Skip to content

Fast and generic implementation using OpenMP and CUDA #45

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
shikishima-TasakiLab wants to merge 14 commits into
base: main
Choose a base branch
from
Open

Fast and generic implementation using OpenMP and CUDA #45

Changes from 2 commits
Commits
Show all changes
14 commits
Select commit Hold shift + click to select a range
221b32b
First commit
Jun 12, 2021
c32b399
CPU-only support
Jun 13, 2021
41a5fb0
Update README.md
shikishima-TasakiLab Jun 13, 2021
1140e73
Remove "AutoNonVariableTypeMode".
Jun 27, 2021
684769a
Merge branch 'main' of https://github.com/shikishima-TasakiLab/Involu…
Jun 27, 2021
4b036d3
Merge remote-tracking branch 'upstream/main' into main
Jun 29, 2021
ce9100f
Compatible with PyTorch 1.7.0 or later
Jul 4, 2021
4dc6c8f
Update README.md
shikishima-TasakiLab Jul 4, 2021
57a6d6a
Merge remote-tracking branch 'upstream/main' into main
Jul 4, 2021
0c190bc
Fixed the value of CUDA_MAX_THREADS
Jul 10, 2021
353c22a
Fixed auto-casting to float32 when using AMP.
Jul 10, 2021
265c309
Merge remote-tracking branch 'upstream/main' into main
Jul 10, 2021
2319dea
Support fatbin.
Jul 13, 2021
7734b83
Merge remote-tracking branch 'upstream/main' into main
Jul 13, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 9 additions & 1 deletion setup.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,12 +8,20 @@

EXTENSION = []

CC = ['52', '53', '60', '61', '62', '70', '72', '75', '80']

if os.getenv('USE_OPENMP', '1') == '1':
EXTRA_COMPILE_ARGS.append('-fopenmp')

if os.getenv('USE_CUDA', '1') == '1':
EXTRA_COMPILE_ARGS.append('-DUSE_CUDA')

GENERATE_CODES = []

for cc in CC:
GENERATE_CODES.append('--generate-code')
GENERATE_CODES.append(f'arch=compute_{cc},code=compute_{cc}')

EXTENSION.append(
CUDAExtension(
name='involution',
Expand All @@ -27,7 +35,7 @@
],
extra_compile_args={
'cxx': EXTRA_COMPILE_ARGS,
'nvcc': ['-O3'],
'nvcc': ['-O3'] + GENERATE_CODES,
}
)
)
Copy link
Author

@shikishima-TasakiLab shikishima-TasakiLab Jul 13, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@d-li14
I changed the arguments of NVCC to be optimized for different architectures.
Hopefully this will increase the speed.

Copy link
Owner

@d-li14 d-li14 Jul 13, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@shikishima-TasakiLab
Thanks for your efforts. However, the new code still does not lead to an expected speedup from my side.

Expand Down