Files
Xiaoze Fan 7e353d11cd fix(attention): keep multi-row triton decode free of register spills on sm_80 to sm_120
- pass maxnreg=255 to multi-row stage-1 launches on CUDA: without .maxnreg ptxas holds these tiles at 80-128 registers for occupancy and spills 8-16 B on sm_80 to sm_120
- pick block_n and warps from a launch sweep under maxnreg, among the variants that do not spill on sm_80, 86, 89, 90, 100 and 120; quantized head_dim 512 takes 16 rows x heads again
2026-11-13 01:52:17 +00:00
..
2026-08-11 22:53:25 +00:00
2026-08-11 22:53:25 +00:00
2026-08-11 22:53:25 +00:00
2026-08-11 22:53:25 +00:00