ViperEkura fda82ee232 perf: drop redundant smem zero-init in paged decode kernel ...

- Removes per-step STAGES*BC*LD smem clear loop (2 buffers x 24 layers)
- cp.async predicated load + softmax mask already exclude padding slots,
  matching the paged prefill kernel which never zero-inits
- Standalone and extension tests pass; decode step time unchanged

2026-08-01 16:17:34 +08:00

..

attn_common.h

feat: SGLang-style paged attention kernels replace page-table path

2026-08-01 15:41:25 +08:00

attn_decode_split_kv_mma.cuh

perf: remove split partials memset and overlap decode tile loads

2026-07-31 22:37:44 +08:00

attn_decode_split_kv.cuh

style: unify decode expf to __expf

2026-07-31 00:19:18 +08:00

attn_decode.cu

refactor: replace magic layout ints with TensorLayout enum

2026-08-01 11:05:52 +08:00

attn_dispatchers.cuh

feat: SGLang-style paged attention kernels replace page-table path

2026-08-01 15:41:25 +08:00

attn_entry_utils.cuh

feat: SGLang-style paged attention kernels replace page-table path

2026-08-01 15:41:25 +08:00

attn_mma_utils.cuh

build: parametrize CUDA version for wheels and docker

2026-07-31 14:10:55 +08:00

attn_paged_decode_split_kv_mma.cuh

perf: drop redundant smem zero-init in paged decode kernel

2026-08-01 16:17:34 +08:00

attn_paged_decode_split_kv.cuh

feat: SGLang-style paged attention kernels replace page-table path

2026-08-01 15:41:25 +08:00

attn_paged_decode.cu

feat: SGLang-style paged attention kernels replace page-table path

2026-08-01 15:41:25 +08:00

attn_paged_prefill_split_q_mma.cuh

feat: SGLang-style paged attention kernels replace page-table path

2026-08-01 15:41:25 +08:00

attn_paged_prefill.cu

feat: SGLang-style paged attention kernels replace page-table path

2026-08-01 15:41:25 +08:00

attn_prefill_split_q_mma.cuh

feat: unify attention backend with multi-dim mask support

2026-07-30 20:38:34 +08:00

attn_prefill_split_q.cuh

feat: unify attention backend with multi-dim mask support

2026-07-30 20:38:34 +08:00

attn_prefill.cu

refactor: replace magic layout ints with TensorLayout enum

2026-08-01 11:05:52 +08:00

attn_warp_utils.cuh

perf: extend MMA decode to arbitrary GQA ratio, add launch bounds, vectorize combine

2026-07-27 00:35:34 +08:00

rotary_emb.cu

refactor: replace magic layout ints with TensorLayout enum

2026-08-01 11:05:52 +08:00