ViperEkura 21ddead238 fix: stabilize paged decode attention kernels ...

- zero-fill split partials so combine skips unwritten splits deterministically
- skip loading masked KV in paged decode kernels to avoid 0*NaN output poisoning
- zero-fill shared memory tile buffers to prevent stale NaN leaking into softmax

2026-07-31 21:01:12 +08:00

..

attn_common.h

feat: unify attention backend with multi-dim mask support

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

attn_decode_split_kv_mma.cuh

feat: unify attention backend with multi-dim mask support

2026-07-30 20:38:34 +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: extract shared dispatcher header, unify MMA/scalar dispatch format

2026-07-21 22:21:39 +08:00

attn_dispatchers.cuh

perf: enable paged MMA kernel for page_size=1

2026-07-30 22:06:41 +08:00

attn_entry_utils.cuh

fix: stabilize paged decode attention kernels

2026-07-31 21:01:12 +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

fix: stabilize paged decode attention kernels

2026-07-31 21:01:12 +08:00

attn_paged_decode_split_kv.cuh

fix: stabilize paged decode attention kernels

2026-07-31 21:01:12 +08:00

attn_paged_decode.cu

refactor: extract shared dispatcher header, unify MMA/scalar dispatch format

2026-07-21 22:21:39 +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: extract shared dispatcher header, unify MMA/scalar dispatch format

2026-07-21 22:21:39 +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: unify rotary embedding interface and update docs

2026-07-31 16:52:25 +08:00