Gregory Shtrasberg 32aa74c09c

[ROCm][FP8][Kernel] FP8 quantization fused into Custom Paged Attention (#17139) ...

Signed-off-by: Gregory Shtrasberg <Gregory.Shtrasberg@amd.com>

2025-05-07 07:12:35 -07:00

..

attention

[NVIDIA] Support Cutlass MLA for Blackwell GPUs (#16032)

2025-04-27 06:29:21 -07:00

core

[torch.compile] Add torch inductor pass for fusing silu_and_mul with subsequent scaled_fp8_quant operations (#10867)

2025-05-01 07:59:28 -07:00

cpu

Make key optional for rotary embedding (#17566)

2025-05-07 00:11:46 -07:00

cutlass_extensions

[Kernel] CUTLASS grouped gemm fp8 MoE kernel (#13972)

2025-03-27 00:54:44 +00:00

mamba

[BugFix] fix some typos found by typos. (#16314)

2025-04-09 03:43:59 -07:00

moe

Removed unused marlin cuda code (#17684)

2025-05-06 17:59:47 -07:00

prepare_inputs

[Misc][Easy] Annotate unused vars in the csrc files (#14798)

2025-03-15 12:40:09 +08:00

quantization

[Kernel] GGUF MoeVec kernel (#16780)

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

rocm

[ROCm][FP8][Kernel] FP8 quantization fused into Custom Paged Attention (#17139)

2025-05-07 07:12:35 -07:00

sparse/cutlass

[BugFix/Build] Fix sparse kernels not getting built on hopper (#14572)

2025-03-11 17:09:03 +00:00

activation_kernels.cu

[Kernel] Support MulAndSilu (#11624)

2025-01-15 02:29:53 +00:00

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

dispatch_utils.h

dynamic distpatch of fp8 kernels (#14245)

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

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

Make key optional for rotary embedding (#17566)

2025-05-07 00:11:46 -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

Make key optional for rotary embedding (#17566)

2025-05-07 00:11:46 -07:00

torch_bindings.cpp

Make key optional for rotary embedding (#17566)

2025-05-07 00:11:46 -07:00

type_convert.cuh

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

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