restore conda-python-tests on CUDA 13 (#395)

Closes #296 

Restores CUDA 13 conda test CI jobs, now that there are conda-forge PyTorch packages with CUDA 13 support (conda-forge/pytorch-cpu-feedstock#477)

Also modifies `pytorch` conda dependency to meet these requirements:

* `cugraph-pyg` must be installable on a system without a GPU
* `cugraph-pyg`'s tests require CUDA-enabled builds of PyTorch

With the following mix of things:

* add a `require_gpu` matrix filter in `dependencies.yaml` which pulls in `pytorch-gpu` opted-into in test CI jobs but otherwise not
  - *`conda-forge::pytorch-gpu` is a metapackage that forces the installation of CUDA variants of `conda-forge::pytorch`... that should replace the "accidentally pulled in a CPU-only variant" case with a loud, clear conda solver error*
* depend on `mkl` in the test x86_64 environment but without version constraints
  - *allow `pytorch` to declare its range of compatible `mkl` versions*
  - *this still prevents OpenBLAS variants from getting installed, which I think was part of the goal of #161*
  - *keeping this out of `cugraph-pyg`'s dependencies still makes it possible to install alongside `nomkl`, even though that combination is untested*
* add comments in the `cugraph-pyg` conda recipe explaining why it doesn't depend on `pytorch-gpu`

I hope this will be a relatively future-proof way to guarantee CI here keeps picking up the PyTorch versions this project wants to tet against.

Authors:
  - James Lamb (https://github.com/jameslamb)

Approvers:
  - Alex Barghi (https://github.com/alexbarghi-nv)
  - Bradley Dice (https://github.com/bdice)

URL: #395

Author: jameslamb

.github/workflows/pr.yaml

@@ -184,9 +184,7 @@ jobs:
     if: fromJSON(needs.changed-files.outputs.changed_file_groups).test_python_conda
     with:
       build_type: pull-request
-      # TODO: remove the CUDA 13 exclusion here once there are pytorch CUDA 13 packages
-      #  ref: https://github.com/pytorch/pytorch/issues/159779
-      matrix_filter: map(select(.ARCH == "amd64" and (.CUDA_VER | startswith("12"))))
+      matrix_filter: map(select(.ARCH == "amd64"))
       script: ci/test_python.sh
       sccache-dist-token-secret-name: GIST_REPO_READ_ORG_GITHUB_TOKEN
   docs-build:

.github/workflows/test.yaml

@@ -42,9 +42,7 @@ jobs:
       date: ${{ inputs.date }}
       script: ci/test_python.sh
       sha: ${{ inputs.sha }}
-      # TODO: remove the CUDA 13 exclusion here once there are pytorch CUDA 13 packages
-      #  ref: https://github.com/pytorch/pytorch/issues/159779
-      matrix_filter: map(select(.ARCH == "amd64" and (.CUDA_VER | startswith("12"))))
+      matrix_filter: map(select(.ARCH == "amd64"))
       sccache-dist-token-secret-name: GIST_REPO_READ_ORG_GITHUB_TOKEN
   wheel-tests-pylibwholegraph:
     secrets: inherit

ci/test_python.sh

@@ -1,5 +1,5 @@
 #!/bin/bash
-# SPDX-FileCopyrightText: Copyright (c) 2022-2025, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2022-2026, NVIDIA CORPORATION.
 # SPDX-License-Identifier: Apache-2.0
 
 set -euo pipefail
@@ -38,7 +38,7 @@ if [[ "${RUNNER_ARCH}" != "ARM64" ]]; then
   rapids-dependency-file-generator \
     --output conda \
     --file-key test_cugraph_pyg \
-    --matrix "cuda=${RAPIDS_CUDA_VERSION%.*};arch=$(arch);py=${RAPIDS_PY_VERSION}"  \
+    --matrix "cuda=${RAPIDS_CUDA_VERSION%.*};arch=$(arch);py=${RAPIDS_PY_VERSION};require_gpu=true" \
     --prepend-channel "${CPP_CHANNEL}" \
     --prepend-channel "${PYTHON_CHANNEL}" \
   | tee env.yaml
@@ -76,7 +76,7 @@ if [[ "${RUNNER_ARCH}" != "ARM64" ]]; then
   rapids-dependency-file-generator \
     --output conda \
     --file-key test_pylibwholegraph \
-    --matrix "cuda=${RAPIDS_CUDA_VERSION%.*};arch=$(arch);py=${RAPIDS_PY_VERSION}"  \
+    --matrix "cuda=${RAPIDS_CUDA_VERSION%.*};arch=$(arch);py=${RAPIDS_PY_VERSION};require_gpu=true" \
     --prepend-channel "${CPP_CHANNEL}" \
     --prepend-channel "${PYTHON_CHANNEL}" \
   | tee env.yaml

conda/recipes/cugraph-pyg/recipe.yaml

@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2025, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2025-2026, NVIDIA CORPORATION.
 # SPDX-License-Identifier: Apache-2.0
 schema_version: 1
 
@@ -37,8 +37,11 @@ requirements:
     - pylibcugraph =${{ minor_version }}
     - pylibwholegraph =${{ minor_version }}
     - python
+    # This is intentionally spelled 'pytorch' (not 'pytorch-gpu' and not using build string selectors)
+    # because we want it to be possible to at least install `cugraph-pyg` in an environment without a GPU,
+    # to support use cases like building container images.
     - pytorch >=2.3
-    - pytorch_geometric >=2.5,<2.7
+    - pytorch_geometric >=2.5,<2.8
 
 tests:
   - python:

dependencies.yaml

@@ -64,7 +64,6 @@ files:
       - depends_on_sentence_transformers
       - py_version
       - test_python_common
-      - depends_on_mkl
       - depends_on_pylibwholegraph
       - depends_on_cugraph_pyg
   test_pylibwholegraph:
@@ -75,7 +74,6 @@ files:
       - depends_on_pytorch
       - py_version
       - test_python_common
-      - depends_on_mkl
       - depends_on_pylibwholegraph
       - test_python_pylibwholegraph
   py_build_libwholegraph:
@@ -376,6 +374,16 @@ dependencies:
           - matrix:
             packages:
               - *pytorch_pip
+      - output_types: [conda]
+        matrices:
+          # Prevent fallback to CPU-only pytorch when we want a CUDA variant.
+          - matrix:
+              require_gpu: "true"
+            packages:
+              - pytorch-gpu
+          # Default to falling back to whatever 'pytorch' is pulled in via cugraph-pyg's dependencies.
+          - matrix:
+            packages:
   depends_on_nccl:
     common:
       - output_types: conda
@@ -648,8 +656,3 @@ dependencies:
       - output_types: conda
         packages:
           - cugraph-pyg==26.4.*,>=0.0.0a0
-  depends_on_mkl:
-    common:
-      - output_types: conda
-        packages:
-          - mkl<2024.1.0