project_6 db0df78580 [gen_patch] fix 3 critical bugs: reduce struct selection, topk bits_per_pass injection, transform/batch_memcpy extraction ...

Bug 1: reduce struct selection — gen_patch selected bi100_plus_accum1_o4 (int8
path, items=32) instead of bi100_plus_float32_o4 (fp32 score accumulator,
items=24). vllm paged_attention always uses fp32 for score accumulation, so
the wrong struct was injecting items=32 into the 83%-weight hot path.

Fix: preference-ordered struct selection — float32 > accum2 > first non-default.
Now correctly selects bi100_plus_float32_o4 → NUM_ITEMS_PER_THREAD=24.

Bug 2: topk bits_per_pass not injected — gen_patch only extracted threads=512
from topk inline policy_selector, missing calc_bits_per_pass(key_size).
For Qwen3.6 float32 logits (key_size=4), bits_per_pass=11 (not 8).

Fix: topk-specific extraction that parses calc_bits_per_pass and returns
bits_per_pass=11. Now generates RADIX_BITS=11 patch for sampling_kernels.cu.

Bug 3: transform/batch_memcpy extraction failed — these headers use
policy struct naming, not bi100_* naming, so extract_bi100_structs was empty.

Fix: algorithm-specific fallback extraction for transform (reads
bi100_bytes_in_flight constexpr) and batch_memcpy (reads threads from
policy_selector return).

Validation: gen_patch 7 patches (was 6), test_smem_safety 191/191 safe

2026-08-05 03:15:12 +00:00

..

include

[muh] delay v2 完成: 全部 8 个 lookback 算法改为 no_delay

2026-08-04 07:18:48 +00:00

schema

[MUH] Bootstrap muh toolchain — extract/parse/gen_yaml/gen_patch + baseline.muh

2026-07-30 10:39:06 +00:00

test

[MUH] Derive BI-V100 tuning values from hardware specs — fix 5 SMEM overflow bugs

2026-07-30 15:08:30 +00:00

tests

test: add SMEM safety validator for all 26 tuning algorithms

2026-08-01 12:37:14 +08:00

bench_bi100.py

feat(muh): add bench_bi100.py — CCCL BruteForceSeeker for BI-V100

2026-08-03 10:24:37 +00:00

bench_triton_prefill.py

feat: asymmetric BLOCK_M/BLOCK_N search + re-add BI-V100 autotune configs

2026-08-03 11:18:18 +00:00

bench_triton_real.py

[muh] bench_triton_real + bi100_configs: REAL tunable surface benchmark

2026-08-04 06:19:07 +00:00

bi100_configs.json

[muh] bench_triton_real + bi100_configs: REAL tunable surface benchmark

2026-08-04 06:19:07 +00:00

bi100_triton_configs.py

[muh] bench_triton_real + bi100_configs: REAL tunable surface benchmark

2026-08-04 06:19:07 +00:00

extract.py

[MUH] Bootstrap muh toolchain — extract/parse/gen_yaml/gen_patch + baseline.muh

2026-07-30 10:39:06 +00:00

gen_patch.py

[gen_patch] fix 3 critical bugs: reduce struct selection, topk bits_per_pass injection, transform/batch_memcpy extraction

2026-08-05 03:15:12 +00:00

gen_yaml.py

[MUH] Bootstrap muh toolchain — extract/parse/gen_yaml/gen_patch + baseline.muh

2026-07-30 10:39:06 +00:00

parse.py

[MUH] Bootstrap muh toolchain — extract/parse/gen_yaml/gen_patch + baseline.muh

2026-07-30 10:39:06 +00:00

run_on_bi100.sh

feat: asymmetric BLOCK_M/BLOCK_N search + re-add BI-V100 autotune configs

2026-08-03 11:18:18 +00:00

vllm_bi100_patch.py

feat(muh): apply CCCL-derived BI-V100 tuning to 5 vllm Python files

2026-08-03 10:27:10 +00:00