dolpm 1b962e2457

[fix] lora benchmarks pass no_lora_flag_cpu (#23774) ...

Signed-off-by: Dylan Maloy <34420038+dolpm@users.noreply.github.com>
Co-authored-by: Jee Jee Li <pandaleefree@gmail.com>

2025-09-17 21:22:25 +08:00

..

deepgemm

[Refactor] Remove Duplicate per_block_cast_to_fp8, Remove Dependencies of DeepGEMM (#21787)

2025-08-01 01:13:27 +00:00

bench_block_fp8_gemm.py

[Perf] Use upstream CUTLASS for SM90 Block FP8 kernel (#23280)

2025-09-11 15:43:14 -07:00

bench_fp8_gemm.py

[Misc] Add SPDX-FileCopyrightText (#20428)

2025-07-04 07:40:42 +00:00

bench_int8_gemm.py

[Benchmark] Refactor benchmark script for fp8 & int8 (#19627)

2025-06-15 15:15:37 +08:00

bench_nvfp4_gemm.py

[Bench] Add NVFP4 GEMM benchmark script (#20578)

2025-07-09 13:23:48 -04:00

bench_per_token_quant_fp8.py

[FP8] Extend per-token-group quantization support to QuantFP8 (#24342)

2025-09-16 18:31:06 -07:00

benchmark_activation.py

[Benchmark] add benchmark for custom activation op (#23908)

2025-09-06 20:12:05 -07:00

benchmark_bitblas.py

[Bugfix] Add proper comparison for package versions (#22314)

2025-08-06 20:31:03 -07:00

benchmark_cutlass_fp4_moe.py

[Hardware][NVIDIA] FP4 MoE kernel optimization (#19110)

2025-06-05 09:48:26 -07:00

benchmark_device_communicators.py

[Kernels] Enable Torch Symmetric Memory All-Reduce By Default (#24111)

2025-09-11 09:45:31 -07:00

benchmark_grouped_gemm_cutlass.py

[Kernel] CUTLASS MoE FP8: Integrate cuda moe permute/unpermute (#23045)

2025-08-20 10:35:26 -04:00

benchmark_layernorm.py

[Misc] Add SPDX-FileCopyrightText (#19100)

2025-06-03 11:20:17 -07:00

benchmark_lora.py

[fix] lora benchmarks pass no_lora_flag_cpu (#23774)

2025-09-17 21:22:25 +08:00

benchmark_machete.py

[kernel] Support W4A8 on Hopper (#23198)

2025-08-24 06:18:04 +00:00

benchmark_marlin.py

[Bugfix][Benchmark] Fix Marlin benchmark (#19929)

2025-06-24 07:25:12 +09:00

benchmark_moe_align_block_size.py

[Refactor] Remove moe_align_block_size_triton (#21335)

2025-07-26 07:08:29 -07:00

benchmark_moe_permute_unpermute.py

Fix CUDA permute/unpermute for use with DeepGemm Moe (#17934)

2025-07-27 07:08:00 -07:00

benchmark_moe.py

[benchmark] Add triton version in the moe tuned config (#24769)

2025-09-16 14:10:54 +08:00

benchmark_mrope.py

[FEAT] [Performance] Add triton mrope to replace the torch code path (#22375)

2025-08-09 11:50:03 -07:00

benchmark_paged_attention.py

[Misc] Add SPDX-FileCopyrightText (#19100)

2025-06-03 11:20:17 -07:00

benchmark_per_token_group_quant.py

[Test] Add Benchmark and Unit Test for per_token_group_quant (#21860)

2025-07-30 07:15:02 -07:00

benchmark_polynorm.py

[Model] New model support for Motif-1-Tiny (#23414)

2025-09-10 23:29:40 -07:00

benchmark_quant.py

[Misc] Add SPDX-FileCopyrightText (#19100)

2025-06-03 11:20:17 -07:00

benchmark_reshape_and_cache_flash.py

[Perf] Optimize reshape_and_cache_flash CUDA Kernel (#22036)

2025-08-01 19:18:51 -04:00

benchmark_rmsnorm.py

[Misc] Add SPDX-FileCopyrightText (#19100)

2025-06-03 11:20:17 -07:00

benchmark_rope.py

[Misc] Add SPDX-FileCopyrightText (#19100)

2025-06-03 11:20:17 -07:00

benchmark_shapes.py

[Misc] Add SPDX-FileCopyrightText (#19100)

2025-06-03 11:20:17 -07:00

benchmark_silu_mul_fp8_quant.py

[Kernels][DP/EP] Optimize Silu Kernel for R1 (#24054)

2025-09-13 00:17:27 -07:00

benchmark_trtllm_decode_attention.py

[Flashinfer] Support Flashinfer TRTLLM FP8-qkv BF16/FP16-out Attention Kernel (#23647)

2025-09-08 20:53:07 -07:00

benchmark_trtllm_prefill_attention.py

[Flashinfer] Support Flashinfer TRTLLM FP8-qkv BF16/FP16-out Attention Kernel (#23647)

2025-09-08 20:53:07 -07:00

benchmark_w8a8_block_fp8.py

[Doc]: fix typos in various files (#24726)

2025-09-12 06:43:12 -07:00

graph_machete_bench.py

[Misc] Add SPDX-FileCopyrightText (#19100)

2025-06-03 11:20:17 -07:00

requirements.txt

[Kernel] (2/N) Machete - Integrate into CompressedTensorsWNA16 and GPTQMarlin (#7701)

2024-09-23 13:46:26 -04:00

utils.py

[Misc] Add SPDX-FileCopyrightText (#19100)

2025-06-03 11:20:17 -07:00

weight_shapes.py

[kernel] Support W4A8 on Hopper (#23198)

2025-08-24 06:18:04 +00:00