Skip to content
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

Jump to bottom

Rebuild for CUDA 12 #4

Merged
merged 13 commits into from
Jun 3, 2023
Merged

Rebuild for CUDA 12 #4

Show file tree
Hide file tree
Changes from 4 commits
Commits
Show all changes
13 commits
Select commit Hold shift + click to select a range
406967e
Rebuild for CUDA 12
regro-cf-autotick-bot May 31, 2023
8d6fba0
MNT: Re-rendered with conda-build 3.25.0, conda-smithy 3.23.1, and co…
regro-cf-autotick-bot May 31, 2023
e40de23
MNT: Re-rendered with conda-build 3.25.0, conda-smithy 3.23.1, and co…
May 31, 2023
4a0941e
Adjust CUDAARCHS
carterbox May 31, 2023
e25d608
Use new cudatoolkit packages for CUDA 12
carterbox May 31, 2023
e53ec19
Try redirecting CUDA_CFLAGS to CUDAFLAGS
carterbox Jun 1, 2023
d99db8d
Use the export keyword
carterbox Jun 1, 2023
f584313
Increase number of target CUDA archs
carterbox Jun 1, 2023
c654424
REF: Use cross-compilation for alternative archs
carterbox Jun 1, 2023
401d4e3
Add more archs
carterbox Jun 2, 2023
dbd5782
Reduce to 7 archs for time
carterbox Jun 2, 2023
b99a4cc
Skip PowerPC
carterbox Jun 2, 2023
d16af7f
MNT: Re-rendered with conda-build 3.24.0, conda-smithy 3.23.1, and co…
carterbox Jun 2, 2023
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
11 changes: 8 additions & 3 deletions .azure-pipelines/azure-pipelines-linux.yml
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

6 changes: 3 additions & 3 deletions .azure-pipelines/azure-pipelines-win.yml
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

1 change: 1 addition & 0 deletions ...er_version11.2cxx_compiler_version10.yaml → ...er_version11.2cxx_compiler_version10.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,6 +27,7 @@ target_platform:
zip_keys:
- - c_compiler_version
- cxx_compiler_version
- cuda_compiler
- cuda_compiler_version
- cdt_name
- docker_image
33 changes: 33 additions & 0 deletions ...piler_version12cuda_compilercuda-nvcccuda_compiler_version12.0cxx_compiler_version12.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,33 @@
c_compiler:
- gcc
c_compiler_version:
- '12'
cdt_name:
- cos7
channel_sources:
- conda-forge
channel_targets:
- conda-forge main
cuda_compiler:
- cuda-nvcc
cuda_compiler_version:
- '12.0'
cxx_compiler:
- gxx
cxx_compiler_version:
- '12'
docker_image:
- quay.io/condaforge/linux-anvil-cos7-x86_64
carterbox marked this conversation as resolved.
Show resolved Hide resolved
libblas:
- 3.9 *netlib
liblapack:
- 3.9 *netlib
target_platform:
- linux-64
zip_keys:
- - c_compiler_version
- cxx_compiler_version
- cuda_compiler
- cuda_compiler_version
- cdt_name
- docker_image
1 change: 1 addition & 0 deletions ...er_version11.2cxx_compiler_version10.yaml → ...er_version11.2cxx_compiler_version10.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,7 @@ target_platform:
zip_keys:
- - c_compiler_version
- cxx_compiler_version
- cuda_compiler
- cuda_compiler_version
- cdt_name
- docker_image
72 changes: 72 additions & 0 deletions .ci_support/migrations/cuda120.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,72 @@
migrator_ts: 1682985063
__migrator:
kind:
version
migration_number:
1
build_number:
1
paused: false
override_cbc_keys:
- cuda_compiler_stub
operation: key_add
check_solvable: false
primary_key: cuda_compiler_version
ordering:
cxx_compiler_version:
- 9
- 8
- 7
c_compiler_version:
- 9
- 8
- 7
fortran_compiler_version:
- 9
- 8
- 7
docker_image:
- quay.io/condaforge/linux-anvil-comp7 # [os.environ.get("BUILD_PLATFORM") == "linux-64"]
- quay.io/condaforge/linux-anvil-aarch64 # [os.environ.get("BUILD_PLATFORM") == "linux-aarch64"]
- quay.io/condaforge/linux-anvil-ppc64le # [os.environ.get("BUILD_PLATFORM") == "linux-ppc64le"]
- quay.io/condaforge/linux-anvil-armv7l # [os.environ.get("BUILD_PLATFORM") == "linux-armv7l"]
- quay.io/condaforge/linux-anvil-cuda:9.2 # [linux64 and os.environ.get("BUILD_PLATFORM") == "linux-64"]
- quay.io/condaforge/linux-anvil-cuda:10.0 # [linux64 and os.environ.get("BUILD_PLATFORM") == "linux-64"]
- quay.io/condaforge/linux-anvil-cuda:10.1 # [linux64 and os.environ.get("BUILD_PLATFORM") == "linux-64"]
- quay.io/condaforge/linux-anvil-cuda:10.2 # [linux64 and os.environ.get("BUILD_PLATFORM") == "linux-64"]
- quay.io/condaforge/linux-anvil-cuda:11.0 # [linux64 and os.environ.get("BUILD_PLATFORM") == "linux-64"]
- quay.io/condaforge/linux-anvil-cuda:11.1 # [linux64 and os.environ.get("BUILD_PLATFORM") == "linux-64"]
- quay.io/condaforge/linux-anvil-cuda:11.2 # [linux64 and os.environ.get("BUILD_PLATFORM") == "linux-64"]
- quay.io/condaforge/linux-anvil-cos7-x86_64 # [linux64 and os.environ.get("BUILD_PLATFORM") == "linux-64"]
cuda_compiler_version:
- None
- 10.2 # [(linux64 or win) and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]
- 11.0 # [(linux64 or win) and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]
- 11.1 # [(linux64 or win) and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]
- 11.2 # [(linux64 or win) and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]
- 12.0 # [(linux64 or win) and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]
commit_message: "Rebuild for CUDA 12"

cuda_compiler: # [(linux64 or win) and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]
- cuda-nvcc # [(linux64 or win) and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]

cuda_compiler_version: # [(linux64 or win) and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]
- 12.0 # [(linux64 or win) and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]

c_compiler_version: # [linux64 and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]
- 12 # [linux64 and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]

cxx_compiler_version: # [linux64 and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]
- 12 # [linux64 and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]

fortran_compiler_version: # [linux64 and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]
- 12 # [linux64 and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]

cudnn: # [(linux64 or win) and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]
- 8 # [(linux64 or win) and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]

cdt_name: # [linux64 and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]
- cos7 # [linux64 and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]

docker_image: # [os.environ.get("BUILD_PLATFORM", "").startswith("linux-") and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]
- quay.io/condaforge/linux-anvil-cos7-x86_64 # [linux64 and os.environ.get("BUILD_PLATFORM") == "linux-64" and os.environ.get("CF_CUDA_ENABLED", "False") == "True"]
3 changes: 3 additions & 0 deletions ...ort/win_64_cuda_compiler_version11.2.yaml → ...ompilernvcccuda_compiler_version11.2.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,3 +16,6 @@ liblapack:
- 3.9 *netlib
target_platform:
- win-64
zip_keys:
- - cuda_compiler
- cuda_compiler_version
2 changes: 1 addition & 1 deletion .travis.yml
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

19 changes: 13 additions & 6 deletions README.md
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

14 changes: 10 additions & 4 deletions recipe/build.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,13 +4,19 @@ set -exv
make generate --jobs ${CPU_COUNT}

# Duplicate lists because of https://bitbucket.org/icl/magma/pull-requests/32
export CUDA_ARCH_LIST="sm_35,sm_50,sm_60,sm_70,sm_80"
export CUDAARCHS="35-virtual;50-virtual;60-virtual;70-virtual;80-virtual"
export CUDA_ARCH_LIST="sm_35,sm_50,sm_60,sm_70,sm_75,sm_80"
export CUDAARCHS="35-virtual;50-virtual;60-virtual;70-virtual;75-virtual;80-virtual"
carterbox marked this conversation as resolved.
Show resolved Hide resolved

# Only build the lowest non-deprecated arch to minimize build time
if [[ "$target_platform" == "linux-ppc64le" || "$target_platform" == "linux-aarch64" ]]; then
export CUDA_ARCH_LIST="sm_60"
export CUDAARCHS="60-virtual"
export CUDA_ARCH_LIST="sm_60;sm_80"
export CUDAARCHS="60-virtual;80-virtual"
Copy link
Member

Choose a reason for hiding this comment

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

Suggested change
export CUDA_ARCH_LIST="sm_60;sm_80"
export CUDAARCHS="60-virtual;80-virtual"
export CUDA_ARCH_LIST="sm_60"
export CUDAARCHS="60-virtual"

Still only enough time for one arch on Travis.

Copy link
Member

Choose a reason for hiding this comment

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

We could try cross-compiling if that is helpful

Copy link
Member

Choose a reason for hiding this comment

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

Does CUDA 11 support cross-compile? If not, I need to wait for the second migrator for arm/ppc.

Copy link
Member

Choose a reason for hiding this comment

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

Yep it does. Several packages are already building for CUDA 11 using cross-compilation. For example nccl uses cross-compilation

It also will be supported for CUDA 12 (once work on the other archs wraps up)

fi

# Only build the lowest non-deprecated arch to minimize build time
if [[ "$cuda_compiler_version" == "12.0" ]]; then
export CUDA_ARCH_LIST="sm_50,sm_60,sm_70,sm_75,sm_80,sm_89"
export CUDAARCHS="50-virtual;60-virtual;70-virtual;75-virtual;80-virtual;89-virtual"
carterbox marked this conversation as resolved.
Show resolved Hide resolved
fi

# Remove CXX standard flags added by conda-forge. std=c++11 is required to
Expand Down
2 changes: 1 addition & 1 deletion recipe/meta.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -96,7 +96,7 @@ source:
- optional-install.patch

build:
number: 1
number: 2
skip: true # [cuda_compiler_version=="None"]
skip: true # [cuda_compiler_version == "10.2"]
skip: true # [cuda_compiler_version == "11.0"]
Expand Down