ViperEkura e9b03f4fca perf: apply cp.async, XOR swizzle, pre-scaled Q to decode MMA kernel ... Decode MMA kernel previously used scalar global→shared loads with LD=HEAD_DIM+8 padding and per-tile scale multiply. This commit brings it in line with the prefill MMA kernel (which already had these optimizations): - cp.async K/V loads (bypasses registers, halves load instructions) - XOR swizzle: LD=HEAD_DIM instead of HEAD_DIM+8 (zero waste smem) - Pre-scale Q during load (removes per-tile scale multiply in softmax) - Clean up prefill MMA kernel comments (no code change) ~2x speedup on decode (0.47ms→0.24ms at seq_len=512)		2026-07-08 16:15:14 +08:00
..
kernels	perf: apply cp.async, XOR swizzle, pre-scaled Q to decode MMA kernel	2026-07-08 16:15:14 +08:00
tests	perf: add fast-math and vectorization nvcc/cxx build flags	2026-07-07 22:28:32 +08:00
__init__.py	feat: add optional CUDA kernel system (csrc/) + fused GQA decode attention	2026-07-06 12:09:58 +08:00
build.py	perf: add fast-math and vectorization nvcc/cxx build flags	2026-07-07 22:28:32 +08:00