root
cdcf115037
feat: CUTLASS Cu10 grouped GEMM — real device verified
BI-V100 real device results:
moe_group_gemm: err=0.000015 PASS
moe_decode_cutlass: NaN=False PASS
cutlass grouped: 4.77ms vs torch.mm loop: 9.38ms → 1.97x speedup
Fix: gemm_grouped.cu ldb=K (not N) for ColumnMajor B view
Link: -lcuinfer from /usr/local/corex-3.2.3/lib64/libcuinfer.so.7
2026-08-17 04:18:29 +00:00
..
2026-08-17 04:18:29 +00:00
2026-08-14 07:48:52 +00:00
2026-08-10 09:12:05 +00:00
2026-08-15 14:26:24 +00:00
2026-08-17 04:18:29 +00:00
2026-08-15 14:26:24 +00:00
2026-08-15 14:26:24 +00:00
2026-08-13 03:19:39 +00:00
2026-08-15 13:08:24 +00:00
2026-08-17 04:18:29 +00:00
2026-08-17 04:18:29 +00:00
2026-08-15 14:15:47 +00:00
2026-08-15 14:15:47 +00:00
2026-08-15 14:13:13 +00:00
2026-08-12 01:33:24 +00:00
2026-08-10 04:01:21 +00:00
2026-08-15 14:15:47 +00:00
2026-08-15 06:15:17 +00:00
2026-08-12 01:33:24 +00:00
2026-08-10 07:50:28 +00:00