Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
Menu
Open sidebar
gaoqiong
flash-attention
Commits
65f723bb
Commit
65f723bb
authored
Jul 23, 2024
by
Tri Dao
Browse files
Split bwd into more .cu files to speed up compilation
parent
5ca83a9c
Changes
89
Hide whitespace changes
Inline
Side-by-side
Showing
20 changed files
with
20 additions
and
20 deletions
+20
-20
csrc/flash_attn/src/flash_fwd_hdim192_bf16_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim192_bf16_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim192_fp16_causal_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim192_fp16_causal_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim192_fp16_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim192_fp16_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim256_bf16_causal_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim256_bf16_causal_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim256_bf16_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim256_bf16_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim256_fp16_causal_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim256_fp16_causal_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim256_fp16_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim256_fp16_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim32_bf16_causal_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim32_bf16_causal_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim32_bf16_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim32_bf16_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim32_fp16_causal_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim32_fp16_causal_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim32_fp16_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim32_fp16_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim64_bf16_causal_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim64_bf16_causal_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim64_bf16_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim64_bf16_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim64_fp16_causal_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim64_fp16_causal_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim64_fp16_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim64_fp16_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim96_bf16_causal_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim96_bf16_causal_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim96_bf16_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim96_bf16_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim96_fp16_causal_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim96_fp16_causal_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_hdim96_fp16_sm80.cu
csrc/flash_attn/src/flash_fwd_hdim96_fp16_sm80.cu
+1
-1
csrc/flash_attn/src/flash_fwd_split_hdim128_bf16_causal_sm80.cu
...lash_attn/src/flash_fwd_split_hdim128_bf16_causal_sm80.cu
+1
-1
No files found.
csrc/flash_attn/src/flash_fwd_hdim192_bf16_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim192_fp16_causal_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim192_fp16_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim256_bf16_causal_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim256_bf16_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim256_fp16_causal_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim256_fp16_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim32_bf16_causal_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim32_bf16_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim32_fp16_causal_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim32_fp16_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim64_bf16_causal_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim64_bf16_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim64_fp16_causal_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim64_fp16_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim96_bf16_causal_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim96_bf16_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim96_fp16_causal_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_hdim96_fp16_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
csrc/flash_attn/src/flash_fwd_split_hdim128_bf16_causal_sm80.cu
View file @
65f723bb
// Copyright (c) 202
3
, Tri Dao.
// Copyright (c) 202
4
, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
// This file is auto-generated. See "generate_kernels.py"
...
...
Prev
1
2
3
4
5
Next
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment