aska-0096 4f7be0f2b5 disable the scheduler in v3; bring it back when compiler feature ready.

2025-05-13 08:37:11 +00:00

..

CMakeLists.txt

disable the scheduler in v3; bring it back when compiler feature ready.

2025-05-13 08:37:11 +00:00

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

blockscale preshuffle, pipeline optimized, compiler issue

2025-04-28 10:20:23 +00:00

gemm_multiply_multiply_xdl_fp8_blockscale_bpreshuffle_v1.cpp

optimized decode pipeline for blockscale gemm

2025-03-13 08:52:12 +00:00

gemm_multiply_multiply_xdl_fp8_blockscale_bpreshuffle.cpp

temp save, a performant version.

2025-05-08 07:44:06 +00:00

gemm_multiply_multiply_xdl_fp8_bpreshuffle.cpp

experimental optimization for scale load in blkscale gemm

2025-05-02 07:12:01 +00:00

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.cpp

tempsave

2025-04-27 08:05:20 +00:00

moe_gemm1_xdl_pk_i4.cpp

enable do top k weights in moe stage1 gemm (#2094)

2025-04-18 10:45:49 +08:00

moe_gemm2_xdl_fp8.cpp

blockscale preshuffle, pipeline optimized, compiler issue

2025-04-28 10:20:23 +00:00

moe_gemm2_xdl_pk_i4.cpp

enable do top k weights in moe stage1 gemm (#2094)

2025-04-18 10:45:49 +08:00