ViperEkura 8055027df7 perf: enable paged MMA kernel for page_size=1 ...

- Replace per-tile page lookup with per-element lookup in load_tile
- Remove page_ok gate and scalar fallback in launch_paged_decode_mma
- Unified path works for any page_size (L1-cached when page_size >= BC)
- HBM BW: 12% → 73%, decode throughput: 2,250 → 2,606 tok/s (B=32)
- Scales to 5,232 tok/s at B=128 (2.54x vs torch native)

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

..

kernels

perf: enable paged MMA kernel for page_size=1

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

tests

feat: unify attention backend with multi-dim mask support

2026-07-30 20:38:34 +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

chore: bump version to 1.3.9

2026-07-12 21:04:20 +08:00