Revert "[Perf][GLM-5.2] Blackwell decode optimizations" (#49768)

This commit is contained in:
Woosuk Kwon
2026-07-24 16:04:59 -07:00
committed by GitHub
parent 33c4f3551c
commit 213f681f81
29 changed files with 138 additions and 2139 deletions
-4
View File
@@ -330,10 +330,6 @@ STABLE_TORCH_LIBRARY_FRAGMENT(_C, ops) {
// conditionally compiled so impl registration is in source file
ops.def("fp32_router_gemm(Tensor! output, Tensor mat_a, Tensor mat_b) -> ()");
// BF16 skinny GEMM (M<=32, weight-BW-bound shapes, e.g. MTP eh_proj).
// conditionally compiled so impl registration is in source file
ops.def("bf16_skinny_gemm(Tensor! output, Tensor mat_a, Tensor mat_b) -> ()");
// reorder weight for AllSpark Ampere W8A16 Fused Gemm kernel
ops.def(
"rearrange_kn_weight_as_n32k16_order(Tensor b_qweight, Tensor b_scales, "