Skip to content

GitLab

Commit 69224aac authored by fsx950223's avatar fsx950223
Browse files

add grouped flash attention api

parent 7409bc5d
Expand all Show whitespace changes
Inline Side-by-side
Showing with 2084 additions and 0 deletions
example/32_batched_gemm_scale_softmax_gemm/CMakeLists.txt
View file @ 69224aac
...@@ -5,7 +5,13 @@ add_example_executable(example_batched_gemm_scale_softmax_gemm_permute_xdl_bf16 ...@@ -5,7 +5,13 @@ add_example_executable(example_batched_gemm_scale_softmax_gemm_permute_xdl_bf16
add_example_executable(example_grouped_gemm_scale_softmax_gemm_permute_xdl_fp16 grouped_gemm_scale_softmax_gemm_permute_xdl_fp16.cpp) add_example_executable(example_grouped_gemm_scale_softmax_gemm_permute_xdl_fp16 grouped_gemm_scale_softmax_gemm_permute_xdl_fp16.cpp)
add_example_executable(example_batched_gemm_lower_triangle_scale_softmax_gemm_permute_xdl_fp16 batched_gemm_lower_triangle_scale_softmax_gemm_permute_xdl_fp16.cpp) add_example_executable(example_batched_gemm_lower_triangle_scale_softmax_gemm_permute_xdl_fp16 batched_gemm_lower_triangle_scale_softmax_gemm_permute_xdl_fp16.cpp)
add_example_executable(example_grouped_gemm_lower_triangle_scale_softmax_gemm_permute_xdl_fp16 grouped_gemm_lower_triangle_scale_softmax_gemm_permute_xdl_fp16.cpp) add_example_executable(example_grouped_gemm_lower_triangle_scale_softmax_gemm_permute_xdl_fp16 grouped_gemm_lower_triangle_scale_softmax_gemm_permute_xdl_fp16.cpp)
add_example_executable(example_grouped_multihead_attention_forward_fp16 grouped_multihead_attention_forward_fp16.cpp)
add_example_executable(example_batched_multihead_attention_forward_fp16 batched_multihead_attention_forward_fp16.cpp)
add_example_executable(example_grouped_multihead_attention_forward_bf16 grouped_multihead_attention_forward_bf16.cpp)
add_example_executable(example_batched_multihead_attention_forward_bf16 batched_multihead_attention_forward_bf16.cpp)
add_example_executable(example_batched_multihead_attention_backward_fp16 batched_multihead_attention_backward_fp16.cpp) add_example_executable(example_batched_multihead_attention_backward_fp16 batched_multihead_attention_backward_fp16.cpp)
add_example_executable(example_grouped_multihead_attention_backward_fp16 grouped_multihead_attention_backward_fp16.cpp)
add_example_executable(example_batched_multihead_attention_backward_pt1_fp16 batched_multihead_attention_backward_pt1_fp16.cpp)
add_custom_target(example_gemm_scale_softmax_gemm) add_custom_target(example_gemm_scale_softmax_gemm)
add_dependencies(example_gemm_scale_softmax_gemm example_batched_gemm_scale_softmax_gemm_xdl_fp16) add_dependencies(example_gemm_scale_softmax_gemm example_batched_gemm_scale_softmax_gemm_xdl_fp16)
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment