Merge branch 'main' into get-ready-for-64-bit-indptr

Author: Intron7

.coderabbit.yaml

@@ -10,7 +10,7 @@ reviews:
   request_changes_workflow: false
   review_status: false
   auto_review:
-    enabled: true
+    enabled: false
     drafts: false
     base_branches:
       - "^main$"
@@ -23,45 +23,21 @@ reviews:
       enabled: true
   path_instructions:
     - path: "src/rapids_singlecell/_cuda/**"
-      instructions: |
-        These are nanobind CUDA C++ kernels. Pay close attention to:
-        - Shared memory usage and device limits
-        - Thread/block configuration
-        - Memory access patterns and coalescing
-        - Correct use of atomicAdd and synchronization
-        - Template parameter correctness (float vs double)
-        - MANDATORY: Every kernel launch (<<<grid, block, shared, stream>>>) MUST be followed by cudaGetLastError() to catch launch failures. Flag any kernel launch missing this check.
-        - MANDATORY: No magic numbers. All block sizes, tile sizes, grid calculations, and thresholds must use named constants (constexpr int BLOCK_SIZE = 256). Flag any raw numeric literal in dim3, grid, or shared memory calculations.
-    - path: "src/rapids_singlecell/**/_kernels/**"
-      instructions: |
-        These are CuPy RawKernel definitions. Review for:
-        - Correct CUDA kernel launch configurations
-        - Shared memory bounds
-        - Type safety (float32 vs float64 mismatches)
-        - No magic numbers in kernel launch configurations or kernel code. Block sizes, tile sizes, and thresholds must use named constants.
-        - After RawKernel calls, check for cp.cuda.runtime.getLastError() to catch silent launch failures.
+      instructions: "Nanobind CUDA C++ kernels. See knowledge base for detailed review guidelines."
+    - path: "src/rapids_singlecell/**"
+      instructions: "GPU-accelerated Python. See knowledge base for detailed review guidelines."
     - path: "tests/**"
-      instructions: |
-        Do not suggest changing test tolerances without strong justification.
-        Tests run on GPU; some numerical differences vs CPU are expected.
+      instructions: "GPU tests — do not suggest tolerance changes without strong justification."
     - path: "docs/**"
-      instructions: |
-        For documentation changes, focus on:
-        - Accuracy of code examples
-        - Completeness of API documentation
-        - Consistency with current code
+      instructions: "Check accuracy of code examples and consistency with current code."
     - path: ".github/**"
-      instructions: |
-        For CI/workflow changes:
-        - Check for proper error handling
-        - Verify GPU availability checks before tests
-        - Watch for semicolon issues in CMAKE_CUDA_ARCHITECTURES
+      instructions: "CI workflows — watch for semicolon issues in CMAKE_CUDA_ARCHITECTURES."
 knowledge_base:
   opt_out: false
   code_guidelines:
     filePatterns:
-      - "python_agents.md"
-      - "cuda_agents.md"
+      - ".coderabbit/python_agents.md"
+      - ".coderabbit/cuda_agents.md"
       - "docs/contributing.md"
 chat:
-  auto_reply: true
+  auto_reply: false

cuda_agents.md .coderabbit/cuda_agents.mdcuda_agents.md renamed to .coderabbit/cuda_agents.md

@@ -71,7 +71,7 @@ jobs:
     strategy:
       matrix:
         RAPIDS_VER:
-          - "26.02"
+          - "26.04"
         CUDA_SUFFIX:
           - { ver: "12.8.0", label: "cuda12", pkg: "cu12" }
           - { ver: "13.0.2", label: "cuda13", pkg: "cu13" }
@@ -143,6 +143,7 @@ jobs:
           labels: ${{ steps.meta-base.outputs.labels }}
           build-args: |
             CUDA_VER=${{ matrix.CUDA_SUFFIX.ver }}
+            GIT_ID=${{ github.event.pull_request.head.sha || github.sha }}
           #cache-from: type=registry,ref=ghcr.io/${{ github.repository }}-deps
 
       - name: Generate artifact attestation for base image
@@ -171,6 +172,7 @@ jobs:
           labels: ${{ steps.meta.outputs.labels }}
           build-args: |
             CUDA_ARCHS=${{ matrix.CUDA_SUFFIX.label == 'cuda12' && '75-real;80-real;86-real;89-real;90' || '75-real;80-real;86-real;89-real;90-real;100-real;120' }}
+            GIT_ID=${{ github.event.pull_request.head.sha || github.sha }}
           #cache-from: type=registry,ref=ghcr.io/${{ github.repository }}
           build-contexts: |
             rapids-singlecell-deps=docker-image://${{ fromJSON(steps.meta-base.outputs.json).tags[0] }}

python_agents.md .coderabbit/python_agents.mdpython_agents.md renamed to .coderabbit/python_agents.md

@@ -106,6 +106,7 @@ jobs:
         uses: pypa/cibuildwheel@v3.1.4
         env:
           SETUPTOOLS_SCM_PRETEND_VERSION: ${{ steps.version.outputs.version }}
+          CIBW_BUILD: 'cp312-*'
           CIBW_SKIP: '*-musllinux*'
           CIBW_MANYLINUX_X86_64_IMAGE: ${{ matrix.os == 'linux-intel' && matrix.cibw_image || '' }}
           CIBW_MANYLINUX_AARCH64_IMAGE: ${{ matrix.os == 'linux-arm' && matrix.cibw_image || '' }}

.github/workflows/docker.yml

@@ -38,7 +38,7 @@ jobs:
       - id: get-envs
         run: |
           # Stable and dev environments
-          STABLE_DEV_JSON=$(uvx --with "virtualenv<21" hatch env show --json | jq -c '
+          STABLE_DEV_JSON=$(uvx --from "hatch==1.16.5" hatch env show --json | jq -c '
             to_entries
             | map(
               select(.key | startswith("hatch-test") and endswith("12") and (contains("prerelease") | not))
@@ -50,7 +50,7 @@ jobs:
           echo "stable-dev=${STABLE_DEV_JSON}" | tee -a $GITHUB_OUTPUT
 
           # Prerelease environments
-          PRERELEASE_JSON=$(uvx --with "virtualenv<21" hatch env show --json | jq -c '
+          PRERELEASE_JSON=$(uvx --from "hatch==1.16.5" hatch env show --json | jq -c '
             to_entries
             | map(
               select(.key | startswith("hatch-test") and endswith("12") and contains("prerelease"))
@@ -89,18 +89,18 @@ jobs:
           echo "LD_LIBRARY_PATH=/usr/local/cuda/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}" >> $GITHUB_ENV
 
       - name: Install dependencies
-        run: uvx --with "virtualenv<21" hatch -v env create ${{ matrix.env.name }}
+        run: uvx --from "hatch==1.16.5" hatch -v env create ${{ matrix.env.name }}
 
       - name: Run tests
         run: |
           mkdir -p test-data
           if [[ "${{ matrix.env.name }}" == *"stable"* ]]; then
-            uvx --with "virtualenv<21" hatch run ${{ matrix.env.name }}:run-cov -v --color=yes
-            uvx --with "virtualenv<21" hatch run ${{ matrix.env.name }}:coverage xml
-            uvx --with "virtualenv<21" hatch run ${{ matrix.env.name }}:cov-report
+            uvx --from "hatch==1.16.5" hatch run ${{ matrix.env.name }}:run-cov -v --color=yes
+            uvx --from "hatch==1.16.5" hatch run ${{ matrix.env.name }}:coverage xml
+            uvx --from "hatch==1.16.5" hatch run ${{ matrix.env.name }}:cov-report
 
           else
-            uvx --with "virtualenv<21" hatch run ${{ matrix.env.name }}:run -v --color=yes
+            uvx --from "hatch==1.16.5" hatch run ${{ matrix.env.name }}:run -v --color=yes
           fi
 
       - name: Upload test results
@@ -157,10 +157,10 @@ jobs:
         run: nvidia-smi
 
       - name: Install dependencies
-        run: uvx --with "virtualenv<21" hatch -v env create ${{ matrix.env.name }}
+        run: uvx --from "hatch==1.16.5" hatch -v env create ${{ matrix.env.name }}
 
       - name: Run tests
-        run: uvx --with "virtualenv<21" hatch run ${{ matrix.env.name }}:run -v --color=yes
+        run: uvx --from "hatch==1.16.5" hatch run ${{ matrix.env.name }}:run -v --color=yes
 
   remove-label:
     name: Remove 'run-gpu-ci' Label

.github/workflows/publish.yml

@@ -4,6 +4,7 @@ __pycache__/
 /*cache/
 .ipynb_checkpoints/
 /data/
+/benchmarks/
 test-data/
 .vscode/
 
@@ -46,6 +47,7 @@ coverage.xml
 .vscode/
 .cursor/
 .claude/
+.codex
 CLAUDE.md
 
 # tmp_scripts

.github/workflows/test-gpu.yml

@@ -1,6 +1,6 @@
 repos:
 -   repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.15.6
+    rev: v0.15.12
     hooks:
     -   id: ruff-check
         args: ["--fix"]
@@ -37,7 +37,7 @@ repos:
               docs/references.bib|
           )
 -   repo: https://github.com/pre-commit/mirrors-clang-format
-    rev: v22.1.1
+    rev: v22.1.5
     hooks:
     -   id: clang-format
         args: [--style=file, -i]

.gitignore

@@ -16,6 +16,5 @@ build:
       - asdf global uv latest
     build:
       html:
-        # pin virtualenv<21 to work around https://github.com/pypa/hatch/issues/2193
-        - CMAKE_ARGS="-DRSC_BUILD_EXTENSIONS=OFF" uvx --with "virtualenv<21" hatch run docs:build
+        - CMAKE_ARGS="-DRSC_BUILD_EXTENSIONS=OFF" uvx --from "hatch==1.16.5" hatch run docs:build
         - mv docs/_build $READTHEDOCS_OUTPUT

.pre-commit-config.yaml

@@ -71,6 +71,8 @@ if (RSC_BUILD_EXTENSIONS)
   add_nb_cuda_module(_qc_dask_cuda      src/rapids_singlecell/_cuda/qc_dask/qc_kernels_dask.cu)
   add_nb_cuda_module(_bbknn_cuda        src/rapids_singlecell/_cuda/bbknn/bbknn.cu)
   add_nb_cuda_module(_norm_cuda         src/rapids_singlecell/_cuda/norm/norm.cu)
+  add_nb_cuda_module(_gmm_cuda          src/rapids_singlecell/_cuda/gmm/gmm.cu)
+  target_link_libraries(_gmm_cuda PRIVATE CUDA::cublas)
   add_nb_cuda_module(_pr_cuda           src/rapids_singlecell/_cuda/pr/pr.cu)
   add_nb_cuda_module(_nn_descent_cuda   src/rapids_singlecell/_cuda/nn_descent/nn_descent.cu)
   add_nb_cuda_module(_aucell_cuda       src/rapids_singlecell/_cuda/aucell/aucell.cu)