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.

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

refactor CUDA versions in dependencies.yaml #517

Merged
merged 3 commits into from
Jan 11, 2024
Merged
Changes from all commits
Commits
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: 39 additions & 21 deletions dependencies.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,8 @@ files:
output: none
includes:
- build
- cudatoolkit
- cuda
- cuda_version
- docs
- py_version
- test
Expand All @@ -17,7 +18,7 @@ files:
docs:
output: none
includes:
- cudatoolkit
- cuda_version
- docs
channels:
- rapidsai
Expand All @@ -35,6 +36,30 @@ dependencies:
- cxx-compiler
- make
specific:
- output_types: conda
matrices:
- matrix:
cuda: "11.[245]"
Copy link
Contributor

Choose a reason for hiding this comment

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

@jameslamb What do you think of this syntax? This is valid glob syntax and should work since it's handled by fnmatch.fnmatch. Just wanted to point it out explicitly to you since it's different than what we've been writing elsewhere.

Copy link
Member Author

Choose a reason for hiding this comment

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

I thought this could lead to surprises if there were ever double-digit releases (11.10+), but I guess that isn't true and [] matches a single character.

On Python 3.10.12:

import fnmatch

fnmatch.fnmatch("11.12", "11.[1245]")
# False

fnmatch.fnmatch("11.12", "11.[1245][12]")
# True

fnmatch.fnmatch("11.12", "11.[245][12]")
# False

Given that, I support this and think it's a nice simplification. Thanks for the @!

packages:
- gcc<11.0.0
- matrix:
cuda: "11.[68]"
packages:
- gcc<12.0.0
- matrix:
cuda: "12.*"
packages:
- gcc<13.0.0
- output_types: conda
matrices:
- matrix:
arch: x86_64
packages:
- sysroot_linux-64==2.17
Copy link
Contributor

Choose a reason for hiding this comment

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

It looks like we were requiring sysroot_linux-64 on ARM builds before this PR. I fixed that.

Copy link
Member Author

Choose a reason for hiding this comment

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

makes sense to me! This makes it clearer that the package name is architecture-dependent.

- matrix:
arch: aarch64
packages:
- sysroot_linux-aarch64==2.17
- output_types: conda
matrices:
- matrix:
Expand Down Expand Up @@ -80,54 +105,47 @@ dependencies:
- matrix:
cuda: "12.0"
packages:
- cuda-version=12.0
- cuda-nvcc
cudatoolkit:
cuda_version:
specific:
- output_types: conda
matrices:
- matrix:
cuda: "11.2"
packages:
- cuda-version=11.2
- cudatoolkit
- gcc<11.0.0
- sysroot_linux-64==2.17
- matrix:
cuda: "11.4"
packages:
- cuda-version=11.4
- cudatoolkit
- gcc<11.0.0
- sysroot_linux-64==2.17
- matrix:
cuda: "11.5"
packages:
- cuda-version=11.5
- cudatoolkit
- gcc<11.0.0
- sysroot_linux-64==2.17
- matrix:
cuda: "11.6"
packages:
- cuda-version=11.6
- cudatoolkit
- gcc<12.0.0
- sysroot_linux-64==2.17
- matrix:
cuda: "11.8"
packages:
- cuda-version=11.8
- cudatoolkit
- gcc<12.0.0
- sysroot_linux-64==2.17
- matrix:
cuda: "12.0"
packages:
- cuda-version=12.0
cuda:
specific:
- output_types: conda
matrices:
- matrix:
cuda: "11.*"
packages:
- cudatoolkit
- matrix:
cuda: "12.*"
packages:
- cuda-cupti-dev
Copy link
Contributor

Choose a reason for hiding this comment

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

We can explore dropping this as a follow-up PR, due to #504.

- gcc<13.0.0
- sysroot_linux-64==2.17
docs:
common:
- output_types: [conda]
Expand Down