is obvious.
- D4 captures the page_size dual-default and its production path
(
_ensure_allocatorsalways wins), letting futuretopology.yamlpage_sizechanges be assessed against ADR-0039's stopgap interaction quickly.