Update CuPy and NumPy dependency specs by bdice · Pull Request #22962 · NVIDIA/cudf

bdice · 2026-06-24T05:08:47Z

Contributes to rapidsai/build-planning#279

RAPIDS is taking on requirements cupy>=14.0.1,!=14.1.0 and numpy>=2.0. Wheel dependencies on cupy-cuda12x and cupy-cuda13x now use the [ctk] extra.

See the linked issue for details.

copy-pr-bot · 2026-06-24T05:08:51Z

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

coderabbitai · 2026-06-24T06:01:01Z

Note

Reviews paused

It looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the reviews.auto_review.auto_pause_after_reviewed_commits setting.

Use the following commands to manage reviews:

@coderabbitai resume to resume automatic reviews.
@coderabbitai review to trigger a single review.

Use the checkboxes below for quick actions:

▶️ Resume reviews
🔍 Trigger review

📝 Walkthrough

Walkthrough

Updates CuPy and NumPy version constraints across dependency matrices, Python package manifests, conda recipes, and generated CUDA environment files. Some package and recipe headers also change to include “NVIDIA CORPORATION & AFFILIATES”.

Changes

CuPy 14 and NumPy 2 Dependency Bump

Layer / File(s)	Summary
Canonical dependency constraints in `dependencies.yaml` `dependencies.yaml`	`numpy_run` is raised to `>=2.0,<3.0`; `depends_on_cupy` raises the shared CuPy constraint to `>=14.0.1,!=14.1.0` and switches CUDA wheel selectors to `[ctk]` variants; test matrices bump the oldest CuPy pin, update NumPy pins for Python tests, and adjust `test_cudf_pandas_compat` NumPy and CUDA version entries.
Python package dependency manifests `python/cudf/pyproject.toml`, `python/dask_cudf/pyproject.toml`, `python/cudf_streaming/pyproject.toml`, `python/pylibcudf/pyproject.toml`	CuPy requirements are updated to the `[ctk]` selector where applicable and NumPy requirements are raised to `>=2.0` while preserving upper bounds. SPDX copyright headers are updated in the files that show header changes.
Conda recipes and generated environment files `conda/recipes/cudf/recipe.yaml`, `conda/recipes/pylibcudf/recipe.yaml`, `conda/environments/all_cuda-129_arch-.yaml`, `conda/environments/all_cuda-133_arch-.yaml`	CuPy and NumPy constraints are updated in the recipe runtime constraints and all four generated CUDA environment YAML files. Recipe copyright headers are updated to the newer NVIDIA text.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Possibly related PRs

rapidsai/cudf#22702: Modifies the same CuPy version constraints, including the !=14.1.0 exclusion, in the same dependency files.
rapidsai/cudf#22189: Adjusts dependencies.yaml compatibility/test dependency pins in the same matrix area used here.

Suggested labels

conda, ci

Suggested reviewers

jameslamb
mroeschke

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly summarizes the main change: updating CuPy and NumPy dependency specs.
Description check	✅ Passed	The description matches the changeset by describing the CuPy/NumPy requirement updates and [ctk] wheel dependency changes.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

Create PR with unit tests

_{Comment @coderabbitai help to get the list of available commands.}

coderabbitai

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@dependencies.yaml`:
- Around line 1060-1073: The test CuPy wheel pins in the dependency matrices
still use plain cupy-cuda12x and cupy-cuda13x, which is inconsistent with the
canonical depends_on_cupy entries that require the [ctk] selector. Update the
affected matrix package entries in dependencies.yaml to use the same [ctk]
contract as depends_on_cupy so test and production dependency resolution stay
aligned.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

Push a commit to this branch (recommended)
Create a new PR with the fixes

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 526ee758-902a-45ff-baa5-e39c036693d7

📥 Commits

Reviewing files that changed from the base of the PR and between 20a16c7 and 26fb661.

📒 Files selected for processing (1)

dependencies.yaml

mroeschke

(Reminding myself that are some CI/code cleanups available once this merged)

jameslamb

Changes look right to me, though there are what look like real failures in the oldest-deps jobs:

Attribute "dtype" are different
[left]:  int64
[right]: float64
FAILED tests/dataframe/test_np_ufuncs.py::test_ufunc_dataframe[ceil-False-False] - AssertionError: Attributes of DataFrame.iloc[:, 0] (column name="foo") are different

Attribute "dtype" are different
[left]:  int64
[right]: float64
FAILED tests/dataframe/test_np_ufuncs.py::test_ufunc_dataframe[floor-True-False] - AssertionError: Attributes of DataFrame.iloc[:, 0] (column name="foo") are different

(build link)

gforsyth · 2026-06-30T16:22:24Z

Ok, so previously, oldest deps was pulling in numpy==1.26.4 and latest deps was pulling in numpy==2.4.6

It seems like numpy==2.0.* only fail a few tests because of dtype inconsistencies.

I don't think this is a new failure, so much as one that hasn't shown up yet because we haven't tested on 2.0.*.

What if we bump the lower bound of numpy to 2.1.*?

I can reproduce this locally if numpy=2.0.* is installed.

On other PRs without the dependency changes proposed here, the oldest deps job passes because it's on numpy 1.26: https://github.com/rapidsai/cudf/actions/runs/28449642641/job/84316150615

numpy 2.1.0 is when numpy dropped support for Python 3.9, but that should be fine, since we've already dropped 3.9

`cupy>=14` and `numpy>=2.1` have unified behavior around casting integers and booleans to floats (they don't do it). This leaves us with a condition where numpy>2,<2.1 where the two libraries mismatch on their behavior. These get xfailed.

gforsyth · 2026-06-30T17:39:00Z

+            and parse(np.__version__) >= parse("2.0")
+            and parse(np.__version__) < parse("2.1"),


Now that we are using cupy>=14, which has unified casting behavior with numpy>=2.1, we only need to xfail these tests when numpy>=2.0,<2.1

gforsyth · 2026-06-30T17:39:06Z

+            and parse(np.__version__) >= parse("2.0")
+            and parse(np.__version__) < parse("2.1"),


Now that we are using cupy>=14, which has unified casting behavior with numpy>=2.1, we only need to xfail these tests when numpy>=2.0,<2.1

gforsyth · 2026-06-30T19:12:37Z

/merge

Follow up to #22962 Authors: - Matthew Roeschke (https://github.com/mroeschke) Approvers: - Bradley Dice (https://github.com/bdice) URL: #23053

bdice added non-breaking Non-breaking change improvement Improvement / enhancement to an existing function labels Jun 24, 2026

github-actions Bot assigned bdice Jun 24, 2026

github-actions Bot added Python Affects Python cuDF API. pylibcudf Issues specific to the pylibcudf package labels Jun 24, 2026

github-project-automation Bot added this to cuDF Python Jun 24, 2026

GPUtester moved this to In Progress in cuDF Python Jun 24, 2026

bdice mentioned this pull request Jun 24, 2026

feat(deps): raise cupy lower bound to >=14 rapidsai/build-planning#279

Closed

bdice force-pushed the cupy-pin branch from 67f9758 to c95d2bc Compare June 24, 2026 05:27

bdice changed the title ~~Update CuPy lower bound to 14.0.1~~ Update CuPy and NumPy lower bounds Jun 24, 2026

Update CuPy and NumPy dependency specs.

20a16c7

bdice force-pushed the cupy-pin branch from c95d2bc to 20a16c7 Compare June 24, 2026 05:44

bdice changed the title ~~Update CuPy and NumPy lower bounds~~ Update CuPy and NumPy dependency specs Jun 24, 2026

bdice marked this pull request as ready for review June 24, 2026 05:54

bdice requested a review from a team as a code owner June 24, 2026 05:54

bdice requested a review from gforsyth June 24, 2026 05:54

Fix CuPy constraints for wheel tests

26fb661

coderabbitai Bot reviewed Jun 25, 2026

View reviewed changes

Comment thread dependencies.yaml

Merge branch 'main' into cupy-pin

c73003d

KyleFromNVIDIA reviewed Jun 25, 2026

View reviewed changes

Comment thread dependencies.yaml

KyleFromNVIDIA approved these changes Jun 25, 2026

View reviewed changes

mroeschke approved these changes Jun 25, 2026

View reviewed changes

Guard CuPy CTK extra for wheel metadata

00b1c56

jameslamb approved these changes Jun 26, 2026

View reviewed changes

Reuse CuPy dependency anchors

a908e2d

This was referenced Jun 30, 2026

Update CuPy and NumPy dependency specs rapidsai/rmm#2454

Merged

Update CuPy and NumPy dependency specs rapidsai/kvikio#989

Merged

Merge branch 'main' into cupy-pin

006a495

github-actions Bot assigned gforsyth Jun 30, 2026

gforsyth removed their assignment Jun 30, 2026

gforsyth requested a review from a team as a code owner June 30, 2026 17:37

gforsyth requested review from galipremsagar and wence- June 30, 2026 17:37

github-actions Bot assigned gforsyth Jun 30, 2026

gforsyth reviewed Jun 30, 2026

View reviewed changes

mroeschke approved these changes Jun 30, 2026

View reviewed changes

mroeschke mentioned this pull request Jun 30, 2026

[MLIR] MaskedType: core type, constructor, accessors #22884

Merged

rapids-bot Bot merged commit 0f28e6d into NVIDIA:main Jun 30, 2026
138 checks passed

github-project-automation Bot moved this from In Progress to Done in cuDF Python Jun 30, 2026

mroeschke mentioned this pull request Jun 30, 2026

cudf-classic and CI cleanups for NumPy/CuPy 2/14 version bump #23053

Merged

3 tasks

yinqingh mentioned this pull request Jul 9, 2026

[BUG] cudf_udf CI fails: conda cannot solve cudf=26.08 environment due to cupy/cuda-pathfinder conflict NVIDIA/cudf-spark#15247

Closed

		and parse(np.__version__) >= parse("2.0")
		and parse(np.__version__) < parse("2.1"),

Uh oh!

Conversation

bdice commented Jun 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

copy-pr-bot Bot commented Jun 24, 2026

Uh oh!

coderabbitai Bot commented Jun 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Reviews paused

Walkthrough

Changes

Estimated code review effort

Possibly related PRs

Suggested labels

Suggested reviewers

Uh oh!

coderabbitai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

mroeschke left a comment

Choose a reason for hiding this comment

Uh oh!

jameslamb left a comment

Choose a reason for hiding this comment

Uh oh!

gforsyth commented Jun 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

gforsyth Jun 30, 2026

Choose a reason for hiding this comment

Uh oh!

gforsyth Jun 30, 2026

Choose a reason for hiding this comment

Uh oh!

gforsyth commented Jun 30, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

bdice commented Jun 24, 2026 •

edited

Loading

coderabbitai Bot commented Jun 24, 2026 •

edited

Loading

gforsyth commented Jun 30, 2026 •

edited

Loading