Merge remote-tracking branch 'upstream/main' into guilhermeleobas/exc…

…eption-handling

.github/workflows/stale.yml

@@ -13,14 +13,25 @@ jobs:
       pull-requests: write  # for actions/stale to close stale PRs
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/stale@v3
+      - uses: actions/stale@v7
         with:
+          # issues
           stale-issue-message: >
             This issue is marked as stale as it has had no activity in the past
             30 days. Please close this issue if no further response or action is
             needed. Otherwise, please respond with any updates and confirm that
             this issue still needs to be addressed.
           stale-issue-label: 'stale'
-          any-of-labels: 'question,needtriage,more info needed'
+          any-of-issue-labels: 'question,needtriage,more info needed'
           days-before-issue-stale: 30
           days-before-issue-close: 7
+          # pull requests
+          stale-pr-message: >
+            This pull request is marked as stale as it has had no activity in
+            the past 3 months. Please respond to this comment if you're still
+            interested in working on this. Many thanks!
+          days-before-pr-stale: 90  # 3 months
+          days-before-pr-close: 7
+          any-of-pr-labels: '2 - In progress,4 - Waiting on author'
+          stale-pr-label: 'stale'
+          close-pr-label: 'abandoned - stale'

azure-pipelines.yml

@@ -12,14 +12,14 @@ jobs:
     name: macOS
     vmImage: macos-11
     matrix:
-      py38_np118:
+      py38_np121:
         PYTHON: '3.8'
-        NUMPY: '1.18'
+        NUMPY: '1.21'
         CONDA_ENV: 'azure_ci'
         TEST_START_INDEX: 0
-      py310_np123:
+      py310_np124:
         PYTHON: '3.10'
-        NUMPY: '1.23'
+        NUMPY: '1.24'
         CONDA_ENV: 'azure_ci'
         TEST_THREADING: 'tbb'
         TEST_START_INDEX: 1
@@ -29,89 +29,89 @@ jobs:
     name: Linux
     vmImage: ubuntu-20.04
     matrix:
-      py38_np118_vanilla:
+      py38_np121_vanilla:
         PYTHON: '3.8'
-        NUMPY: '1.18'
+        NUMPY: '1.21'
         CONDA_ENV: azure_ci
         VANILLA_INSTALL: yes
         TEST_START_INDEX: 2
-      py38_np118_cov:
+      py38_np121_cov:
         PYTHON: '3.8'
-        NUMPY: '1.18'
+        NUMPY: '1.21'
         CONDA_ENV: azure_ci
         RUN_COVERAGE: yes
         RUN_FLAKE8: yes
         RUN_MYPY: yes
         TEST_START_INDEX: 3
-      py38_np119_tbb:
+      py38_np122_tbb:
         PYTHON: '3.8'
-        NUMPY: '1.19.2=*_0'
+        NUMPY: '1.22'
         CONDA_ENV: azure_ci
         TEST_THREADING: 'tbb'
         TEST_START_INDEX: 4
-      py38_np119_omp:
+      py38_np122_omp:
         PYTHON: '3.8'
-        NUMPY: '1.19.2=*_0'
+        NUMPY: '1.22'
         CONDA_ENV: azure_ci
         TEST_THREADING: omp
         TEST_START_INDEX: 5
-      py38_np119_workqueue:
+      py38_np122_workqueue:
         PYTHON: '3.8'
-        NUMPY: '1.19.2=*_0'
+        NUMPY: '1.22'
         CONDA_ENV: azure_ci
         TEST_THREADING: workqueue
         TEST_START_INDEX: 6
-      py38_np120_doc:
+      py38_np123_doc:
         PYTHON: '3.8'
-        NUMPY: '1.20'
+        NUMPY: '1.23'
         CONDA_ENV: azure_ci
         BUILD_DOC: yes
         TEST_START_INDEX: 7
-      py38_np120:
+      py38_np122:
         PYTHON: '3.8'
-        NUMPY: '1.20'
+        NUMPY: '1.22'
         CONDA_ENV: azure_ci
         TEST_START_INDEX: 8
-      py38_np120_svml:
+      py38_np123_svml:
         PYTHON: '3.8'
-        NUMPY: '1.20'
+        NUMPY: '1.23'
         CONDA_ENV: azure_ci
         TEST_SVML: yes
         TEST_START_INDEX: 9
-      py38_np122:
+      py38_np124:
         PYTHON: '3.8'
-        NUMPY: '1.22'
+        NUMPY: '1.24'
         CONDA_ENV: azure_ci
         TEST_START_INDEX: 10
-      py39_np119:
+      py39_np123:
         PYTHON: '3.9'
-        NUMPY: '1.19.2=*_0'
+        NUMPY: '1.23'
         CONDA_ENV: azure_ci
         TEST_START_INDEX: 11
-      py39_np120_typeguard:
+      py39_np123_typeguard:
         PYTHON: '3.9'
-        NUMPY: '1.20'
+        NUMPY: '1.23'
         CONDA_ENV: azure_ci
         RUN_TYPEGUARD: yes
         TEST_START_INDEX: 12
-      py39_np121:
+      py39_np122:
         PYTHON: '3.9'
-        NUMPY: '1.21'
+        NUMPY: '1.22'
         CONDA_ENV: azure_ci
         TEST_START_INDEX: 13
-      py39_np123:
+      py39_np124:
         PYTHON: '3.9'
-        NUMPY: '1.23'
+        NUMPY: '1.24'
         CONDA_ENV: azure_ci
         TEST_START_INDEX: 14
-      py310_np121:
+      py310_np123:
         PYTHON: '3.10'
-        NUMPY: '1.21'
+        NUMPY: '1.23'
         CONDA_ENV: azure_ci
         TEST_START_INDEX: 15
-      py310_np123:
+      py310_np124:
         PYTHON: '3.10'
-        NUMPY: '1.23'
+        NUMPY: '1.24'
         CONDA_ENV: azure_ci
         TEST_START_INDEX: 16
 

buildscripts/azure/azure-windows.yml

@@ -8,14 +8,14 @@ jobs:
     vmImage: ${{ parameters.vmImage }}
   strategy:
     matrix:
-      py310_np123:
+      py310_np124:
         PYTHON: '3.10'
-        NUMPY: '1.23'
+        NUMPY: '1.24'
         CONDA_ENV: 'testenv'
         TEST_START_INDEX: 17
-      py38_np118:
+      py38_np121:
         PYTHON: '3.8'
-        NUMPY: '1.18'
+        NUMPY: '1.21'
         CONDA_ENV: 'testenv'
         TEST_START_INDEX: 18
 

buildscripts/gpuci/axis.yaml

@@ -9,6 +9,7 @@ CUDA_TOOLKIT_VER:
 - "11.1"
 - "11.2"
 - "11.5"
+- "11.8"
 
 LINUX_VER:
 - ubuntu18.04

buildscripts/gpuci/build.sh

@@ -24,12 +24,29 @@ else
   export NUMBA_CUDA_USE_NVIDIA_BINDING=0;
 fi;
 
+# Test with Minor Version Compatibility on CUDA 11.8
+if [ $CUDA_TOOLKIT_VER == "11.8" ]
+then
+  export NUMBA_CUDA_ENABLE_MINOR_VERSION_COMPATIBILITY=1;
+else
+  export NUMBA_CUDA_ENABLE_MINOR_VERSION_COMPATIBILITY=0;
+fi;
+
 # Test with different NumPy versions with each toolkit (it's not worth testing
 # the Cartesian product of versions here, we just need to test with different
 # CUDA and NumPy versions).
-declare -A CTK_NUMPY_VMAP=( ["11.0"]="1.19" ["11.1"]="1.21" ["11.2"]="1.22" ["11.5"]="1.23")
+declare -A CTK_NUMPY_VMAP=( ["11.0"]="1.21" ["11.1"]="1.22" ["11.2"]="1.23" ["11.5"]="1.24" ["11.8"]="1.24")
 NUMPY_VER="${CTK_NUMPY_VMAP[$CUDA_TOOLKIT_VER]}"
 
+# Use conda-forge for NumPy 1.24 - at the time of writing it is not available
+# on the defaults channel.
+
+if [ "${NUMPY}" == "1.24" ]; then
+  NUMPY_CHANNEL_PKG=conda-forge::numpy
+else
+  NUMPY_CHANNEL_PKG=numpy
+fi
+
 
 ################################################################################
 # SETUP - Check environment
@@ -46,8 +63,10 @@ gpuci_logger "Create testing env"
 gpuci_mamba_retry create -n numba_ci -y \
                   "python=${PYTHON_VER}" \
                   "cudatoolkit=${CUDA_TOOLKIT_VER}" \
+                  "rapidsai::cubinlinker" \
+                  "conda-forge::ptxcompiler" \
                   "numba/label/dev::llvmlite" \
-                  "numpy=${NUMPY_VER}" \
+                  "${NUMPY_CHANNEL_PKG}=${NUMPY_VER}" \
                   "scipy" \
                   "cffi" \
                   "psutil" \

buildscripts/incremental/setup_conda_environment.cmd

@@ -14,14 +14,18 @@ cmd /C conda info
 set CONDA_INSTALL=cmd /C conda install -q -y
 set PIP_INSTALL=pip install -q
 
+@rem Use conda-forge for NumPy 1.24 - at the time of writing it is not available
+@rem on the defaults channel.
+if %NUMPY%==1.24 (set NUMPY_CHANNEL_PKG="conda-forge::numpy") else (set NUMPY_CHANNEL_PKG="numpy")
+
 @echo on
 
 @rem Deactivate any environment
 call deactivate
 @rem Display root environment (for debugging)
 conda list
 @rem Scipy, CFFI, jinja2 and IPython are optional dependencies, but exercised in the test suite
-conda create -n %CONDA_ENV% -q -y python=%PYTHON% numpy=%NUMPY% cffi pip scipy jinja2 ipython gitpython pyyaml
+conda create -n %CONDA_ENV% -q -y python=%PYTHON% %NUMPY_CHANNEL_PKG%=%NUMPY% cffi pip scipy jinja2 ipython gitpython pyyaml
 
 call activate %CONDA_ENV%
 @rem Install latest llvmlite build

buildscripts/incremental/setup_conda_environment.sh

@@ -27,13 +27,23 @@ source deactivate
 # Display root environment (for debugging)
 conda list
 
+# Use conda-forge for NumPy 1.24 - at the time of writing it is not available
+# on the defaults channel.
+
+if [ "${NUMPY}" == "1.24" ]; then
+  NUMPY_CHANNEL_PKG=conda-forge::numpy
+else
+  NUMPY_CHANNEL_PKG=numpy
+fi
+
 # If VANILLA_INSTALL is yes, then only Python, NumPy and pip are installed, this
 # is to catch tests/code paths that require an optional package and are not
 # guarding against the possibility that it does not exist in the environment.
 # Create a base env first and then add to it...
 # NOTE: gitpython is needed for CI testing to do the test slicing
 # NOTE: pyyaml is used to ensure that the Azure CI config is valid
-conda create -n $CONDA_ENV -q -y ${EXTRA_CHANNELS} python=$PYTHON numpy=$NUMPY pip gitpython pyyaml
+
+conda create -n $CONDA_ENV -q -y ${EXTRA_CHANNELS} python=$PYTHON $NUMPY_CHANNEL_PKG=$NUMPY pip gitpython pyyaml
 
 # Activate first
 set +v

docs/source/cuda-reference/kernel.rst

@@ -59,8 +59,8 @@ creating a specialized instance:
 .. autoclass:: numba.cuda.dispatcher.CUDADispatcher
    :members: inspect_asm, inspect_llvm, inspect_sass, inspect_types,
              get_regs_per_thread, specialize, specialized, extensions, forall,
-             get_shared_mem_per_block, get_const_mem_size,
-             get_local_mem_per_thread
+             get_shared_mem_per_block, get_max_threads_per_block,
+             get_const_mem_size, get_local_mem_per_thread
 
 
 

docs/source/cuda/index.rst

@@ -26,4 +26,5 @@ Numba for CUDA GPUs
    bindings.rst
    cuda_ffi.rst
    caching.rst
+   minor_version_compatibility.rst
    faq.rst

docs/source/cuda/minor_version_compatibility.rst

@@ -0,0 +1,64 @@
+.. _minor-version-compatibility:
+
+CUDA Minor Version Compatiblity
+===============================
+
+CUDA `Minor Version Compatibility
+<https://docs.nvidia.com/deploy/cuda-compatibility/index.html#minor-version-compatibility>`_
+(MVC) enables the use of a newer CUDA toolkit version than the CUDA version
+supported by the driver, provided that the toolkit and driver both have the same
+major version. For example, use of CUDA toolkit 11.5 with CUDA driver 450 (CUDA
+version 11.0) is supported through MVC.
+
+Numba supports MVC for CUDA 11 on Linux using the external ``cubinlinker`` and
+``ptxcompiler`` packages, subject to the following limitations:
+
+- Linking of archives is unsupported.
+- Cooperative Groups are unsupported, because they require an archive to be
+  linked.
+
+MVC is not supported on Windows.
+
+
+Enabling MVC Support
+--------------------
+
+To use MVC support, the ``cubinlinker`` and ``ptxcompiler`` compiler packages
+must be installed from the appropriate channels. To install using conda, use:
+
+.. code:: bash
+
+   conda install rapidsai::cubinlinker conda-forge::ptxcompiler
+
+To install with pip, use the NVIDIA package index:
+
+.. code:: bash
+
+   pip install ptxcompiler-cu11 cubinlinker-cu11 --extra-index-url=https://pypi.ngc.nvidia.com
+
+MVC support is enabled by setting the environment variable:
+
+.. code:: bash
+
+   export NUMBA_CUDA_ENABLE_MINOR_VERSION_COMPATIBILITY=1
+
+
+or by setting a configuration variable prior to using any CUDA functionality in
+Numba:
+
+.. code:: python
+
+   from numba import config
+   config.CUDA_ENABLE_MINOR_VERSION_COMPATIBILITY = True
+
+
+References
+----------
+
+Further information about Minor Version Compatibility may be found in:
+
+- The `CUDA Compatibility Guide
+  <https://docs.nvidia.com/deploy/cuda-compatibility/index.html>`_.
+- The `README for ptxcompiler
+  <https://github.com/rapidsai/ptxcompiler/blob/main/README.md>`_.
+

docs/source/cuda/overview.rst

@@ -55,8 +55,8 @@ Software
 --------
 
 Numba aims to support CUDA Toolkit versions released within the last 3 years.
-An NVIDIA driver sufficient for the toolkit version is also required.
-Presently:
+An NVIDIA driver sufficient for the toolkit version is also required (see also
+:ref:`minor-version-compatibility`). Presently:
 
 * 11.0 is the minimum required toolkit version.
 * 11.2 or later is recommended, as it uses an NVVM version based on LLVM 7 (as