ViperEkura/AstrAI
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

refactor: remove ineffective __launch_bounds__ from prefill kernel

Browse Source 
- Remove MIN_BLOCKS template param and __launch_bounds__ attribute
- Profiling shows smem (not registers) is the occupancy bottleneck for D>=64, making the hint a no-op
- D=64 sees 2-4% speedup, D=128 unchanged (smem-capped at 1 block/SM)
- Update comment blocks in kernel header and both dispatch sites
- Verified correctness via standalone CUDA test (max_err ~1e-4)
This commit is contained in:
ViperEkura
2026-07-12 15:00:53 +08:00
parent 69fecaf387
commit b5cdea98ad
3 changed files with 5 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files
csrc/tests/attn_prefill_test.cu
+1 -3
View File
@@ -18,11 +18,9 @@ static void launch_prefill(AttentionParams<bf16>& p) {
#ifndef ASTRAI_NO_MMA
constexpr int WARPS = 4, BR = 16;
constexpr int BC = (HEAD_DIM <= 128) ? 32 : 16;
constexpr int MIN_BLOCKS = (HEAD_DIM <= 32) ? 6 : (HEAD_DIM <= 64) ? 4
: (HEAD_DIM <= 128) ? 3 : 2;
dim3 grid((p.q_len + BR * WARPS - 1) / (BR * WARPS), p.q_head, p.batch);
dim3 block(WARPS * 32, 1, 1);
attn_prefill_split_q_mma_kernel<HEAD_DIM, WARPS, BC, MIN_BLOCKS><<<grid, block>>>(p);
attn_prefill_split_q_mma_kernel<HEAD_DIM, WARPS, BC><<<grid, block>>>(p);
#else
constexpr int G = 8, ROWS = 32, P_BC = 32;
dim3 grid((p.q_len + ROWS - 1) / ROWS, p.q_head, p.batch);