-
-
Notifications
You must be signed in to change notification settings - Fork 13.2k
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
ginkgo-hpc: init at 1.6.0 #261155
base: master
Are you sure you want to change the base?
ginkgo-hpc: init at 1.6.0 #261155
Conversation
Thanks for working on this.
|
|
Gladly taking the maintainer role here. Thanks for your work. |
4793e9f
to
8c5ee98
Compare
That should fix most of what you asked, @markuskowa. |
fa7dd6b
to
4695722
Compare
I have been preparing openmpi-5.0 (https://github.com/markuskowa/nixpkgs/tree/openmpi-5.0), so once it is released, were are actually ready to update.
The additions of rocm to ucc, ucx, and openmpi could be its own PR first, for example. |
Looking through your commits, I think I see why pmix works on your branch. It's at 4.0, which apparently is good enough for openmpi 5.0. I still see no reason to get rid of pmix_5, but that's good to know.
This is indev right now anyway, so sure alright. |
Keeping the commits here until those two PRs are merged so I can continue to build and test. |
Move pmix to pmix_3, and make pmix_3 the default
I'm using this package ( |
cuda = finalAttrs.finalPackage.overrideAttrs (callPackage ./overlays/cuda.nix { }); | ||
hip = finalAttrs.finalPackage.overrideAttrs (callPackage ./overlays/hip.nix { inherit rocmUnfree; }); | ||
dpc = finalAttrs.finalPackage.overrideAttrs (callPackage ./overlays/dpc.nix { }); | ||
doc = finalAttrs.finalPackage.overrideAttrs (callPackage ./overlays/doc.nix { }); | ||
test = finalAttrs.finalPackage.overrideAttrs (callPackage ./overlays/test.nix { }); | ||
example = finalAttrs.finalPackage.overrideAttrs (callPackage ./overlays/example.nix { }); | ||
benchmark = finalAttrs.finalPackage.overrideAttrs (callPackage ./overlays/benchmark.nix { }); |
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.
the derivations in passthru
won't be spliced because
#211340
so having one of them in nativeBuildInputs
when cross-compiling won't pick the on build version
nix-repl> pkgsCross.aarch64-multiplatform.__splicedPackages.ginkgo-hpc.doc ? __spliced
false
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.
So, unless I'm reading this completely wrong...
- I need to make a scope like https://github.com/NixOS/nixpkgs/blob/master/pkgs/top-level/cuda-packages.nix
- I need to use
makeScopeWithSplicing
somehow?
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.
If those two options won't work, would just setting it to x86-64-linux
for now suffice?
I don't really understand this at all.
hip = finalAttrs.finalPackage.overrideAttrs (callPackage ./overlays/hip.nix { inherit rocmUnfree; }); | ||
dpc = finalAttrs.finalPackage.overrideAttrs (callPackage ./overlays/dpc.nix { }); | ||
doc = finalAttrs.finalPackage.overrideAttrs (callPackage ./overlays/doc.nix { }); | ||
test = finalAttrs.finalPackage.overrideAttrs (callPackage ./overlays/test.nix { }); |
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.
tests should be in passthru.tests
attrset
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.
The test
sub-derivation just builds the tests, we have to use impureTests
for actually running them.
This is because said tests require a GPU.
Description of changes
Tracking: #197885
For: #216655
@keldu I added you to the maintainers for
ginkgo-hpc
since you expressed interest in making a PR for this yourself.Things done
sandbox = true
set innix.conf
? (See Nix manual)nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD"
. Note: all changes have to be committed, also see nixpkgs-review usage./result/bin/
)