Merge pull request #1457 from IntelPython/main

Merge 0.23 into gold/2021

.github/workflows/conda-package.yml

@@ -30,7 +30,7 @@ jobs:
       fail-fast: false
       matrix:
         python: ['3.9', '3.10', '3.11']
-        os: [ubuntu-latest, windows-latest]
+        os: [ubuntu-latest, windows-2019]
 
     runs-on: ${{ matrix.os }}
 

.github/workflows/coverage.yml

@@ -54,7 +54,7 @@ jobs:
       # Ignoring test due to opencl driver optimization bug
       - name: Run tests with coverage
         run: |
-          pytest  -q --cov --cov-report xml --pyargs numba_dpex \
+          pytest  -q --cov=./ --cov-report xml --pyargs numba_dpex \
             -k 'not test_1d_strided_dpnp_array_in_kernel[2]'
 
       - name: Coveralls

.github/workflows/cpp_style_checks.yml

@@ -18,7 +18,7 @@ jobs:
     steps:
     - uses: actions/checkout@v4
     - name: Run clang-format style check for C/C++ programs.
-      uses: jidicula/clang-format-action@v4.11.0
+      uses: jidicula/clang-format-action@v4.13.0
       with:
         clang-format-version: '14'
         check-path: 'numba_dpex/dpctl_iface'

.github/workflows/openssf-scorecard.yml

@@ -40,7 +40,7 @@ jobs:
           persist-credentials: false
 
       - name: "Run analysis"
-        uses: ossf/scorecard-action@0864cf19026789058feabb7e87baa5f140aac736 # v2.3.1
+        uses: ossf/scorecard-action@dc50aa9510b46c811795eb24b2f1ba02a914e534 # v2.3.3
         with:
           results_file: results.sarif
           results_format: sarif
@@ -71,6 +71,6 @@ jobs:
       # Upload the results to GitHub's code scanning dashboard.
       - name: "Upload to code-scanning"
         if: ${{ github.event_name == 'push' }}
-        uses: github/codeql-action/upload-sarif@d39d31e687223d841ef683f52467bd88e9b21c14 # v3.25.3
+        uses: github/codeql-action/upload-sarif@9fdb3e49720b44c48891d036bb502feb25684276 # v3.25.6
         with:
           sarif_file: results.sarif

.gitignore

@@ -26,6 +26,9 @@ _skbuild
 
 docs/source/developer/autogen*
 
+# Ignore versioneer generated files
+numba_dpex/_version.py
+
 # Ignore generated cpp files
 numba_dpex/dpnp_iface/*.cpp
 numba_dpex/dpnp_iface/*.h

CHANGELOG.md

@@ -4,13 +4,21 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
-## [0.23.0] - 2024-04-XX
+## [0.23.0] - 2024-05-28
 
 ### Fixed
 * Array alignment problem for stack arrays allocated for kernel arguments. (#1357)
 * Issue #892, #906 caused by incorrect code generation for indexing (#1377)
-* Fix `KernelHasReturnValueError` inside `KernelDispatcher`. (#1394)
+* Generation of `KernelHasReturnValueError` error inside `KernelDispatcher`. (#1394)
 * Issue #1390: broken support for slicing into `dpctl.tensor.usm_ndarray` in kernels (#1425)
+* Support for Wheels package on Windows (#1430)
+* Incorrect mangled name for kernel function arguments (#1443)
+* Remove artifacts from conda/wheel packages residing in root level (#1450)
+* GDB tests to work properly on Intel Max GPU (#1451)
+* Improper wheels installation on unsupported platforms (#1452)
+* Ref-counting of Python object temporaries in unboxing code (#1454)
+* Segfault caused by using `malloc` to allocate `NRT_MemInfo`. Replaced with Numba's NRT `alloc` (#1458)
+* Incorrect package name in README.md (#1463)
 
 ### Added
 * A new overloaded `dimensions` attribute for all index-space id classes (#1359)
@@ -24,18 +32,24 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 * A `sycl::local_accessor`-like API (`kernel_api.LocalAccessor`) for numba-dpex kernel (#1331)
 * Specialization support for `device_func` decorator (#1398)
 * Support for all `kernel_api` functions inside the `numba_dpex.kernel` decorator. (#1400)
+* Support for dpnp 0.15 (#1434, #1464)
+* Improvements to pyproject.toml configs to build numba-dpex from source. (#1449)
+* Load the `SPV_INTEL_variable_length_array` SPIR-V extension to supporting arrays in private address-space on Intel Max GPU. (#1451)
 
 ### Changed
 * Default inline threshold value set to `2` from `None`. (#1385)
 * Port parfor kernel templates to `kernel_api` (#1416), (#1424)
+* Use `SPIRVKernelDispatcher` for parfor kernel dispatch (#1435, #1448)
+* All examples use the latest dpctl API (#1431)
 * Minimum required dpctl version is now 0.16.1
-* Minimum required numba version is now 0.59.0
+* Minimum required numba version is now 0.59.0 (#1462)
 
 ### Removed
 * OpenCL-like kernel API functions (#1420)
 * `func` decorator (replaced by `device_func`) (#1400)
 * `numba_dpex.experimental.kernel` and `numba_dpex.experimental.device_func` (#1400)
 
+
 ## [0.22.0] - 2024-02-19
 
 ### Fixed

README.md

@@ -12,7 +12,7 @@
 
 
 
-Data-parallel Extension for Numba* (numba-dpex) is an open-source standalone
+Data Parallel Extension for Numba* (numba-dpex) is an open-source standalone
 extension for the [Numba](http://numba.pydata.org) Python JIT compiler.
 Numba-dpex provides a [SYCL*](https://sycl.tech/)-like API for kernel
 programming Python. SYCL* is an open standard developed by the [Unified

conda-recipe/bld.bat

@@ -35,7 +35,13 @@ for /f %%f in ('dir /b /S .\dist') do (
 
 :: wheel file was renamed
 for /f %%f in ('dir /b /S .\dist') do (
-    %PYTHON% -m pip install %%f
+    %PYTHON% -m pip install %%f ^
+      --no-build-isolation ^
+      --no-deps ^
+      --only-binary :all: ^
+      --no-index ^
+      --prefix %PREFIX% ^
+      -vv
     if %ERRORLEVEL% neq 0 exit 1
 )
 

conda-recipe/build.sh

@@ -9,6 +9,9 @@ echo "--gcc-toolchain=${BUILD_PREFIX} --sysroot=${BUILD_PREFIX}/${HOST}/sysroot
 ICPXCFG="$(pwd)/icpx_for_conda.cfg"
 ICXCFG="$(pwd)/icpx_for_conda.cfg"
 
+read -r GLIBC_MAJOR GLIBC_MINOR <<<"$(conda list '^sysroot_linux-64$' \
+    | tail -n 1 | awk '{print $2}' | grep -oP '\d+' | head -n 2 | tr '\n' ' ')"
+
 export ICXCFG
 export ICPXCFG
 
@@ -26,8 +29,15 @@ export PATH=$CONDA_PREFIX/bin-llvm:$PATH
 # -wnx flags mean: --wheel --no-isolation --skip-dependency-check
 ${PYTHON} -m build -w -n -x
 ${PYTHON} -m wheel tags --remove --build "$GIT_DESCRIBE_NUMBER" \
-    --platform-tag manylinux2014_x86_64 dist/numba_dpex*.whl
-${PYTHON} -m pip install dist/numba_dpex*.whl
+    --platform-tag "manylinux_${GLIBC_MAJOR}_${GLIBC_MINOR}_x86_64" \
+    dist/numba_dpex*.whl
+${PYTHON} -m pip install dist/numba_dpex*.whl \
+    --no-build-isolation \
+    --no-deps \
+    --only-binary :all: \
+    --no-index \
+    --prefix "${PREFIX}" \
+    -vv
 
 # Copy wheel package
 if [[ -v WHEELS_OUTPUT_FOLDER ]]; then

conda-recipe/meta.yaml

@@ -3,6 +3,9 @@
 {% set excluded_compiler_version2 = "2024.0.2" %}
 {% set excluded_compiler_version3 = "2024.0.3" %}
 
+{% set pyproject = load_file_data('pyproject.toml') %}
+{% set py_build_deps = pyproject.get('build-system', {}).get('requires', []) %}
+
 package:
     name: numba-dpex
     version: {{ GIT_DESCRIBE_TAG }}
@@ -21,38 +24,39 @@ requirements:
         - {{ compiler('cxx') }}
         - {{ compiler('dpcpp') }} >={{ required_compiler_version }},!={{ excluded_compiler_version1 }},!={{ excluded_compiler_version2 }},!={{ excluded_compiler_version3 }}  # [win]
         - {{ compiler('dpcpp') }} >={{ required_compiler_version }},!={{ excluded_compiler_version1 }},!={{ excluded_compiler_version2 }}  # [linux]
-        # specific version of sysroot required by dpcpp, but 2024.0.0 package
-        # does not have it in meta data
-        - sysroot_linux-64 >=2.28  # [linux]
+        # Minimal supported version of sysroot (which is version of glibc) to
+        # have compatibility with wider range of linux distributions.
+        # 2.28 is the minimal supported version by dpcpp
+        - sysroot_linux-64 =2.28  # [linux]
     host:
+        - python
+        - pip >=24.0
         - dpcpp-cpp-rt >={{ required_compiler_version }},!={{ excluded_compiler_version1 }},!={{ excluded_compiler_version2 }},!={{ excluded_compiler_version3 }}  # [win]
         - dpcpp-cpp-rt >={{ required_compiler_version }},!={{ excluded_compiler_version1 }},!={{ excluded_compiler_version2 }}  # [linux]
-        - python
+        # ensure we are using latest version of setuptools, since we don't need
+        # editable environments for release.
         - setuptools >=69
-        - scikit-build >=0.17
-        - ninja >=1.11.1 # [not win]
-        - cmake >=3.29
-        - numba >=0.59
-        - llvmlite >=0.42.0
-        - dpctl >=0.16.1
-        - dpnp >=0.14
-        - numpy >=1.24
-        # TODO: temporary fix, because IGC is broken for output produced by llvm-spirv 2024.1 on windows
-        - dpcpp-llvm-spirv >={{ required_compiler_version }} # [not win]
-        - dpcpp-llvm-spirv >={{ required_compiler_version }},<2024.1 # [win]
-        - wheel >=0.43
-        - pip >=24.0
-        - python-build >=1.1
-        - versioneer==0.29
+        {% for dep in py_build_deps %}
+        {% if dep.startswith('ninja') %}
+        - {{ dep.split(';')[0] }} # [not win]
+        {% elif dep.startswith('cmake') %}
+        - {{ dep }}
+        {% elif dep.startswith('build>=') %}
+        - {{ 'python-' ~ dep }}
+        {% else %}
+        - {{ dep|replace('_','-') }}
+        {% endif %}
+        {% endfor %}
         # versioneer dependency
         - tomli # [py<311]
+        # While we don't need it for build, but install it here, so we can
+        # pin_compatible at run section.
+        - dpcpp-llvm-spirv >={{ required_compiler_version }}
     run:
         - {{ pin_compatible('dpcpp-cpp-rt', min_pin='x.x', max_pin='x') }}
         - {{ pin_compatible('intel-cmplr-lib-rt', min_pin='x.x', max_pin='x') }}
+        # TODO: pick up min version from dep
         - {{ pin_compatible('dpcpp-llvm-spirv', min_pin='x.x', max_pin='x') }}
-        # TODO: temporary fix, because IGC is broken for output produced by llvm-spirv 2024.1 on windows
-        - {{ pin_compatible('dpcpp-llvm-spirv', min_pin='x.x', max_pin='x') }} # [not win]
-        - {{ pin_compatible('dpcpp-llvm-spirv', min_pin='x.x', max_pin='x', upper_bound='2024.1') }} # [win]
         - {{ pin_compatible('dpnp', min_pin='x.x.x', max_pin='x.x') }}
         - {{ pin_compatible('dpctl', min_pin='x.x.x', max_pin='x.x') }}
         - {{ pin_compatible('numba', min_pin='x.x.x', max_pin='x.x') }}

numba_dpex/__init__.py

@@ -17,7 +17,7 @@
 from numba import __version__ as numba_version
 
 from .kernel_api_impl.spirv import target as spirv_kernel_target
-from .numba_patches import patch_is_ufunc
+from .numba_patches import patch_ufuncs
 from .register_kernel_api_overloads import init_kernel_api_spirv_overloads
 
 
@@ -70,7 +70,7 @@ def parse_sem_version(version_string: str) -> Tuple[int, int, int]:
 dpctl_sem_version = parse_sem_version(dpctl.__version__)
 
 # Monkey patches
-patch_is_ufunc.patch()
+patch_ufuncs.patch()
 
 from numba import prange  # noqa E402