huaiguxu b12ae84a40 Huaiguxu/moe fp8 pertoken scale fix (#2391) ...

* fix pertoken_scale a_scale dimension

* clang-format

* Fix moe_gemm2_fp8 perTokenScale reference and example.

[ROCm/composable_kernel commit: e1c5172fdb]

2025-06-27 10:24:34 +08:00

..

CMakeLists.txt

check for if misched-bottomup flag is valid (#2341)

2025-06-13 13:34:22 -07:00

gemm_add_add_xdl_fp16.cpp

[GEMM] gemm_universal related optimization (#1453)

2024-08-14 10:42:30 +08:00

gemm_multiply_multiply_xdl_fp8_ab_scale.cpp

Add MoE & FP8 Blockscale WP Kernels for GFX950 (#2297)

2025-06-12 09:25:59 +08:00

gemm_multiply_multiply_xdl_fp8_blockscale_bpreshuffle.cpp

Add MoE & FP8 Blockscale WP Kernels for GFX950 (#2297)

2025-06-12 09:25:59 +08:00

gemm_multiply_multiply_xdl_fp8_bpreshuffle.cpp

Add MoE & FP8 Blockscale WP Kernels for GFX950 (#2297)

2025-06-12 09:25:59 +08:00

gemm_multiply_multiply_xdl_fp8.cpp

Add the instance of MBlock=144 for GemmMultiplyMultiply (#1955)

2025-03-07 13:44:06 -08:00

gemm_multiply_multiply_xdl_fp16_bpreshuffle.cpp

add preshuffle gemm fp16 (#2036)

2025-04-16 10:53:21 +08:00

gemm_multiply_multiply_xdl_int8.cpp

[A8W8 GEMM] Optimized weight-preshuffled implementation & add quantization datatype for CK TILE rms_norm (#1862)

2025-02-20 14:00:27 -08:00

moe_gemm1_xdl_fp8_blockscale.cpp

Add MoE & FP8 Blockscale WP Kernels for GFX950 (#2297)

2025-06-12 09:25:59 +08:00

moe_gemm1_xdl_fp8.cpp

Add MoE & FP8 Blockscale WP Kernels for GFX950 (#2297)

2025-06-12 09:25:59 +08:00

moe_gemm1_xdl_pk_i4.cpp

fix moe i4 example bug (#2139)

2025-04-28 09:49:31 -07:00

moe_gemm2_xdl_fp8_blockscale.cpp

Add MoE & FP8 Blockscale WP Kernels for GFX950 (#2297)

2025-06-12 09:25:59 +08:00

moe_gemm2_xdl_fp8.cpp

Huaiguxu/moe fp8 pertoken scale fix (#2391)

2025-06-27 10:24:34 +08:00

moe_gemm2_xdl_pk_i4.cpp

fix moe gemm2 for gfx950 (#2164)

2025-05-09 08:25:31 -07:00