Skip to content

Use EasyBlock for PyTorch 1.3.1 and remove glog & gflags #11325

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

Merged
lexming merged 7 commits into from
Oct 1, 2020
Merged

Use EasyBlock for PyTorch 1.3.1 and remove glog & gflags #11325

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
e7c4c54
Use EasyBlock for PyTorch 1.3.1 and remove glog&gflags
Flamefire Sep 22, 2020
fc50da8
Exclude autograd test
Flamefire Sep 23, 2020
d0835e4
Exclude jit and cuda test on POWER
Flamefire Sep 23, 2020
defb558
Exclude test_nn on POWER
Flamefire Sep 23, 2020
fb9a181
Exclude test_torch
Flamefire Sep 24, 2020
c926fc6
Set OMP_NUM_THREADS=1 to work around POWER issues
Flamefire Sep 24, 2020
2159380
Avoid downloading PSIMD
Flamefire Sep 25, 2020
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
60 changes: 20 additions & 40 deletions easybuild/easyconfigs/p/PyTorch/PyTorch-1.3.1-foss-2019b-Python-3.7.4.eb
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,3 @@
easyblock = 'PythonPackage'

name = 'PyTorch'
version = '1.3.1'
versionsuffix = '-Python-%(pyver)s'
Expand Down Expand Up @@ -73,12 +71,6 @@ sources = [
'filename': 'onnx-tensorrt-20190425.tar.gz',
'extract_cmd': local_extract_cmd_pattern % (local_pytorchthirdpartydir, 'onnx-tensorrt'),
},
{
'source_urls': ['https://github.com/google/protobuf/archive'],
'download_filename': '48cb18e5c419ddd23d9badcfe4e9df7bde1979b2.tar.gz',
'filename': 'protobuf-20180727.tar.gz',
'extract_cmd': local_extract_cmd_pattern % (local_pytorchthirdpartydir, 'protobuf'),
},
{
'source_urls': ['https://github.com/Maratyszcza/pthreadpool/archive'],
'download_filename': '13da0b4c21d17f94150713366420baaf1b5a46f4.tar.gz',
Expand Down Expand Up @@ -169,12 +161,6 @@ sources = [
'filename': 'gemmlowp-20181127.tar.gz',
'extract_cmd': local_extract_cmd_pattern % (local_pytorchthirdpartydir, 'gemmlowp/gemmlowp'),
},
{
'source_urls': ['https://github.com/NVIDIA/nccl/archive'],
'download_filename': '7c72dee660e4d055b81721dd6b03e4e1c0a983cf.tar.gz',
'filename': 'nccl-20190625.tar.gz',
'extract_cmd': local_extract_cmd_pattern % (local_pytorchthirdpartydir, 'nccl/nccl'),
},
{
'source_urls': ['https://github.com/NVlabs/cub/archive'],
'download_filename': '285aeebaa34b0e8a7670867a2e66c1a52d998d6a.tar.gz',
Expand Down Expand Up @@ -210,6 +196,7 @@ patches = [
'%(name)s-1.2.0_fix-findAVX.patch',
'%(name)s-1.2.0_disable-tests-ppc64le.patch',
'%(name)s-%(version)s_fix-missing-sleef_h.patch',
'%(name)s-1.4.0_fix-missing-source-dir.patch',
]
checksums = [
'ab6feb5044f7d36f6e93dce4668d8c593e89d34aca7023fd99a38d215ca9dfc0', # v1.3.1.tar.gz
Expand All @@ -222,7 +209,6 @@ checksums = [
'c7682e9007ddfd94072647abab3e89ffd9084089460ae47d67060974467b58bf', # benchmark-20180525.tar.gz
'02407529fd8431cf627d45638b4d9a6c02b8aa5bf50d73b8fa12217918c57ff4', # pybind11-20170919.tar.gz
'1a98c6fece7878c88b59ffe651b427c15b4929c801a09655cd14ca44553787c3', # onnx-tensorrt-20190425.tar.gz
'f5a35e17fb07f3b13517264cd17a089636fcbb2912f9df7bef7414058969a8d2', # protobuf-20180727.tar.gz
'90ab4f60ae0e99c0172dd5201dccc6de6c8f384a21b3d959588f4c1e00beec0f', # pthreadpool-20181008.tar.gz
'07625551202a1f35fe5e826669ad495962b26f6a139290e90b321c5996f002ef', # FXdiv-20181016.tar.gz
'afd27417c6df1c85f79a68361055dd3adc292174913c39c1f58b698b9cfd7926', # psimd-20180906.tar.gz
Expand All @@ -238,7 +224,6 @@ checksums = [
'0de43b6eaddd356f1d6cd164f73f37faf2f6c96fd684e1f7ea543ce49c1d144e', # benchmark-20180606.tar.gz
'2ec954f18cec50a7063a7358ce555f7e11788a7f6d4e7e597d83687dc2f3b989', # eigen-20190126.tar.gz
'fdd6f08bdb33d33f4df516ffb91730fdb163479c19502cfc983083fd9cf43bfa', # gemmlowp-20181127.tar.gz
'ef8b624ef28e9e2d0e162065932dd2f25de8d6d1dc58df91bd36977e4bc0224c', # nccl-20190625.tar.gz
'7224b03af4acbc54525105bb42e3ecd75c66a3a5b47e8a725ab008467c4109f9', # cub-20170829.tar.gz
'b444b4428adc4befc777d9a1af2fd9bbf5f5a2339913668357196c5133e9cc7b', # foxi-20190424.tar.gz
'b8fb8b4871cc3da4bded01c476957a08f51949c20f834f39f90f9e10811d6dfe', # ideep-20190912.tar.gz
Expand All @@ -249,48 +234,43 @@ checksums = [
'001c9bf604aebe4b39ccad15332a71130b07b780c539ceca84d6c64cd6fc8a68', # PyTorch-1.2.0_fix-findAVX.patch
'c4183bcb29a8bcbadea0341e93a3a32afdf860aa31331b768e787d899183da92', # PyTorch-1.2.0_disable-tests-ppc64le.patch
'1337647ff64a1208d1e401fc84052d0bc6174b133cec2f3521319cb593f524fa', # PyTorch-1.3.1_fix-missing-sleef_h.patch
'797987fb9c9bf9f1d75a1be878ddf9f418f9524006b0985ca8e6d65d4e2b6998', # PyTorch-1.4.0_fix-missing-source-dir.patch
]

builddependencies = [
('CMake', '3.15.3'),
('hypothesis', '4.44.2', versionsuffix),
('Ninja', '1.9.0'),
]

dependencies = [
('Ninja', '1.9.0'),
('protobuf', '3.10.0'),
('Python', '3.7.4'),
('SciPy-bundle', '2019.10', versionsuffix),
('PyYAML', '5.1.2'),
('MPFR', '4.0.2'),
('GMP', '6.1.2'),
('numactl', '2.0.12'),
('FFmpeg', '4.2.1'),
('gflags', '2.2.2'),
('glog', '0.4.0'),
('Pillow', '6.2.1'),
]

download_dep_fail = True

prebuildopts = 'PYTORCH_BUILD_VERSION=%(version)s PYTORCH_BUILD_NUMBER=1 VERBOSE=1 '
prebuildopts += 'MAX_JOBS=%(parallel)s '
prebuildopts += 'LDFLAGS="$LDFLAGS -ldl" '
prebuildopts += 'USE_FFMPEG=ON USE_IBVERBS=1 USE_GFLAGS=ON USE_GLOG=ON '

preinstallopts = prebuildopts

runtest = 'export PYTHONPATH=%(builddir)s/%(namelower)s-%(version)s/' \
'build/lib.linux-%(arch)s-%(pyshortver)s:$PYTHONPATH '
runtest += '&& export LD_LIBRARY_PATH=%(builddir)s/%(namelower)s-%(version)s/torch/lib64:$LD_LIBRARY_PATH '
runtest += '&& python test/run_test.py --verbose '

options = {'modulename': 'torch'}

postinstallcmds = ['cp -a %(builddir)s/%(namelower)s-%(version)s/torch/lib*/* %(installdir)s/lib/']

sanity_check_paths = {
'files': [],
'dirs': ['lib/python%(pyshortver)s/site-packages'],
excluded_tests = {
'POWER': [
# https://github.com/pytorch/pytorch/issues/41186
'autograd',
# https://github.com/pytorch/pytorch/issues/41400
'jit',
# https://github.com/pytorch/pytorch/issues/41469
'nn',
# Sporadically fails for unknown reason. Fixed in newer versions
'torch',
],
}

# OpenMP causes e.g. torch.matmul to return invalid results at least on POWER
# See e.g. https://github.com/pytorch/pytorch/issues/3716#issuecomment-387776917
runtest = 'cd test && OMP_NUM_THREADS=1 %(python)s run_test.py --verbose %(excluded_tests)s'
tests = ['PyTorch-check-cpp-extension.py']

moduleclass = 'devel'
59 changes: 24 additions & 35 deletions easybuild/easyconfigs/p/PyTorch/PyTorch-1.3.1-fosscuda-2019b-Python-3.7.4.eb
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,3 @@
easyblock = 'PythonPackage'

name = 'PyTorch'
version = '1.3.1'
versionsuffix = '-Python-%(pyver)s'
Expand Down Expand Up @@ -73,12 +71,6 @@ sources = [
'filename': 'onnx-tensorrt-20190425.tar.gz',
'extract_cmd': local_extract_cmd_pattern % (local_pytorchthirdpartydir, 'onnx-tensorrt'),
},
{
'source_urls': ['https://github.com/google/protobuf/archive'],
'download_filename': '48cb18e5c419ddd23d9badcfe4e9df7bde1979b2.tar.gz',
'filename': 'protobuf-20180727.tar.gz',
'extract_cmd': local_extract_cmd_pattern % (local_pytorchthirdpartydir, 'protobuf'),
},
{
'source_urls': ['https://github.com/Maratyszcza/pthreadpool/archive'],
'download_filename': '13da0b4c21d17f94150713366420baaf1b5a46f4.tar.gz',
Expand Down Expand Up @@ -204,6 +196,7 @@ patches = [
'%(name)s-1.2.0_fix-findAVX.patch',
'%(name)s-1.2.0_disable-tests-ppc64le.patch',
'%(name)s-%(version)s_fix-missing-sleef_h.patch',
'%(name)s-1.4.0_fix-missing-source-dir.patch',
]
checksums = [
'ab6feb5044f7d36f6e93dce4668d8c593e89d34aca7023fd99a38d215ca9dfc0', # v1.3.1.tar.gz
Expand All @@ -216,7 +209,6 @@ checksums = [
'c7682e9007ddfd94072647abab3e89ffd9084089460ae47d67060974467b58bf', # benchmark-20180525.tar.gz
'02407529fd8431cf627d45638b4d9a6c02b8aa5bf50d73b8fa12217918c57ff4', # pybind11-20170919.tar.gz
'1a98c6fece7878c88b59ffe651b427c15b4929c801a09655cd14ca44553787c3', # onnx-tensorrt-20190425.tar.gz
'f5a35e17fb07f3b13517264cd17a089636fcbb2912f9df7bef7414058969a8d2', # protobuf-20180727.tar.gz
'90ab4f60ae0e99c0172dd5201dccc6de6c8f384a21b3d959588f4c1e00beec0f', # pthreadpool-20181008.tar.gz
'07625551202a1f35fe5e826669ad495962b26f6a139290e90b321c5996f002ef', # FXdiv-20181016.tar.gz
'afd27417c6df1c85f79a68361055dd3adc292174913c39c1f58b698b9cfd7926', # psimd-20180906.tar.gz
Expand All @@ -242,54 +234,51 @@ checksums = [
'001c9bf604aebe4b39ccad15332a71130b07b780c539ceca84d6c64cd6fc8a68', # PyTorch-1.2.0_fix-findAVX.patch
'c4183bcb29a8bcbadea0341e93a3a32afdf860aa31331b768e787d899183da92', # PyTorch-1.2.0_disable-tests-ppc64le.patch
'1337647ff64a1208d1e401fc84052d0bc6174b133cec2f3521319cb593f524fa', # PyTorch-1.3.1_fix-missing-sleef_h.patch
'797987fb9c9bf9f1d75a1be878ddf9f418f9524006b0985ca8e6d65d4e2b6998', # PyTorch-1.4.0_fix-missing-source-dir.patch
]

builddependencies = [
('CMake', '3.15.3'),
('hypothesis', '4.44.2', versionsuffix),
('Ninja', '1.9.0'),
]

dependencies = [
('Ninja', '1.9.0'),
('protobuf', '3.10.0'),
('Python', '3.7.4'),
('SciPy-bundle', '2019.10', versionsuffix),
('PyYAML', '5.1.2'),
('MPFR', '4.0.2'),
('GMP', '6.1.2'),
('numactl', '2.0.12'),
('FFmpeg', '4.2.1'),
('gflags', '2.2.2'),
('glog', '0.4.0'),
('Pillow', '6.2.1'),
('cuDNN', '7.6.4.38'),
('magma', '2.5.1'),
('NCCL', '2.4.8'),
]

download_dep_fail = True

prebuildopts = 'PYTORCH_BUILD_VERSION=%(version)s PYTORCH_BUILD_NUMBER=1 VERBOSE=1 '
prebuildopts += 'MAX_JOBS=%(parallel)s '
prebuildopts += 'LDFLAGS="$LDFLAGS -ldl" '
prebuildopts += 'USE_FFMPEG=ON USE_IBVERBS=1 USE_GFLAGS=ON USE_GLOG=ON '
prebuildopts += 'CUDNN_LIB_DIR=$EBROOTCUDNN/lib64 CUDNN_INCLUDE_DIR=$EBROOTCUDNN/include '
prebuildopts += 'USE_SYSTEM_NCCL=1 NCCL_INCLUDE_DIR=$EBROOTNCCL/include '
prebuildopts += 'TORCH_CUDA_ARCH_LIST="3.5 3.7 5.2 6.0 6.1 7.0 7.2 7.5" '
# default CUDA compute capabilities to use (override via --cuda-compute-capabilities)
cuda_compute_capabilities = ['3.5', '3.7', '5.2', '6.0', '6.1', '7.0', '7.2', '7.5']

preinstallopts = prebuildopts

runtest = 'export PYTHONPATH=%(builddir)s/%(namelower)s-%(version)s/' \
'build/lib.linux-%(arch)s-%(pyshortver)s:$PYTHONPATH '
runtest += '&& export LD_LIBRARY_PATH=%(builddir)s/%(namelower)s-%(version)s/torch/lib64:$LD_LIBRARY_PATH '
runtest += '&& python test/run_test.py --verbose '

options = {'modulename': 'torch'}

postinstallcmds = ['cp -a %(builddir)s/%(namelower)s-%(version)s/torch/lib*/* %(installdir)s/lib/']

sanity_check_paths = {
'files': [],
'dirs': ['lib/python%(pyshortver)s/site-packages'],
excluded_tests = {
'POWER': [
# https://github.com/pytorch/pytorch/issues/41186
'autograd',
# https://github.com/pytorch/pytorch/issues/41400
'jit',
# https://github.com/pytorch/pytorch/issues/41469
'nn',
# Sporadically fails for unknown reason. Fixed in newer versions
'torch',
# Unknown
'cuda',
],
}

# OpenMP causes e.g. torch.matmul to return invalid results at least on POWER
# See e.g. https://github.com/pytorch/pytorch/issues/3716#issuecomment-387776917
runtest = 'cd test && OMP_NUM_THREADS=1 %(python)s run_test.py --verbose %(excluded_tests)s'
tests = ['PyTorch-check-cpp-extension.py']

moduleclass = 'devel'