Thanks to visit codestin.com
Credit goes to github.com

Skip to content

[v13] BUG: Hot fix for numpy 2 support in some fusion paths#9012

Merged
asi1024 merged 1 commit into
cupy:v13from
seberg:np2-fuse-scalar
Mar 5, 2025
Merged

[v13] BUG: Hot fix for numpy 2 support in some fusion paths#9012
asi1024 merged 1 commit into
cupy:v13from
seberg:np2-fuse-scalar

Conversation

@seberg

@seberg seberg commented Mar 4, 2025

Copy link
Copy Markdown
Member

Scalars are a problem, and I am not convinced that this is a complete fix (for example w.r.t. integers). OTOH v14 should already be working here, I believe.

But I think this is safe and could be back-portable.


Closes gh-8536

Need to still check if v14 is indeed fine, we may want to forward port the test to make sure. v14 is fine as expected.

Scalars are a problem, and I am not convinced that this is a complete
fix (for example w.r.t. integers).  OTOH v14 should already be working
here I believe.

But I think this is safe and could be back-portable.
@seberg seberg force-pushed the np2-fuse-scalar branch from 855faa9 to 19c914e Compare March 4, 2025 13:33
@jakirkham

Copy link
Copy Markdown
Member

/test mini

@kmaehashi kmaehashi added cat:bug Bugs blocking Issue/pull-request is mandatory for the upcoming release prio:high labels Mar 5, 2025
@kmaehashi kmaehashi changed the title BUG: Hot fix for numpy 2 support in some fusion paths [v13] BUG: Hot fix for numpy 2 support in some fusion paths Mar 5, 2025
@kmaehashi kmaehashi added the to-be-backported Pull-requests to be backported to stable branch label Mar 5, 2025
@kmaehashi kmaehashi added this to the v13.5.0 milestone Mar 5, 2025
@asi1024

asi1024 commented Mar 5, 2025

Copy link
Copy Markdown
Member

Test failures are unrelated. I confirmed that 7 test cases have been resolved in this PR. Thank you for the pull request!

@asi1024 asi1024 removed the to-be-backported Pull-requests to be backported to stable branch label Mar 5, 2025
@asi1024 asi1024 merged commit c41319e into cupy:v13 Mar 5, 2025
@kmaehashi

Copy link
Copy Markdown
Member

This is to-be-backported to v13.4.1 branch.

@kmaehashi kmaehashi added the to-be-backported Pull-requests to be backported to stable branch label Mar 5, 2025
kmaehashi pushed a commit to kmaehashi/cupy that referenced this pull request Mar 5, 2025
[v13] BUG: Hot fix for numpy 2 support in some fusion paths
@github-actions

Copy link
Copy Markdown

@asi1024 This pull-request is marked as to-be-backported, but the corresponding backport PR could not be found. Could you check?

@seberg seberg deleted the np2-fuse-scalar branch March 13, 2025 08:23
@github-actions

Copy link
Copy Markdown

@asi1024 This pull-request is marked as to-be-backported, but the corresponding backport PR could not be found. Could you check?

@kmaehashi kmaehashi removed the to-be-backported Pull-requests to be backported to stable branch label Mar 21, 2025
@kmaehashi

Copy link
Copy Markdown
Member

This is backported to v13.4.1 in #9016.
(The bot doesn't handle the irregular hotfix case 😅)

rapids-bot Bot pushed a commit to rapidsai/cucim that referenced this pull request Mar 26, 2025
…els (#856)

Fix for recent test failures after CuPy 13.4.1 was released as seen here:
https://github.com/rapidsai/cucim/actions/runs/14052308871

There is a data type casting change in fused kernels used by `cucim.skimage.segmentation.chan_vese`. Likely this is related to the changes made in cupy/cupy#9012. We can ensure the correct output dtype from the fused kernel by making sure all float scalars used as input to the fused kernels are device scalars of the desired precision.

Authors:
  - Gregory Lee (https://github.com/grlee77)

Approvers:
  - https://github.com/jakirkham

URL: #856
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

blocking Issue/pull-request is mandatory for the upcoming release cat:bug Bugs prio:high

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants