-
Notifications
You must be signed in to change notification settings - Fork 720
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.
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
{compiler}[AOMP/11.12-0] Add AOMP compiler module #12018
Closed
Closed
Changes from all commits
Commits
Show all changes
5 commits
Select commit
Hold shift + click to select a range
File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
59 changes: 59 additions & 0 deletions
59
easybuild/easyconfigs/a/AOMP/AOMP-11.12-0-GCCcore-9.3.0.eb
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,59 @@ | ||
easyblock = 'Binary' | ||
|
||
name = 'AOMP' | ||
version = '11.12-0' | ||
|
||
homepage = 'https://github.com/ROCm-Developer-Tools/aomp' | ||
description = "AMD fork of LLVM, setup for OpenMP offloading to Accelerators" | ||
|
||
toolchain = {'name': 'GCCcore', 'version': '9.3.0'} | ||
|
||
source_urls = ['https://github.com/ROCm-Developer-Tools/aomp/releases/download/rel_%(version)s/'] | ||
sources = ['aomp-%(version)s.tar.gz'] | ||
patches = ['build_aomp-%(version)s.patch'] | ||
checksums = [ | ||
'b01bbed55c4851cfebc69d5be42bafdfb8b43b82e0b6715fe5e1e86db47272e2', # aomp-11.12-0.tar.gz | ||
'c5ccdd00e270731a353461a8f158cdec556689742d501163c8059e739d0588b7', # build_aomp-11.12-0.patch | ||
] | ||
|
||
builddependencies = [ | ||
('CMake', '3.16.4'), | ||
('Perl', '5.30.2'), | ||
('Python', '3.8.2'), | ||
('pkg-config', '0.29.2'), | ||
] | ||
|
||
dependencies = [ | ||
('binutils', '2.34'), | ||
('libffi', '3.3'), | ||
('numactl', '2.0.13'), | ||
] | ||
|
||
extract_sources = True | ||
dontcreateinstalldir = True | ||
|
||
preinstallopts = 'AOMP=%(installdir)s ' | ||
preinstallopts += 'AOMP_REPOS="%(builddir)s/aomp%(version_major)s" ' | ||
preinstallopts += 'AOMP_CHECK_GIT_BRANCH=0 ' | ||
preinstallopts += 'AOMP_APPLY_ROCM_PATCHES=0 ' | ||
preinstallopts += 'AOMP_STANDALONE_BUILD=1 ' | ||
preinstallopts += 'NUM_THREADS=%(parallel)s ' | ||
preinstallopts += 'AOMP_BUILD_CUDA=0' | ||
|
||
# Core components that must be built for the compiler to work | ||
local_comps = ['roct', 'rocr', 'project', 'libdevice', 'extras', 'openmp', | ||
'pgmath', 'flang', 'flang_runtime', 'comgr', 'rocminfo'] | ||
# Debug components, (optional) | ||
# The debugging compoentes require additonal build dependences: | ||
# texinfo, bison, flex, ncurses, expat, xz, libbabeltrace | ||
# local_comps += ['rocdbgapi', 'rocgdb'] | ||
# Profiling components, (optional) | ||
# local_comps += ['roctracer', 'rocprofiler'] | ||
install_cmd = "./aomp/bin/build_aomp.sh select %s" % ' '.join(local_comps) | ||
|
||
sanity_check_paths = { | ||
'files': ['bin/clang', 'bin/ld.lld', 'include/omp.h'], | ||
'dirs': ['amdgcn', 'bin', 'include', 'lib'], | ||
} | ||
|
||
moduleclass = 'compiler' |
63 changes: 63 additions & 0 deletions
63
easybuild/easyconfigs/a/AOMP/AOMP-11.12-0-gcccuda-2019b.eb
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,63 @@ | ||
easyblock = 'Binary' | ||
|
||
name = 'AOMP' | ||
version = '11.12-0' | ||
|
||
homepage = 'https://github.com/ROCm-Developer-Tools/aomp' | ||
description = "AMD fork of LLVM, setup for OpenMP offloading to Accelerators" | ||
|
||
toolchain = {'name': 'gcccuda', 'version': '2019b'} | ||
|
||
source_urls = ['https://github.com/ROCm-Developer-Tools/aomp/releases/download/rel_%(version)s/'] | ||
sources = ['aomp-%(version)s.tar.gz'] | ||
patches = ['build_aomp-%(version)s.patch'] | ||
checksums = [ | ||
'b01bbed55c4851cfebc69d5be42bafdfb8b43b82e0b6715fe5e1e86db47272e2', # aomp-11.12-0.tar.gz | ||
'c5ccdd00e270731a353461a8f158cdec556689742d501163c8059e739d0588b7', # build_aomp-11.12-0.patch | ||
] | ||
|
||
builddependencies = [ | ||
('CMake', '3.15.3'), | ||
('Perl', '5.30.0'), | ||
('Python', '3.7.4'), | ||
('pkg-config', '0.29.2'), | ||
] | ||
|
||
dependencies = [ | ||
('binutils', '2.32'), | ||
('libffi', '3.2.1'), | ||
('numactl', '2.0.12'), | ||
] | ||
|
||
extract_sources = True | ||
dontcreateinstalldir = True | ||
|
||
preinstallopts = 'AOMP=%(installdir)s ' | ||
preinstallopts += 'AOMP_REPOS="%(builddir)s/aomp%(version_major)s" ' | ||
preinstallopts += 'AOMP_CHECK_GIT_BRANCH=0 ' | ||
preinstallopts += 'AOMP_APPLY_ROCM_PATCHES=0 ' | ||
preinstallopts += 'AOMP_STANDALONE_BUILD=1 ' | ||
preinstallopts += 'NUM_THREADS=%(parallel)s' | ||
# 'NVPTXGPUS' should be set to the NVPTX architecture you would like to | ||
# support, e.g. '50,75'. | ||
preinstallopts += ' AOMP_BUILD_CUDA=1 ' | ||
preinstallopts += 'CUDA="$CUDA_ROOT" ' | ||
preinstallopts += 'NVPTXGPUS="70"' | ||
|
||
# Core components that must be built for the compiler to work | ||
local_comps = ['roct', 'rocr', 'project', 'libdevice', 'extras', 'openmp', | ||
'pgmath', 'flang', 'flang_runtime', 'comgr', 'rocminfo'] | ||
# Debug components, (optional) | ||
# The debugging compoentes require additonal build dependences: | ||
# texinfo, bison, flex, ncurses, expat, xz, libbabeltrace | ||
# local_comps += ['rocdbgapi', 'rocgdb'] | ||
# Profiling components, (optional) | ||
# local_comps += ['roctracer', 'rocprofiler'] | ||
install_cmd = "./aomp/bin/build_aomp.sh select %s" % ' '.join(local_comps) | ||
|
||
sanity_check_paths = { | ||
'files': ['bin/clang', 'bin/ld.lld', 'include/omp.h'], | ||
'dirs': ['amdgcn', 'bin', 'include', 'lib'], | ||
} | ||
|
||
moduleclass = 'compiler' |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,68 @@ | ||
Patch to support EB number of threads, remove hard coded path for GCC and | ||
support EB version of GCC | ||
--- a/aomp/bin/aomp_common_vars | ||
+++ b/aomp/bin/aomp_common_vars | ||
@@ -128,16 +128,16 @@ GFXLIST=${GFXLIST:-"gfx700 gfx701 gfx801 gfx803 gfx900 gfx902 gfx906 gfx908"} | ||
export GFXLIST | ||
|
||
# Calculate the number of threads to use for make | ||
-NUM_THREADS= | ||
if [ ! -z `which "getconf"` ]; then | ||
- NUM_THREADS=$(`which "getconf"` _NPROCESSORS_ONLN) | ||
+ _THREADS=$(`which "getconf"` _NPROCESSORS_ONLN) | ||
if [ "$AOMP_PROC" == "ppc64le" ] ; then | ||
- NUM_THREADS=8 | ||
+ _THREADS=8 | ||
fi | ||
if [ "$AOMP_PROC" == "aarch64" ] ; then | ||
- NUM_THREADS=$(( NUM_THREADS / 4)) | ||
+ _THREADS=$(( _THREADS / 4)) | ||
fi | ||
fi | ||
+NUM_THREADS=${NUM_THREADS:-$_THREADS} | ||
|
||
# These are the web sites where the AOMP git repos are pulled from | ||
GITROC="https://github.com/radeonopencompute" | ||
|
||
--- a/aomp/bin/build_aomp.sh | ||
+++ b/aomp/bin/build_aomp.sh | ||
@@ -28,19 +28,6 @@ thisdir=$(getdname $0) | ||
# --- end standard header ---- | ||
|
||
function build_aomp_component() { | ||
- osversion=$(cat /etc/os-release | grep -e ^VERSION_ID) | ||
- | ||
- if [[ $osversion =~ '"7.' ]]; then | ||
- echo "OS version 7 found `cat /etc/os-release`" | ||
- [ -f /opt/rh/devtoolset-7/enable ] && . /opt/rh/devtoolset-7/enable | ||
- elif [[ $osversion =~ '"8' ]]; then | ||
- echo "OS version 8 found `cat /etc/os-release`" | ||
- echo | ||
- echo "Get updated gcc 8: export PATH=/usr/bin:\$PATH" | ||
- export PATH=/usr/bin:$PATH | ||
- gcc --version | ||
- fi | ||
- | ||
$AOMP_REPOS/$AOMP_REPO_NAME/bin/build_$COMPONENT.sh "$@" | ||
rc=$? | ||
if [ $rc != 0 ] ; then | ||
|
||
--- a/aomp/bin/build_openmp.sh | ||
+++ b/aomp/bin/build_openmp.sh | ||
@@ -88,14 +88,14 @@ function getgcc9orless(){ | ||
_loc=`which gcc` | ||
[ "$_loc" == "" ] && return | ||
gccver=`$_loc --version | grep gcc | cut -d")" -f2 | cut -d"." -f1` | ||
- [ $gccver -gt $GCCMIN ] && _loc=`which gcc-$GCCMIN` | ||
+ [ $gccver -gt $GCCMIN ] | ||
echo $_loc | ||
} | ||
function getgxx9orless(){ | ||
_loc=`which g++` | ||
[ "$_loc" == "" ] && return | ||
gxxver=`$_loc --version | grep g++ | cut -d")" -f2 | cut -d"." -f1` | ||
- [ $gxxver -gt $GCCMIN ] && _loc=`which g++-$GCCMIN` | ||
+ [ $gxxver -gt $GCCMIN ] | ||
echo $_loc | ||
} | ||
|
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think this should be configurable. Does it support 7.0 or sm_70 format? Then you can use the templates cuda_compute_capabilities etc.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Based on the pre-configured targets, I assumed one had to use whole numbers only.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hm, then I'd say an EasyBlock is better here. I'm afraid short of patching the EC (e.g. via hooks) this would limit the GPU usability unusually, especially after we introduced the
--cuda-compute-capabilities
optionThis has the additional benefit that those hard-coded, repeated preinstallopts can be unified and handled better (e.g. detecting presence of CUDA)