Stefan He db7343c992 fix per token cuda kernel hidden dim cannot divide by 16 (#8543)

2025-08-01 09:27:18 -07:00

..

bench_activation.py

[AMD] Add silu_and_mul, gelu_and_mul, gelu_tanh_and_mul, and gelu_quick kernels for AMD GPUs (#7135)

2025-07-24 23:44:28 -07:00

bench_awq_dequant.py

…

bench_cutlass_mla.py

[fix] fix cutlass_mla_backend with cuda_graph and add sm_scale for sgl-kernel cutlass_mla (#7184)

2025-06-14 12:45:41 -07:00

bench_dsv3_fused_a_gemm.py

Add dsv3 fused a gemm to sgl-kernel (#7630)

2025-06-29 02:52:24 -07:00

bench_dsv3_router_gemm.py

[Kimi K2] dsv3_router_gemm supports NUM_EXPERTS == 384 (#8013)

2025-08-01 22:01:24 +08:00

bench_fp4_gemm.py

Use FlashInfer FP4 gemm. (#8241)

2025-07-27 01:05:22 -07:00

bench_fp8_blockwise_gemm.py

…

bench_fp8_blockwise_group_gemm.py

Add CUTLASS FP8 Blockscale MoE kernel for Hopper architecture (#7278)

2025-07-02 23:27:03 -07:00

bench_fp8_gemm.py

…

bench_int8_gemm.py

…

bench_lightning_attention_decode.py

…

bench_moe_align_block_size.py

update sgl-kernel for EP: kernel part (#8514)

2025-07-30 22:19:55 -07:00

bench_moe_ep_post_reorder.py

feat: integrate deepgemm into EPMoE (#6821)

2025-06-23 01:38:58 -07:00

bench_moe_ep_pre_reorder.py

…

bench_moe_fused_gate.py

[fix] benchmark : routed_scaling_factor is None (#8059)

2025-07-22 08:55:35 -07:00

bench_moe_silu_and_mul.py

…

bench_moe_topk_softmax.py

[optimize] fuse renormalize into moe_topk_softmax (#7744)

2025-07-03 12:42:44 -07:00

bench_nvfp4_scaled_gemm.py

Add nvfp4 scaled mm benchmark. (#8401)

2025-07-26 23:18:04 -07:00

bench_per_tensor_quant_fp8.py

…

bench_per_token_group_quant_8bit.py

Fix bench script making input data on L2 cache (#7739)

2025-07-27 00:30:24 -07:00

bench_per_token_quant_fp8.py

fix per token cuda kernel hidden dim cannot divide by 16 (#8543)

2025-08-01 09:27:18 -07:00

bench_qserve_w4a8_gemm.py

…