Files

T

mukesh cc1bbd0ab7 eval: fold GEMM/allreduce harnesses into self-contained milestone benches

Move the GEMM + allreduce sweep/render logic out of scripts/ and tests/
into two self-contained eval benches so a user can regenerate every
result + figure with one command:

  kernbench run --bench milestone-1h-gemm   (MILESTONE_FAST=1 reuses JSON)
  kernbench run --bench milestone-1h-ccl

- benches/milestone_1h_{gemm,ccl}.py: single home for each domain; the
  run(torch) entry drives the sweeps and writes figures into
  benches/1H_milestone_output/{gemm,ccl}/ (gitignored), then submits a
  sentinel tensor to satisfy the run_bench contract.
- tests/gemm + tests/sccl helpers and scripts/gemm_sweep.py become thin
  re-export/wrapper shims over the benches (single source preserved); the
  pytest-only param builders + _run_distributed wrapper stay in the shim.
- eval-bench pattern: a bench may drive many configs + build its own
  per-config engines (extends ADR-0045 D5; reverses ADR-0044 D1/D2).

ADR-0054 (EN+KO) records the design; ADR-0043/0044/0045 + CLAUDE.md CLI
Semantics amended; ADR INDEX regenerated. Verified: milestone benches run
clean (ok=True, all artifacts), full suite 67 passed, lang-pairs OK.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

2026-05-22 15:19:52 -07:00

7.0 KiB

Raw Blame History

ADR Index

Auto-generated by tools/generate_adr_index.py. Total ADRs: 47.

Classification mirrors the /report skill's section assignment. When adding a new ADR, also add an entry to the CLASSIFICATION table in tools/generate_adr_index.py.

Design Principles

ADR-0013 — Verification Strategy and Phase 1 Test Plan
ADR-0033 — Latency Model: Assumptions and Known Simplifications

High-level Architecture

ADR-0003 — Target System Hierarchy & Modeling Scope (System hierarchy (Tray / SIP / CUBE / PE))
ADR-0007 — Runtime API and Simulation Engine Boundaries (Runtime API ↔ sim_engine boundaries)
ADR-0016 — IOChiplet NOC and Memory Data Path (IOChiplet NOC and memory data path)
ADR-0017 — Cube NOC and HBM Connectivity (Cube NOC and HBM connectivity)

Detailed Architecture

One subsection per component file under src/kernbench/components/builtin/.

forwarding

ADR-0037 — Forwarding Component (forwarding_v1)

hbm_ctrl

ADR-0034 — HBM Controller Internal Design

io_cpu

ADR-0036 — IO_CPU Component Model

m_cpu

ADR-0035 — M_CPU and M_CPU.DMA Component Model

pcie_ep

ADR-0038 — PCIE_EP Component Model

pe_cpu

ADR-0014 — PE Pipeline Execution Model

pe_dma

ADR-0014 — PE Pipeline Execution Model
ADR-0023 — PE-level IPCQ — Inter-PE Collective Communication

pe_fetch_store

ADR-0014 — PE Pipeline Execution Model

pe_gemm

ADR-0014 — PE Pipeline Execution Model

pe_ipcq

ADR-0023 — PE-level IPCQ — Inter-PE Collective Communication

pe_math

ADR-0014 — PE Pipeline Execution Model

pe_mmu

ADR-0039 — PE_MMU Component Model — Component + Utility Dual Role

pe_scheduler

ADR-0014 — PE Pipeline Execution Model

pe_tcm

ADR-0040 — PE_TCM Component Model — Dual-Channel BW Serialization

sram

ADR-0041 — Cube SRAM Component Model — terminal scratchpad on cube NoC

tiling

ADR-0042 — Tile Plan Generators — GEMM/Math Pipeline Plan Builders

Implementation Decisions

Address Scheme

ADR-0001 — 51-bit Physical Address Layout & Decoding Contract
ADR-0011 — Memory Addressing — PA / VA / LA Address Models

Routing & Helper API

ADR-0002 — Routing Distance, Ordering & Bypass Rules
ADR-0051 — Routing Helper API — AddressResolver + PathRouter

Memory Semantics & Local-HBM Bandwidth

ADR-0004 — Memory Semantics & Local-HBM Bandwidth Guarantee

Topology Compilation, Diagrams & Builder Algorithms

ADR-0005 — Diagram Views & Distance-Aware Layout Rules
ADR-0006 — Topology Compilation, Distance Extraction, and Automatic Diagram Generation
ADR-0053 — Topology Builder + Visualizer Algorithms

Tensor Deployment and Allocation

ADR-0008 — Tensor Deployment and Allocation (Host Allocator, PA-first)

Kernel Execution and Host-Device Messaging

ADR-0009 — Kernel Execution Messaging and Completion Semantics
ADR-0012 — Host ↔ IO_CPU Message Schema (PA-first, PE-tagged)

CLI Surface and Semantics

ADR-0010 — Command Line Interface and Execution Semantics

Component Port/Wire Fabric Model

ADR-0015 — Component Port/Wire Model and Fabric Routing

Two-Pass Data Execution

ADR-0020 — 2-Pass Data Execution Model (Timing / Data Separation)

2D Grid Program Identity

ADR-0022 — 2D Grid program_id Semantics

Parallelism (Launcher, DP, TP, AHBM backend, CCL algorithm)

ADR-0024 — SIP-level Launcher — rank = SIP
ADR-0026 — DPPolicy = Intra-Device Only — remove sip/num_sips fields
ADR-0027 — Megatron-style Tensor Parallelism API
ADR-0047 — AHBM CCL Backend — torch.distributed-compat shim
ADR-0050 — CCL Algorithm Module Contract — ccl/algorithms/*.py

IPCQ Direction Addressing

ADR-0025 — IPCQ Direction Addressing — address-based matching

Intercube All-Reduce

ADR-0032 — Intercube All-Reduce — pe0 cube-mesh reduce + multi-SIP exchange

Evaluation Harnesses

ADR-0043 — Allreduce Evaluation Harness — tests/sccl/
ADR-0044 — GEMM Evaluation Harness — scripts/gemm_sweep.py + tests/gemm/
ADR-0054 — Milestone Eval Benches — self-contained sweep + figure benches

Bench Module Contract

ADR-0045 — Bench Module Contract — registration, dispatch, and authoring

Kernel-side tl.* API (TLContext)

ADR-0046 — TLContext — Kernel-side tl.* API Contract

Memory Allocator Algorithms

ADR-0048 — Memory Allocator Algorithms — VirtualAllocator + PEMemAllocator

Probe Subcommand

ADR-0049 — kernbench probe Subcommand — Traffic-Pattern Verification Harness

Sim-engine Op Log and Memory Store Schemas

ADR-0052 — OpLog + MemoryStore Schemas — sim_engine internals

7.0 KiB Raw Blame History