Duncan Moss 5923ab9524

[fix]: disable cutlass block scaled group gemm for EP (#20781) ...

Signed-off-by: Duncan Moss <djm.moss@gmail.com>

2025-07-11 02:39:18 +00:00

..

attention

[Bugfix] Fix some narrowing conversion warnings (#20141)

2025-06-27 09:01:28 -07:00

core

[Kernel] fp4 marlin kernel (#17687)

2025-05-10 19:58:49 -07:00

cpu

[Hardware][CPU] Vllm int8 quantization enablement for ARM CPU (#14129)

2025-07-10 15:59:04 +00:00

cutlass_extensions

Replace multiply_add with homogeneous_multiply_add to Address Clang Template Parameter Issue (#20142)

2025-07-09 00:30:18 +00:00

mamba/mamba_ssm

[Kernel] Triton implementation of causal-conv1d for Mamba-based models (#18218)

2025-07-09 12:53:55 -07:00

moe

remove unused variables in marlin_template.h (#20236)

2025-07-02 00:51:52 +00:00

prepare_inputs

[MISC] Remove unused variableds in C++ (#19609)

2025-06-15 20:05:28 -07:00

quantization

[fix]: disable cutlass block scaled group gemm for EP (#20781)

2025-07-11 02:39:18 +00:00

quickreduce

[Feature] add quick all reduce (#19744)

2025-06-26 20:54:24 -07:00

rocm

[Fix][ROCm] Remove unused variables to fix build error on GFX11/12 (#19891)

2025-06-27 07:14:44 -07:00

sparse/cutlass

[feat]: CUTLASS block scaled group gemm for SM100 (#19757)

2025-07-04 12:58:04 -06:00

activation_kernels.cu

Modularize fused experts and integrate PPLX kernels (#15956)

2025-05-14 13:11:54 -07:00

cache_kernels.cu

Allocate kv_cache with stride order (#16605)

2025-04-25 22:03:31 -07:00

cache.h

[Attention] MLA with chunked prefill (#12639)

2025-02-21 15:30:12 -08:00

cuda_compat.h

[Kernel][ROCm][AMD] enable fused topk_softmax kernel for moe layer (#4927)

2024-06-02 14:13:26 -07:00

cuda_utils_kernels.cu

[NVIDIA] Support nvfp4 quantization (#12784)

2025-02-12 19:51:51 -08:00

cuda_utils.h

[Attention] MLA with chunked prefill (#12639)

2025-02-21 15:30:12 -08:00

cuda_view.cu

[V1] Fully Transparent Implementation of CPU Offloading (#15354)

2025-03-31 20:22:34 +08:00

cumem_allocator.cpp

[core] improve error handling when wake up from sleep mode (#12981)

2025-02-10 09:38:57 +08:00

custom_all_reduce_test.cu

[Distributed] Add custom allreduce support for ROCM (#14125)

2025-03-31 22:49:12 -07:00

custom_all_reduce.cu

[Distributed] Add custom allreduce support for ROCM (#14125)

2025-03-31 22:49:12 -07:00

custom_all_reduce.cuh

fix: spelling (#16466)

2025-04-11 23:24:22 -07:00

custom_quickreduce.cu

[Feature] add quick all reduce (#19744)

2025-06-26 20:54:24 -07:00

dispatch_utils.h

Modularize fused experts and integrate PPLX kernels (#15956)

2025-05-14 13:11:54 -07:00

layernorm_kernels.cu

[Kernel] Use fused rmsnorm for some models like qwen3 series (#17735)

2025-05-06 23:10:02 -07:00

layernorm_quant_kernels.cu

dynamic distpatch of fp8 kernels (#14245)

2025-03-11 10:54:56 -04:00

ops.h

[Kernel] Triton implementation of causal-conv1d for Mamba-based models (#18218)

2025-07-09 12:53:55 -07:00

permute_cols.cu

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

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

pos_encoding_kernels.cu

[Kernel] Have rotary embeddings support tensors (#18046)

2025-05-14 15:43:55 -07:00

sampler.cu

[BUG] Fix #20484. Support empty sequence in cuda penalty kernel (#20491)

2025-07-05 19:38:02 -07:00

torch_bindings.cpp

[Kernel] Triton implementation of causal-conv1d for Mamba-based models (#18218)

2025-07-09 12:53:55 -07:00

type_convert.cuh

[torch.compile] Fuse RMSNorm with quant (#9138)

2024-11-08 21:20:08 +00:00