ViperEkura 9ebaea840f perf: cp.async K/V loads, shared sQ staging, causal skip, XOR swizzle ... - cp.async global→shared for K/V full-tile loads, eliminates 99.6% of shared-store bank conflicts (612K→2.7K per ncu) - add cp_async_16/commit/wait_all/wait_group<N> helpers in mma utils - shared sQ staging (single area, serialized per-warp load), cuts smem from (2*BC + WARPS*BR)*LD to (2*BC + BR)*LD bf16 - pre-scale Q by attention scale during Q load, removes per-tile scale multiply in softmax loop - causal tile skipping: block-level early break + warp-level skip - scalar fallback only for last partial tile - XOR swizzle (swiz_col) at 8-bf16 chunk granularity, eliminates ldmatrix bank conflicts without LD padding, LD=HEAD_DIM (zero smem waste), saves 1280 bytes/block vs HEAD_DIM+8 padding		2026-07-08 12:30:32 +08:00
..
gqa_common.cuh	fix: correct prefill mask index, unify GQA kernel interface	2026-07-06 17:21:23 +08:00
gqa_decode_attn.cu	feat: MMA head-packing decode kernel with scalar fallback dispatch	2026-07-07 22:56:02 +08:00
gqa_decode_attn.cuh	fix: correct prefill mask index, unify GQA kernel interface	2026-07-06 17:21:23 +08:00
gqa_decode_attn_mma.cuh	refactor: extract shared MMA utils into gqa_mma_utils.cuh	2026-07-07 23:01:15 +08:00
gqa_mma_utils.cuh	perf: cp.async K/V loads, shared sQ staging, causal skip, XOR swizzle	2026-07-08 12:30:32 +08:00
gqa_prefill_attn.cu	perf: cp.async K/V loads, shared sQ staging, causal skip, XOR swizzle	2026-07-08 12:30:32 +08:00
gqa_prefill_attn.cuh	feat: tensor-core mma prefill with build-time dispatch	2026-07-06 20:33:24 +08:00
gqa_prefill_attn_mma.cuh	perf: cp.async K/V loads, shared sQ staging, causal skip, XOR swizzle	2026-07-08 12:30:32 +08:00