ywkang

ywkang pushed to master at ywkang/kernbench2

2026-04-14 16:00:30 +00:00

4ba0a83e71 Implement ADR-0024 Phase A: SIP-level TP launcher MVP

ywkang pushed to master at ywkang/kernbench2

2026-04-14 07:38:45 +00:00

32536daf2e Fix ADR-0025: IPCQ direction addressing via address-based matching

e1084800ab docs: add ADRs 0024–0031 for SIP-TP launcher stack

Compare 2 commits »

ywkang pushed to master at ywkang/kernbench2

2026-04-13 23:31:35 +00:00

b2c52f0e34 Add English translations for ADR-0018, 0019, 0020, 0021

ywkang pushed to master at ywkang/kernbench2

2026-04-13 06:52:05 +00:00

10b33b44ba Add Tensor indexing + hierarchical 3-level all-reduce kernel

ywkang pushed to master at ywkang/kernbench2

2026-04-13 06:02:21 +00:00

1c8ddc2d03 Fix Phase 1 slot-overwrite race + PE_MATH latency model (n_slots=4 safe)

ywkang pushed to master at ywkang/kernbench2

2026-04-13 04:13:27 +00:00

74f5f5cf08 Add session-scoped topology fixture in tests/conftest.py

ywkang pushed to master at ywkang/kernbench2

2026-04-13 04:06:42 +00:00

372c987995 Reduce test time to 12s: shrink GEMM dims + enable pytest-xdist

ywkang pushed to master at ywkang/kernbench2

2026-04-13 03:52:14 +00:00

bcf941dcee Speed up regression: 25min → 6min (test matrix + DataExecutor cleanup)

ywkang pushed to master at ywkang/kernbench2

2026-04-13 02:37:05 +00:00

998cc85762 Add PE-level IPCQ collective infra + unified ccl_allreduce bench (ADR-0023)

ywkang pushed to master at ywkang/kernbench2

2026-04-09 23:49:58 +00:00

ff2c677a9c Add 2D grid program_id semantics (ADR-0022)

ywkang pushed to master at ywkang/kernbench2

2026-04-09 16:34:07 +00:00

dc3fb02aed Add --verify-data CLI flag, Tensor.data property, parallel DataExecutor

ywkang pushed to master at ywkang/kernbench2

2026-04-09 07:28:05 +00:00

59e36f0c34 Add E2E pipeline tests: greenlet op_log, GEMM accuracy, latency regression

ywkang pushed to master at ywkang/kernbench2

2026-04-09 07:16:26 +00:00

81ce55571d Rename impl names: add builtin. prefix for clear provenance

ywkang pushed to master at ywkang/kernbench2

2026-04-09 07:02:28 +00:00

1d95df4bee Restructure legacy backups, remove pe_accel, fix DMA self-routing

ywkang pushed to master at ywkang/kernbench2

2026-04-09 06:49:29 +00:00

95d583ef9f Add Phase 1→Phase 2 e2e data tests + GraphEngine enable_data mode

ywkang pushed to master at ywkang/kernbench2

2026-04-09 06:40:22 +00:00

f5d1606f9d Add ADR-0021 pipeline tests: self-routing, tiling, overlap

ywkang pushed to master at ywkang/kernbench2

2026-04-09 06:35:40 +00:00

b6eb97c49a Implement ADR-0021: PE pipeline refactor with token self-routing

ywkang pushed to master at ywkang/kernbench2

2026-04-09 06:21:47 +00:00

161132cdcb ADR-0021: PE pipeline refactor — component separation + token self-routing

ywkang pushed to master at ywkang/kernbench2

2026-04-08 07:22:50 +00:00

51004c311c Implement ADR-0020: 2-pass data execution with greenlet kernel runner

ywkang pushed to master at ywkang/kernbench2

2026-04-08 06:53:58 +00:00

140b85436a ADR-0020: 2-Pass data execution model with greenlet kernel runner