ENH: Adopt new macOS Accelerate BLAS/LAPACK Interfaces, including ILP…

…64 (numpy#24053) macOS 13.3 shipped with an updated Accelerate framework that provides BLAS / LAPACK. The new version is aligned with Netlib's v3.9.1 and also supports ILP64. The changes here adopt those new interfaces when available. - New interfaces are used when ACCELERATE_NEW_LAPACK is defined. - ILP64 interfaces are used when both ACCELERATE_NEW_LAPACK and ACCELERATE_LAPACK_ILP64 are defined. macOS 13.3 now ships with 3 different sets of BLAS / LAPACK interfaces: - LP64 / LAPACK v3.2.1 - legacy interfaces kept for compatibility - LP64 / LAPACK v3.9.1 - new interfaces - ILP64 / LAPACK v3.9.1 - new interfaces with ILP64 support For LP64, we want to support building against macOS 13.3+ SDK, but having it work on pre-13.3 systems. To that end, we created wrappers for each API that do a runtime check on which set of API is available and should be used. However, these were deemed potentially too complex to include during review of numpygh-24053, and left out in this commit. Please see numpygh-24053 for those. ILP64 is only supported on macOS 13.3+ and does not use additional wrappers. We've included support for both distutils and Meson builds. All tests pass on Apple silicon and Intel based Macs. A new CI job for Accelerate ILP64 on x86-64 was added as well. Benchmarks ILP64 Accelerate vs OpenBLAS before after ratio [73f0cf4f] [d1572653] <openblas-ilp64> <accelerate-ilp64> n/a n/a n/a bench_linalg.Linalg.time_op('det', 'float16') n/a n/a n/a bench_linalg.Linalg.time_op('pinv', 'float16') n/a n/a n/a bench_linalg.Linalg.time_op('svd', 'float16') failed failed n/a bench_linalg.LinalgSmallArrays.time_det_small_array + 3.96±0.1μs 5.04±0.4μs 1.27 bench_linalg.Linalg.time_op('norm', 'float32') 1.43±0.04ms 1.43±0ms 1.00 bench_linalg.Einsum.time_einsum_outer(<class 'numpy.float32'>) 12.7±0.4μs 12.7±0.3μs 1.00 bench_linalg.Einsum.time_einsum_sum_mul2(<class 'numpy.float32'>) 24.1±0.8μs 24.1±0.04μs 1.00 bench_linalg.Linalg.time_op('norm', 'float16') 9.48±0.2ms 9.48±0.3ms 1.00 bench_linalg.Einsum.time_einsum_outer(<class 'numpy.float64'>) 609±20μs 609±2μs 1.00 bench_linalg.Einsum.time_einsum_noncon_outer(<class 'numpy.float32'>) 64.9±2μs 64.7±0.07μs 1.00 bench_linalg.Einsum.time_einsum_contig_outstride0(<class 'numpy.float64'>) 1.24±0.03ms 1.24±0.01ms 1.00 bench_linalg.Einsum.time_einsum_noncon_outer(<class 'numpy.float64'>) 102±3μs 102±0.2μs 1.00 bench_linalg.Einsum.time_einsum_contig_contig(<class 'numpy.float64'>) 21.9±0.8μs 21.8±0.02μs 1.00 bench_linalg.Einsum.time_einsum_multiply(<class 'numpy.float64'>) 22.8±0.2ms 22.7±0.3ms 0.99 bench_linalg.Eindot.time_einsum_ijk_jil_kl 13.3±0.4μs 13.3±0.02μs 0.99 bench_linalg.Einsum.time_einsum_sum_mul2(<class 'numpy.float64'>) 9.56±0.3μs 9.49±0.2μs 0.99 bench_linalg.Einsum.time_einsum_noncon_contig_contig(<class 'numpy.float64'>) 7.31±0.2μs 7.26±0.08μs 0.99 bench_linalg.Einsum.time_einsum_noncon_contig_outstride0(<class 'numpy.float32'>) 5.60±0.2ms 5.55±0.02ms 0.99 bench_linalg.Eindot.time_einsum_ij_jk_a_b 37.1±1μs 36.7±0.1μs 0.99 bench_linalg.Einsum.time_einsum_contig_outstride0(<class 'numpy.float32'>) 13.5±0.4μs 13.4±0.05μs 0.99 bench_linalg.Einsum.time_einsum_sum_mul(<class 'numpy.float64'>) 1.03±0.03μs 1.02±0μs 0.99 bench_linalg.LinalgSmallArrays.time_norm_small_array 51.6±2μs 51.0±0.09μs 0.99 bench_linalg.Einsum.time_einsum_contig_contig(<class 'numpy.float32'>) 15.2±0.5μs 15.0±0.04μs 0.99 bench_linalg.Einsum.time_einsum_noncon_sum_mul2(<class 'numpy.float64'>) 13.9±0.4μs 13.7±0.02μs 0.99 bench_linalg.Einsum.time_einsum_noncon_sum_mul2(<class 'numpy.float32'>) 415±10μs 409±0.4μs 0.99 bench_linalg.Eindot.time_einsum_i_ij_j 9.29±0.3μs 9.01±0.03μs 0.97 bench_linalg.Einsum.time_einsum_noncon_mul(<class 'numpy.float64'>) 18.2±0.6μs 17.6±0.04μs 0.97 bench_linalg.Einsum.time_einsum_multiply(<class 'numpy.float32'>) 509±40μs 492±10μs 0.97 bench_linalg.Einsum.time_einsum_mul(<class 'numpy.float64'>) 9.63±0.3μs 9.28±0.09μs 0.96 bench_linalg.Einsum.time_einsum_noncon_contig_contig(<class 'numpy.float32'>) 9.08±0.2μs 8.73±0.02μs 0.96 bench_linalg.Einsum.time_einsum_noncon_mul(<class 'numpy.float32'>) 15.6±0.5μs 15.0±0.04μs 0.96 bench_linalg.Einsum.time_einsum_noncon_sum_mul(<class 'numpy.float64'>) 7.74±0.2μs 7.39±0.04μs 0.95 bench_linalg.Einsum.time_einsum_noncon_contig_outstride0(<class 'numpy.float64'>) 18.6±0.6μs 17.7±0.03μs 0.95 bench_linalg.Einsum.time_einsum_noncon_multiply(<class 'numpy.float32'>) 14.5±0.4μs 13.7±0.03μs 0.95 bench_linalg.Einsum.time_einsum_noncon_sum_mul(<class 'numpy.float32'>) 13.3±0.6μs 12.5±0.3μs 0.94 bench_linalg.Einsum.time_einsum_sum_mul(<class 'numpy.float32'>) 23.5±0.5μs 21.9±0.05μs 0.93 bench_linalg.Einsum.time_einsum_noncon_multiply(<class 'numpy.float64'>) 264±20μs 243±4μs 0.92 bench_linalg.Einsum.time_einsum_mul(<class 'numpy.float32'>) - 177±50μs 132±0.6μs 0.75 bench_linalg.Eindot.time_dot_trans_at_a - 10.7±0.3μs 7.13±0.01μs 0.67 bench_linalg.Linalg.time_op('norm', 'int16') - 97.5±2μs 64.7±0.1μs 0.66 bench_linalg.Eindot.time_matmul_trans_a_at - 8.87±0.3μs 5.76±0μs 0.65 bench_linalg.Linalg.time_op('norm', 'longfloat') - 8.90±0.3μs 5.77±0.01μs 0.65 bench_linalg.Linalg.time_op('norm', 'float64') - 8.48±0.3μs 5.40±0.01μs 0.64 bench_linalg.Linalg.time_op('norm', 'int64') - 106±2μs 66.5±8μs 0.63 bench_linalg.Eindot.time_inner_trans_a_a - 8.25±0.3μs 5.16±0μs 0.62 bench_linalg.Linalg.time_op('norm', 'int32') - 103±5ms 64.6±0.5ms 0.62 bench_import.Import.time_linalg - 106±3μs 66.0±0.1μs 0.62 bench_linalg.Eindot.time_dot_trans_a_at - 202±20μs 124±0.6μs 0.61 bench_linalg.Eindot.time_matmul_trans_at_a - 31.5±10μs 19.3±0.02μs 0.61 bench_linalg.Eindot.time_dot_d_dot_b_c - 32.4±20μs 19.7±0.03μs 0.61 bench_linalg.Eindot.time_matmul_d_matmul_b_c - 5.05±1ms 3.06±0.09ms 0.61 bench_linalg.Linalg.time_op('svd', 'complex128') - 5.35±0.9ms 3.09±0.09ms 0.58 bench_linalg.Linalg.time_op('svd', 'complex64') - 6.37±3ms 3.27±0.1ms 0.51 bench_linalg.Linalg.time_op('pinv', 'complex128') - 7.26±8ms 3.24±0.1ms 0.45 bench_linalg.Linalg.time_op('pinv', 'complex64') - 519±100μs 219±0.8μs 0.42 bench_linalg.Linalg.time_op('det', 'complex64') - 31.3±0.9μs 12.8±0.1μs 0.41 bench_linalg.Linalg.time_op('norm', 'complex128') - 2.44±0.7ms 924±1μs 0.38 bench_linalg.Linalg.time_op('pinv', 'float64') - 29.9±0.8μs 10.8±0.01μs 0.36 bench_linalg.Linalg.time_op('norm', 'complex64') - 2.56±0.5ms 924±1μs 0.36 bench_linalg.Linalg.time_op('pinv', 'float32') - 2.63±0.5ms 924±0.6μs 0.35 bench_linalg.Linalg.time_op('pinv', 'int64') - 2.68±0.7ms 927±10μs 0.35 bench_linalg.Linalg.time_op('pinv', 'int32') - 2.68±0.5ms 927±10μs 0.35 bench_linalg.Linalg.time_op('pinv', 'int16') - 2.93±0.6ms 925±2μs 0.32 bench_linalg.Linalg.time_op('pinv', 'longfloat') - 809±500μs 215±0.2μs 0.27 bench_linalg.Linalg.time_op('det', 'complex128') - 3.67±0.9ms 895±20μs 0.24 bench_linalg.Eindot.time_tensordot_a_b_axes_1_0_0_1 - 489±100μs 114±20μs 0.23 bench_linalg.Eindot.time_inner_trans_a_ac - 3.64±0.7ms 777±0.3μs 0.21 bench_linalg.Lstsq.time_numpy_linalg_lstsq_a__b_float64 - 755±90μs 157±10μs 0.21 bench_linalg.Eindot.time_dot_a_b - 4.63±1ms 899±9μs 0.19 bench_linalg.Linalg.time_op('svd', 'longfloat') - 5.19±1ms 922±10μs 0.18 bench_linalg.Linalg.time_op('svd', 'float64') - 599±200μs 89.4±2μs 0.15 bench_linalg.Eindot.time_matmul_trans_atc_a - 956±200μs 140±10μs 0.15 bench_linalg.Eindot.time_matmul_a_b - 6.45±3ms 903±10μs 0.14 bench_linalg.Linalg.time_op('svd', 'float32') - 6.42±3ms 896±0.7μs 0.14 bench_linalg.Linalg.time_op('svd', 'int32') - 6.47±4ms 902±5μs 0.14 bench_linalg.Linalg.time_op('svd', 'int64') - 6.52±1ms 899±2μs 0.14 bench_linalg.Linalg.time_op('svd', 'int16') - 799±300μs 109±2μs 0.14 bench_linalg.Eindot.time_dot_trans_atc_a - 502±100μs 65.0±0.2μs 0.13 bench_linalg.Eindot.time_dot_trans_a_atc - 542±300μs 64.2±0.05μs 0.12 bench_linalg.Eindot.time_matmul_trans_a_atc - 458±300μs 41.6±0.09μs 0.09 bench_linalg.Linalg.time_op('det', 'int32') - 471±100μs 41.9±0.03μs 0.09 bench_linalg.Linalg.time_op('det', 'float32') - 510±100μs 43.6±0.06μs 0.09 bench_linalg.Linalg.time_op('det', 'int16') - 478±200μs 39.6±0.05μs 0.08 bench_linalg.Linalg.time_op('det', 'longfloat') - 599±200μs 39.6±0.09μs 0.07 bench_linalg.Linalg.time_op('det', 'float64') - 758±300μs 41.6±0.1μs 0.05 bench_linalg.Linalg.time_op('det', 'int64') Co-authored-by: Ralf Gommers <ralf.gommers@gmail.com>
charris · Sep 1, 2023 · fdbed5e · fdbed5e
1 parent 4fb4d7a
commit fdbed5e
Show file tree

Hide file tree

Showing 9 changed files with 303 additions and 14 deletions.
diff --git a/.cirrus.star b/.cirrus.star
@@ -48,4 +48,7 @@ def main(ctx):
     if wheel:
         return fs.read("tools/ci/cirrus_wheels.yml")
 
-    return fs.read("tools/ci/cirrus_macosx_arm64.yml")
+    if int(pr_number) < 0:
+        return []
+
+    return fs.read("tools/ci/cirrus_arm.yml")
diff --git a/.github/workflows/macos.yml b/.github/workflows/macos.yml
@@ -0,0 +1,135 @@
+name: macOS tests (meson)
+
+on:
+  pull_request:
+    branches:
+      - main
+      - maintenance/**
+
+permissions:
+   contents: read  # to fetch code (actions/checkout)
+
+env:
+  CCACHE_DIR: "${{ github.workspace }}/.ccache"
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
+  cancel-in-progress: true
+
+jobs:
+  x86_conda:
+    name: macOS x86-64 conda
+    if: "github.repository == 'numpy/numpy'"
+    runs-on: macos-latest
+    strategy:
+      matrix:
+        python-version: ["3.11"]
+
+    steps:
+    - uses: actions/checkout@f43a0e5ff2bd294095638e18286ca9a3d1956744 # v3.6.0
+      with:
+        submodules: recursive
+        fetch-depth: 0
+
+    - name:  Prepare cache dirs and timestamps
+      id:    prep-ccache
+      shell: bash -l {0}
+      run: |
+        mkdir -p "${CCACHE_DIR}"
+        echo "dir=$CCACHE_DIR" >> $GITHUB_OUTPUT
+        NOW=$(date -u +"%F-%T")
+        echo "timestamp=${NOW}" >> $GITHUB_OUTPUT
+        echo "today=$(/bin/date -u '+%Y%m%d')" >> $GITHUB_OUTPUT
+
+    - name: Setup compiler cache
+      uses: actions/cache@88522ab9f39a2ea568f7027eddc7d8d8bc9d59c8 # v3.3.1
+      id:    cache-ccache
+      with:
+        path: ${{ steps.prep-ccache.outputs.dir }}
+        key:  ${{ github.workflow }}-${{ matrix.python-version }}-ccache-macos-${{ steps.prep-ccache.outputs.timestamp }}
+        restore-keys: |
+          ${{ github.workflow }}-${{ matrix.python-version }}-ccache-macos-
+
+    - name: Setup Mambaforge
+      uses: conda-incubator/setup-miniconda@3b0f2504dd76ef23b6d31f291f4913fb60ab5ff3 # v2.2.0
+      with:
+        python-version: ${{ matrix.python-version }}
+        channels: conda-forge
+        channel-priority: true
+        activate-environment: numpy-dev
+        use-only-tar-bz2: false
+        miniforge-variant: Mambaforge
+        miniforge-version: latest
+        use-mamba: true
+
+    # Updates if `environment.yml` or the date changes. The latter is needed to
+    # ensure we re-solve once a day (since we don't lock versions). Could be
+    # replaced by a conda-lock based approach in the future.
+    - name: Cache conda environment
+      uses: actions/cache@88522ab9f39a2ea568f7027eddc7d8d8bc9d59c8 # v3.3.1
+      env:
+        # Increase this value to reset cache if environment.yml has not changed
+        CACHE_NUMBER: 1
+      with:
+        path: ${{ env.CONDA }}/envs/numpy-dev
+        key:
+          ${{ runner.os }}--${{ steps.prep-ccache.outputs.today }}-conda-${{ env.CACHE_NUMBER }}-${{ hashFiles('environment.yml') }}
+      id: envcache
+
+    - name: Update Conda Environment
+      run: mamba env update -n numpy-dev -f environment.yml
+      if: steps.envcache.outputs.cache-hit != 'true'
+
+    - name: Build and Install NumPy
+      shell: bash -l {0}
+      run: |
+        conda activate numpy-dev
+        CC="ccache $CC" spin build -j2
+
+    - name: Run test suite (full)
+      shell: bash -l {0}
+      run: |
+        conda activate numpy-dev
+        export OMP_NUM_THREADS=2
+        spin test -j2 -m full
+
+    - name: Ccache statistics
+      shell: bash -l {0}
+      run: |
+        conda activate numpy-dev
+        ccache -s
+
+  accelerate:
+    name: Accelerate ILP64
+    if: "github.repository == 'numpy/numpy'"
+    runs-on: macos-13
+    steps:
+    - uses: actions/checkout@f43a0e5ff2bd294095638e18286ca9a3d1956744 # v3.6.0
+      with:
+        submodules: recursive
+        fetch-depth: 0
+
+    - uses: actions/setup-python@61a6322f88396a6271a6ee3565807d608ecaddd1 # v4.7.0
+      with:
+        python-version: '3.10'
+
+    - uses: maxim-lobanov/setup-xcode@9a697e2b393340c3cacd97468baa318e4c883d98 # v1.5.1
+      with:
+        xcode-version: '14.3'
+
+    - name: Install dependencies
+      run: |
+        pip install -r build_requirements.txt
+        pip install pytest pytest-xdist hypothesis
+
+    - name: Build NumPy against Accelerate (ILP64)
+      run: |
+        spin build -- -Dblas=accelerate -Dlapack=accelerate -Duse-ilp64=true
+
+    - name: Show meson-log.txt
+      if: always()
+      run: 'cat build/meson-logs/meson-log.txt'
+
+    - name: Test
+      run: |
+        spin test -j2
diff --git a/build_requirements.txt b/build_requirements.txt
@@ -3,3 +3,4 @@ Cython>=3.0
 wheel==0.38.1
 ninja
 spin==0.5
+build
diff --git a/doc/release/upcoming_changes/24053.new_feature.rst b/doc/release/upcoming_changes/24053.new_feature.rst
@@ -0,0 +1,5 @@
+Support for the updated Accelerate BLAS/LAPACK library, including ILP64 (64-bit
+integer) support, in macOS 13.3 has been added. This brings arm64 support, and
+significant performance improvements of up to 10x for commonly used linear
+algebra operations. When Accelerate is selected at build time, the 13.3+
+version will automatically be used if available.
diff --git a/numpy/core/src/common/npy_cblas.h b/numpy/core/src/common/npy_cblas.h
@@ -25,6 +25,21 @@ enum CBLAS_SIDE {CblasLeft=141, CblasRight=142};
 
 #define CBLAS_INDEX size_t  /* this may vary between platforms */
 
+#ifdef ACCELERATE_NEW_LAPACK
+    #if __MAC_OS_X_VERSION_MAX_ALLOWED < 130300
+        #ifdef HAVE_BLAS_ILP64
+            #error "Accelerate ILP64 support is only available with macOS 13.3 SDK or later"
+        #endif
+    #else
+        #define NO_APPEND_FORTRAN
+        #ifdef HAVE_BLAS_ILP64
+            #define BLAS_SYMBOL_SUFFIX $NEWLAPACK$ILP64
+        #else
+            #define BLAS_SYMBOL_SUFFIX $NEWLAPACK
+        #endif
+    #endif
+#endif
+
 #ifdef NO_APPEND_FORTRAN
 #define BLAS_FORTRAN_SUFFIX
 #else

diff --git a/numpy/distutils/system_info.py b/numpy/distutils/system_info.py
@@ -47,6 +47,7 @@
     _numpy_info:Numeric
     _pkg_config_info:None
     accelerate_info:accelerate
+    accelerate_lapack_info:accelerate
     agg2_info:agg2
     amd_info:amd
     atlas_3_10_blas_info:atlas
@@ -534,6 +535,7 @@ def get_info(name, notfound_action=0):
           'lapack_ssl2': lapack_ssl2_info,      
           'blas_ssl2': blas_ssl2_info,          
           'accelerate': accelerate_info,      # use blas_opt instead
+          'accelerate_lapack': accelerate_lapack_info,
           'openblas64_': openblas64__info,
           'openblas64__lapack': openblas64__lapack_info,
           'openblas_ilp64': openblas_ilp64_info,
@@ -2015,14 +2017,17 @@ def _check_info(self, info):
 
 class lapack_ilp64_opt_info(lapack_opt_info, _ilp64_opt_info_mixin):
     notfounderror = LapackILP64NotFoundError
-    lapack_order = ['openblas64_', 'openblas_ilp64']
+    lapack_order = ['openblas64_', 'openblas_ilp64', 'accelerate']
     order_env_var_name = 'NPY_LAPACK_ILP64_ORDER'
 
     def _calc_info(self, name):
+        print('lapack_ilp64_opt_info._calc_info(name=%s)' % (name))
         info = get_info(name + '_lapack')
         if self._check_info(info):
             self.set_info(**info)
             return True
+        else:
+            print('%s_lapack does not exist' % (name))
         return False
 
 
@@ -2163,7 +2168,7 @@ def calc_info(self):
 
 class blas_ilp64_opt_info(blas_opt_info, _ilp64_opt_info_mixin):
     notfounderror = BlasILP64NotFoundError
-    blas_order = ['openblas64_', 'openblas_ilp64']
+    blas_order = ['openblas64_', 'openblas_ilp64', 'accelerate']
     order_env_var_name = 'NPY_BLAS_ILP64_ORDER'
 
     def _calc_info(self, name):
@@ -2625,13 +2630,27 @@ def calc_info(self):
                 link_args.extend(['-Wl,-framework', '-Wl,vecLib'])
 
             if args:
+                macros = [
+                    ('NO_ATLAS_INFO', 3),
+                    ('HAVE_CBLAS', None),
+                    ('ACCELERATE_NEW_LAPACK', None),
+                ]
+                if(os.getenv('NPY_USE_BLAS_ILP64', None)):
+                    print('Setting HAVE_BLAS_ILP64')
+                    macros += [
+                        ('HAVE_BLAS_ILP64', None),
+                        ('ACCELERATE_LAPACK_ILP64', None),
+                    ]
                 self.set_info(extra_compile_args=args,
                               extra_link_args=link_args,
-                              define_macros=[('NO_ATLAS_INFO', 3),
-                                             ('HAVE_CBLAS', None)])
+                              define_macros=macros)
 
         return
 
+class accelerate_lapack_info(accelerate_info):
+    def _calc_info(self):
+        return super()._calc_info()
+
 class blas_src_info(system_info):
     # BLAS_SRC is deprecated, please do not use this!
     # Build or install a BLAS library via your package manager or from

diff --git a/numpy/linalg/meson.build b/numpy/linalg/meson.build
@@ -1,10 +1,10 @@
 # Note that `python_xerbla.c` was excluded on Windows in setup.py;
 # unclear why and it seems needed, so unconditionally used here.
-lapack_lite_sources = [
-  'lapack_lite/python_xerbla.c',
-]
+python_xerbla_sources = ['lapack_lite/python_xerbla.c']
+
+lapack_lite_sources = []
 if not have_lapack
-  lapack_lite_sources += [
+  lapack_lite_sources = [
     'lapack_lite/f2c.c',
     'lapack_lite/f2c_c_lapack.c',
     'lapack_lite/f2c_d_lapack.c',
@@ -19,6 +19,7 @@ endif
 py.extension_module('lapack_lite',
   [
     'lapack_litemodule.c',
+    python_xerbla_sources,
     lapack_lite_sources,
   ],
   dependencies: [np_core_dep, blas_dep, lapack_dep],
@@ -29,6 +30,7 @@ py.extension_module('lapack_lite',
 py.extension_module('_umath_linalg',
   [
     'umath_linalg.cpp',
+    python_xerbla_sources,
     lapack_lite_sources,
   ],
   dependencies: [np_core_dep, blas_dep, lapack_dep],

diff --git a/numpy/meson.build b/numpy/meson.build
@@ -48,6 +48,13 @@ else
   ]
 endif
 
+macOS13_3_or_later = false
+if host_machine.system() == 'darwin'
+  r = run_command('xcrun', '-sdk', 'macosx', '--show-sdk-version', check: true)
+  sdkVersion = r.stdout().strip()
+
+  macOS13_3_or_later = sdkVersion.version_compare('>=13.3')
+endif
 
 # This is currently injected directly into CFLAGS/CXXFLAGS for wheel builds
 # (see cibuildwheel settings in pyproject.toml), but used by CI jobs already
@@ -81,6 +88,7 @@ endif
 # https://github.com/mesonbuild/meson/issues/2835
 blas_name = get_option('blas')
 lapack_name = get_option('lapack')
+
 # pkg-config uses a lower-case name while CMake uses a capitalized name, so try
 # that too to make the fallback detection with CMake work
 if blas_name == 'openblas'
@@ -90,6 +98,23 @@ if blas_name == 'openblas'
     _openblas_names = ['openblas', 'OpenBLAS']
   endif
   blas = dependency(_openblas_names, required: false)
+elif blas_name.to_lower() == 'accelerate'
+  # macOS 13.3+ has updated interfaces aligned with BLAS/LAPACK 3.9.1.  Use them if available.
+  if macOS13_3_or_later
+    accelerate_compile_args = ['-DACCELERATE_NEW_LAPACK']
+    if(use_ilp64)
+      accelerate_compile_args += '-DACCELERATE_LAPACK_ILP64'
+    endif
+    blas = declare_dependency(
+            compile_args: accelerate_compile_args,
+            dependencies: dependency('Accelerate')
+          )
+  else
+    if(use_ilp64)
+      error('macOS SDK 13.3+ is required for ILP64 support.')
+    endif
+    blas = dependency('Accelerate')
+  endif
 else
   blas = dependency(blas_name, required: false)
 endif
@@ -112,14 +137,22 @@ if have_blas
   #   `dependency('blas', modules: cblas)`
   # see https://github.com/mesonbuild/meson/pull/10921.
   have_cblas = false
-  if cc.links('''
+  if blas_name.to_lower() == 'accelerate'
+    _cblas_header = '<Accelerate/Accelerate.h>'
+  elif blas_name.to_lower().startswith('mkl')
+    _cblas_header = '<mkl_cblas.h>'
+  else
+    _cblas_header = '<cblas.h>'
+  endif
+  if cc.links(f'''
       #ifndef BLAS_SYMBOL_SUFFIX
       #  define BLAS_SYMBOL_SUFFIX
       #endif
       #define EXPAND(suffix) cblas_ddot ## suffix
       #define DDOT(suffix) EXPAND(suffix)
 
-      #include <cblas.h>
+      #include @_cblas_header@
+
       int main(int argc, const char *argv[])
       {
           double a[4] = {1,2,3,4};
@@ -178,9 +211,27 @@ else
 endif
 
 if lapack_name == 'openblas'
-  lapack_name = ['openblas', 'OpenBLAS']
+  lapack_dep = dependency(['openblas', 'OpenBLAS'], required: false)
+elif lapack_name.to_lower() == 'accelerate'
+  # macOS 13.3+ has updated interfaces aligned with BLAS/LAPACK 3.9.1.  Use them if available.
+  if macOS13_3_or_later
+    accelerate_compile_args = ['-DACCELERATE_NEW_LAPACK']
+    if(use_ilp64)
+      accelerate_compile_args += '-DACCELERATE_LAPACK_ILP64'
+    endif
+    lapack_dep = declare_dependency(
+               compile_args: accelerate_compile_args,
+               dependencies: dependency('Accelerate')
+             )
+  else
+    if(use_ilp64)
+      error('macOS SDK 13.3+ is required for ILP64 support.')
+    endif
+    lapack_dep = dependency('Accelerate')
+  endif
+else
+  lapack_dep = dependency(lapack_name, required: false)
 endif
-lapack_dep = dependency(lapack_name, required: false)
 have_lapack = lapack_dep.found()
 if not have_lapack and not allow_noblas
   error('No LAPACK library detected! Install one, or use the ' + \