Jinzhen Lin 750f4cabfa

[Kernel] optimize moe_align_block_size for cuda graph and large num_experts (e.g. DeepSeek-V3) (#12222) ...

Signed-off-by: Jinzhen Lin <linjinzhen@hotmail.com>
Co-authored-by: Michael Goin <mgoin@redhat.com>
Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>

2025-01-20 16:42:16 -08:00

..

marlin_kernels

[Kernel] Zero point support in fused MarlinMoE kernel + AWQ Fused MoE (#8973)

2024-10-04 12:34:44 -06:00

marlin_moe_ops.cu

[Bugfix] Fix support for dimension like integers and ScalarType (#9299)

2024-10-17 19:08:34 +00:00

moe_align_sum_kernels.cu

[Kernel] optimize moe_align_block_size for cuda graph and large num_experts (e.g. DeepSeek-V3) (#12222)

2025-01-20 16:42:16 -08:00

moe_ops.h

[Performance][Kernel] Fused_moe Performance Improvement (#9384)

2024-10-24 15:37:52 -07:00

topk_softmax_kernels.cu

[Kernel][Misc] Use TORCH_LIBRARY instead of PYBIND11_MODULE for custom ops (#5047)

2024-06-09 16:23:30 -04:00

torch_bindings.cpp

[Performance][Kernel] Fused_moe Performance Improvement (#9384)

2024-10-24 15:37:52 -07:00