.flake8

 [flake8]
 max-line-length=120
-exclude=pystencils/jupyter.py,
-        pystencils/plot.py
-        pystencils/session.py
+exclude=src/pystencils/jupyter.py,
+        src/pystencils/plot.py
+        src/pystencils/session.py
 ignore = W293 W503 W291 C901 E741

.gitattributes

-pystencils/_version.py export-subst
+src/pystencils/_version.py export-subst

.gitignore

@@ -5,7 +5,7 @@ __pycache__
 *.vti
 /build
 /dist
-/*.egg-info
+*.egg-info
 .cache
 _build
 /html_doc
@@ -15,15 +15,16 @@ _build
 _local_tmp
 RELEASE-VERSION
 test-report
-pystencils/boundaries/createindexlistcython.c
-pystencils/boundaries/createindexlistcython.*.so
-pystencils_tests/tmp
-pystencils_tests/var
-pystencils_tests/kerncraft_inputs/.2d-5pt.c_kerncraft/
-pystencils_tests/kerncraft_inputs/.3d-7pt.c_kerncraft/
+src/pystencils/boundaries/createindexlistcython.c
+src/pystencils/boundaries/createindexlistcython.*.so
+tests/tmp
+tests/var
+tests/kerncraft_inputs/.2d-5pt.c_kerncraft/
+tests/kerncraft_inputs/.3d-7pt.c_kerncraft/
 report.xml
 coverage_report/
 
 
 # macOS
 **/.DS_Store
+*.uuid

.gitlab-ci.yml

 stages:
   - pretest
   - test
+  - nightly
+  - docs
   - deploy
 
 
+# --------------------------  Templates ------------------------------------------------------------------------------------
+
+# Base configuration for jobs meant to run at every commit
+.every-commit:
+  rules:
+    - if: $CI_PIPELINE_SOURCE != "schedule"
+
+# Configuration for jobs meant to run on each commit to pycodegen/pystencils/master
+.every-commit-master:
+  rules:
+    - if: '$CI_PIPELINE_SOURCE != "schedule" && $CI_PROJECT_PATH == "pycodegen/pystencils" && $CI_COMMIT_BRANCH == "master"'
+
+# Base configuration for jobs meant to run at a schedule
+.scheduled:
+  rules:
+    - if: $CI_PIPELINE_SOURCE == "schedule"
+
 # --------------------------  Tests ------------------------------------------------------------------------------------
 
 # Normal test - runs on every commit all but "long run" tests
 tests-and-coverage:
   stage: pretest
-  except:
-    variables:
-      - $ENABLE_NIGHTLY_BUILDS
-  image: i10git.cs.fau.de:5005/pycodegen/pycodegen/full
+  extends: .every-commit
+  image: i10git.cs.fau.de:5005/pycodegen/pycodegen/full:cupy12.3
+  before_script:
+    - pip install -e .
   script:
-    - pip install sympy --upgrade
     - env
     - pip list
     - export NUM_CORES=$(nproc --all)
     - mkdir -p ~/.config/matplotlib
     - echo "backend:template" > ~/.config/matplotlib/matplotlibrc
     - mkdir public
-    - py.test -v -n $NUM_CORES --cov-report html --cov-report term --cov=. -m "not longrun" --html test-report/index.html --junitxml=report.xml
-    - python3 -m coverage xml
+    - pytest -v -n $NUM_CORES --cov-report html --cov-report xml --cov-report term --cov=. -m "not longrun" --html test-report/index.html --junitxml=report.xml
+    - python -m coverage xml
   tags:
     - docker
     - cuda11
@@ -44,9 +62,11 @@ tests-and-coverage-with-longrun:
   stage: test
   when: manual
   allow_failure: true
-  image: i10git.cs.fau.de:5005/pycodegen/pycodegen/full
-  script:
+  image: i10git.cs.fau.de:5005/pycodegen/pycodegen/full:cupy12.3
+  before_script:
     - pip install sympy --upgrade
+    - pip install -e .
+  script:
     - env
     - pip list
     - export NUM_CORES=$(nproc --all)
@@ -62,13 +82,14 @@ tests-and-coverage-with-longrun:
 # pipeline with latest python version
 latest-python:
   stage: test
-  except:
-    variables:
-      - $ENABLE_NIGHTLY_BUILDS
+  extends: .every-commit
   image: i10git.cs.fau.de:5005/pycodegen/pycodegen/latest_python
+  before_script:
+    - pip install -e .
   script:
     - env
     - pip list
+    - pip install -e .
     - export NUM_CORES=$(nproc --all)
     - mkdir -p ~/.config/matplotlib
     - echo "backend:template" > ~/.config/matplotlib/matplotlibrc
@@ -84,41 +105,34 @@ latest-python:
 
 
 # Minimal tests in windows environment
-minimal-windows:
-  stage: test
-  except:
-    variables:
-      - $ENABLE_NIGHTLY_BUILDS
-  tags:
-    - win
-  script:
-    - export NUM_CORES=$(nproc --all)
-    - source /cygdrive/c/Users/build/Miniconda3/Scripts/activate
-    - source activate pystencils
-    - pip install joblib
-    - pip list
-    - python -c "import numpy"
-    - py.test -v -m "not (notebook or longrun)"
+#minimal-windows:
+#  stage: test
+#  tags:
+#    - win
+#  script:
+#    - export NUM_CORES=$(nproc --all)
+#    - source /cygdrive/c/Users/build/Miniconda3/Scripts/activate
+#    - source activate pystencils
+#    - pip install joblib
+#    - pip list
+#    - python -c "import numpy"
+#    - py.test -v -m "not (notebook or longrun)"
 
 ubuntu:
   stage: test
-  except:
-    variables:
-      - $ENABLE_NIGHTLY_BUILDS
+  extends: .every-commit
   image: i10git.cs.fau.de:5005/pycodegen/pycodegen/ubuntu
   before_script:
-    - apt-get -y remove python3-sympy
     - ln -s /usr/include/locale.h /usr/include/xlocale.h
-    - pip3 install `grep -Eo 'sympy[>=]+[0-9\.]+' setup.py | sed 's/>/=/g'`
-    # - pip3 install `grep -Eo 'sympy[>=]+[0-9\.]+' setup.py | sed 's/>/=/g'`
+    - pip3 install -e .
   script:
     - export NUM_CORES=$(nproc --all)
     - mkdir -p ~/.config/matplotlib
     - echo "backend:template" > ~/.config/matplotlib/matplotlibrc
     - sed -i 's/--doctest-modules //g' pytest.ini
     - env
-    - pip3 list
-    - pytest-3 -v -n $NUM_CORES -m "not longrun" --junitxml=report.xml
+    - pip list
+    - pytest -v -n $NUM_CORES -m "not longrun" --junitxml=report.xml
   tags:
     - docker
     - cuda11
@@ -130,10 +144,11 @@ ubuntu:
 
 .multiarch_template:
   stage: test
-  except:
-    variables:
-      - $ENABLE_NIGHTLY_BUILDS
+  extends: .every-commit
+  allow_failure: true
   before_script: &multiarch_before_script
+    # - pip3 install -v .
+    - export PYTHONPATH=src
     - python3 -c "import pystencils as ps; ps.cpu.cpujit.read_config()"
     - sed -i '/^fail_under.*/d' pytest.ini
   script:
@@ -143,20 +158,26 @@ ubuntu:
     - sed -i 's/--doctest-modules //g' pytest.ini
     - env
     - pip3 list
-    - pytest-3 -v -n $NUM_CORES --junitxml=report.xml pystencils_tests/test_*vec*.py pystencils_tests/test_random.py
+    - python3 -m pytest -v -n $NUM_CORES --cov-report html --cov-report xml --cov=. --junitxml=report.xml tests/test_*vec*.py tests/test_random.py tests/test_half_precision.py
+    - python3 -m coverage xml
   tags:
     - docker
     - AVX
   artifacts:
     when: always
+    paths:
+      - coverage_report
     reports:
+      coverage_report:
+        coverage_format: cobertura
+        path: coverage.xml
       junit: report.xml
 
 arm64v8:
   extends: .multiarch_template
   image: i10git.cs.fau.de:5005/pycodegen/pycodegen/arm64
   variables:
-    PYSTENCILS_SIMD: "neon"
+    QEMU_CPU: "cortex-a76"
   before_script:
     - *multiarch_before_script
     - sed -i s/march=native/march=armv8-a/g ~/.config/pystencils/config.json
@@ -164,45 +185,41 @@ arm64v8:
 ppc64le:
   extends: .multiarch_template
   image: i10git.cs.fau.de:5005/pycodegen/pycodegen/ppc64le
-  variables:
-    PYSTENCILS_SIMD: "vsx"
   before_script:
     - *multiarch_before_script
     - sed -i s/mcpu=native/mcpu=power8/g ~/.config/pystencils/config.json
 
 arm64v9:
   # SVE support is still unreliable in GCC 11 (incorrect code for fixed-width vectors, internal compiler errors).
+  # For half precision Clang is necessary
   extends: .multiarch_template
   image: i10git.cs.fau.de:5005/pycodegen/pycodegen/arm64
-  variables:
-    PYSTENCILS_SIMD: "sve128,sve256,sve512,sve"
   before_script:
     - *multiarch_before_script
-    - sed -i s/march=native/march=armv8-a+sve/g ~/.config/pystencils/config.json
+    - sed -i s/march=native/march=armv9-a+sve2+sme/g ~/.config/pystencils/config.json
     - sed -i s/g\+\+/clang++/g ~/.config/pystencils/config.json
 
 riscv64:
   # RISC-V vector extension are currently not supported by GCC.
-  # Also, the image is built without the libomp package which is not yet available on Ubuntu.
   extends: .multiarch_template
   image: i10git.cs.fau.de:5005/pycodegen/pycodegen/riscv64
   variables:
+    # explicitly set SIMD as detection requires QEMU >= 8.1
     PYSTENCILS_SIMD: "rvv"
-    QEMU_CPU: "rv64,v=true"
+    QEMU_CPU: "rv64,v=true,zicboz=true"
   before_script:
     - *multiarch_before_script
-    - sed -i 's/march=native/march=rv64imfdv/g' ~/.config/pystencils/config.json
-    - sed -i s/g\+\+/clang++/g ~/.config/pystencils/config.json
-    - sed -i 's/fopenmp/fopenmp=libgomp -I\/usr\/include\/riscv64-linux-gnu/g' ~/.config/pystencils/config.json
+    - sed -i 's/march=native/march=rv64imfdvzicboz/g' ~/.config/pystencils/config.json
+    - sed -i s/g\+\+/clang++-15/g ~/.config/pystencils/config.json
 
 minimal-conda:
   stage: pretest
-  except:
-    variables:
-      - $ENABLE_NIGHTLY_BUILDS
+  extends: .every-commit
   image: i10git.cs.fau.de:5005/pycodegen/pycodegen/minimal_conda
+  before_script:
+    - pip install -e .
   script:
-    - python setup.py quicktest
+    - python quicktest.py
   tags:
     - docker
     - cuda
@@ -210,13 +227,13 @@ minimal-conda:
 
 minimal-sympy-master:
   stage: test
-  except:
-    variables:
-      - $ENABLE_NIGHTLY_BUILDS
+  extends: .every-commit
   image: i10git.cs.fau.de:5005/pycodegen/pycodegen/minimal_conda
+  before_script:
+    - pip install -e .
   script:
     - python -m pip install --upgrade git+https://github.com/sympy/sympy.git
-    - python setup.py quicktest
+    - python quicktest.py
   allow_failure: true
   tags:
     - docker
@@ -267,26 +284,57 @@ pycodegen-integration:
     reports:
       junit: pycodegen/*/report.xml
 
+
+# -------------------- Scheduled Tasks --------------------------------------------------------------------------
+
+
+# Nightly test against the latest (pre-release) version of SymPy published on PyPI
+nightly-sympy:
+  stage: nightly
+  needs: []
+  extends: .scheduled
+  image: i10git.cs.fau.de:5005/pycodegen/pycodegen/latest_python
+  before_script:
+    - pip install -e .
+    - pip install --upgrade --pre sympy
+  script:
+    - env
+    - pip list
+    - export NUM_CORES=$(nproc --all)
+    - mkdir -p ~/.config/matplotlib
+    - echo "backend:template" > ~/.config/matplotlib/matplotlibrc
+    - mkdir public
+    - pytest -v -n $NUM_CORES -m "not longrun" --junitxml=report.xml
+  tags:
+    - docker
+    - AVX
+    - cuda
+  artifacts:
+    when: always
+    reports:
+      junit: report.xml
+
 # -------------------- Linter & Documentation --------------------------------------------------------------------------
 
 
 flake8-lint:
   stage: pretest
-  except:
-    variables:
-      - $ENABLE_NIGHTLY_BUILDS
+  extends: .every-commit
   image: i10git.cs.fau.de:5005/pycodegen/pycodegen/full
   script:
-    - flake8 pystencils
+    - flake8 src/pystencils
   tags:
     - docker
 
 
 build-documentation:
-  stage: test
+  stage: docs
+  extends: .every-commit
   image: i10git.cs.fau.de:5005/pycodegen/pycodegen/documentation
+  needs: []
+  before_script:
+    - pip install -e .
   script:
-    - export PYTHONPATH=`pwd`
     - mkdir html_doc
     - sphinx-build -b html doc  html_doc
     - sphinx-build -W -b html doc  html_doc
@@ -299,7 +347,9 @@ build-documentation:
 
 pages:
   image: i10git.cs.fau.de:5005/pycodegen/pycodegen/full
+  extends: .every-commit-master
   stage: deploy
+  needs: ["tests-and-coverage", "build-documentation"]
   script:
     - ls -l
     - mv coverage_report html_doc
@@ -309,5 +359,3 @@ pages:
       - public
   tags:
     - docker
-  only:
-    - master@pycodegen/pystencils

MANIFEST.in

-include README.md
-include COPYING.txt
 include AUTHORS.txt
 include CONTRIBUTING.md
-CHANGELOG.md
-global-include *.pyx
-include versioneer.py
-include pystencils/_version.py
+include CHANGELOG.md

README.md

@@ -52,7 +52,7 @@ pip install pystencils[interactive]
 Without `[interactive]` you get a minimal version with very little dependencies.
 
 All options:
-- `gpu`: use this if an NVIDIA GPU is available and CUDA is installed
+- `gpu`: use this if an NVIDIA or AMD GPU is available and CUDA or ROCm is installed
 - `alltrafos`: pulls in additional dependencies for loop simplification e.g. libisl
 - `bench_db`: functionality to store benchmark result in object databases
 - `interactive`: installs dependencies to work in Jupyter including image I/O, plotting etc.
@@ -63,7 +63,7 @@ Options can be combined e.g.
 pip install pystencils[interactive, gpu, doc]
 ```
 
-pystencils is also fully compatible with Windows machines. If working with visual studio and pycuda makes sure to run example files first to ensure that pycuda can find the compiler's executable.
+pystencils is also fully compatible with Windows machines. If working with visual studio and cupy makes sure to run example files first to ensure that cupy can find the compiler's executable.
 
 Documentation
 -------------
@@ -81,4 +81,7 @@ Many thanks go to the [contributors](https://i10git.cs.fau.de/pycodegen/pystenci
 If you use pystencils in a publication, please cite the following articles:
 
 Overview:
-  - M. Bauer et al, Code Generation for Massively Parallel Phase-Field Simulations. Association for Computing Machinery, 2019. https://doi.org/10.1145/3295500.3356186
\ No newline at end of file
+  - M. Bauer et al, Code Generation for Massively Parallel Phase-Field Simulations. Association for Computing Machinery, 2019. https://doi.org/10.1145/3295500.3356186
+
+Performance Modelling:
+  - D. Ernst et al, Analytical performance estimation during code generation on modern GPUs. Journal of Parallel and Distributed Computing, 2023. https://doi.org/10.1016/j.jpdc.2022.11.003

binder/environment.yml

@@ -7,7 +7,7 @@
 #     conda env create -f conda_environment_user.yml
 #     . activate pystencils
 #
-# If you have CUDA installed and want to use your GPU, uncomment the last line to install pycuda
+# If you have CUDA or ROCm installed and want to use your GPU, uncomment the last line to install cupy
 #
 # ----------------------------------------------------------------------------------------------------------------------
 
@@ -32,4 +32,4 @@ dependencies:
       - ipy_table  # HTML tables for jupyter notebooks
       - pyevtk # VTK output for serial simulations
       - blitzdb # file-based No-SQL database to store simulation results
-      #- pycuda # add this if you have CUDA installed
+      #- cupy # add this if you have CUDA or ROCm installed

conftest.py

@@ -8,7 +8,7 @@ import nbformat
 import pytest
 from nbconvert import PythonExporter
 
-from pystencils.boundaries.createindexlistcython import *  # NOQA
+from pystencils.boundaries.createindexlist import *  # NOQA
 # Trigger config file reading / creation once - to avoid race conditions when multiple instances are creating it
 # at the same time
 from pystencils.cpu import cpujit
@@ -40,47 +40,48 @@ def add_path_to_ignore(path):
 
 
 collect_ignore = [os.path.join(SCRIPT_FOLDER, "doc", "conf.py"),
-                  os.path.join(SCRIPT_FOLDER, "pystencils", "opencl", "opencl.autoinit")]
-add_path_to_ignore('pystencils_tests/benchmark')
+                  os.path.join(SCRIPT_FOLDER, "src", "pystencils", "opencl", "opencl.autoinit")]
+add_path_to_ignore('tests/benchmark')
 add_path_to_ignore('_local_tmp')
 
 
 try:
-    import pycuda
+    import cupy
 except ImportError:
-    collect_ignore += [os.path.join(SCRIPT_FOLDER, "pystencils_tests/test_cudagpu.py")]
-    add_path_to_ignore('pystencils/gpucuda')
+    collect_ignore += [os.path.join(SCRIPT_FOLDER, "tests/test_gpu.py")]
+    add_path_to_ignore('src/pystencils/gpu')
 
 try:
     import waLBerla
 except ImportError:
-    collect_ignore += [os.path.join(SCRIPT_FOLDER, "pystencils_tests/test_aligned_array.py"),
-                       os.path.join(SCRIPT_FOLDER, "pystencils_tests/test_datahandling_parallel.py"),
+    collect_ignore += [os.path.join(SCRIPT_FOLDER, "tests/test_aligned_array.py"),
+                       os.path.join(SCRIPT_FOLDER, "tests/test_datahandling_parallel.py"),
                        os.path.join(SCRIPT_FOLDER, "doc/notebooks/03_tutorial_datahandling.ipynb"),
-                       os.path.join(SCRIPT_FOLDER, "pystencils/datahandling/parallel_datahandling.py"),
-                       os.path.join(SCRIPT_FOLDER, "pystencils_tests/test_small_block_benchmark.ipynb")]
+                       os.path.join(SCRIPT_FOLDER, "src/pystencils/datahandling/parallel_datahandling.py"),
+                       os.path.join(SCRIPT_FOLDER, "tests/test_small_block_benchmark.ipynb")]
 
 try:
     import blitzdb
 except ImportError:
-    add_path_to_ignore('pystencils/runhelper')
-    collect_ignore += [os.path.join(SCRIPT_FOLDER, "pystencils_tests/test_parameterstudy.py")]
+    add_path_to_ignore('src/pystencils/runhelper')
+    collect_ignore += [os.path.join(SCRIPT_FOLDER, "tests/test_parameterstudy.py")]
+    collect_ignore += [os.path.join(SCRIPT_FOLDER, "tests/test_json_serializer.py")]
 
 try:
     import islpy
 except ImportError:
-    collect_ignore += [os.path.join(SCRIPT_FOLDER, "pystencils/integer_set_analysis.py")]
+    collect_ignore += [os.path.join(SCRIPT_FOLDER, "src/pystencils/integer_set_analysis.py")]
 
 try:
     import graphviz
 except ImportError:
-    collect_ignore += [os.path.join(SCRIPT_FOLDER, "pystencils/backends/dot.py")]
+    collect_ignore += [os.path.join(SCRIPT_FOLDER, "src/pystencils/backends/dot.py")]
     collect_ignore += [os.path.join(SCRIPT_FOLDER, "doc/notebooks/01_tutorial_getting_started.ipynb")]
 
 try:
     import pyevtk
 except ImportError:
-    collect_ignore += [os.path.join(SCRIPT_FOLDER, "pystencils/datahandling/vtk.py")]
+    collect_ignore += [os.path.join(SCRIPT_FOLDER, "src/pystencils/datahandling/vtk.py")]
 
 collect_ignore += [os.path.join(SCRIPT_FOLDER, 'setup.py')]
 

doc/conf.py

@@ -26,14 +26,14 @@ templates_path = ['_templates']
 source_suffix = '.rst'
 master_doc = 'index'
 
-copyright = f'{datetime.datetime.now().year}, Martin Bauer'
-author = 'Martin Bauer'
+copyright = f'{datetime.datetime.now().year}, Martin Bauer, Markus Holzer, Frederik Hennig'
+author = 'Martin Bauer, Markus Holzer, Frederik Hennig'
 # The short X.Y version (including .devXXXX, rcX, b1 suffixes if present)
 version = re.sub(r'(\d+\.\d+)\.\d+(.*)', r'\1\2', pystencils.__version__)
 version = re.sub(r'(\.dev\d+).*?$', r'\1', version)
 # The full version, including alpha/beta/rc tags.
 release = pystencils.__version__
-language = None
+language = 'en'
 exclude_patterns = ['_build', 'Thumbs.db', '.DS_Store', '**.ipynb_checkpoints']
 default_role = 'any'
 pygments_style = 'sphinx'

doc/sphinx/kernel_compile_and_call.rst

@@ -27,11 +27,11 @@ Code printing
 GPU Indexing
 -------------
 
-.. autoclass:: pystencils.gpucuda.AbstractIndexing
+.. autoclass:: pystencils.gpu.AbstractIndexing
    :members:
 
-.. autoclass:: pystencils.gpucuda.BlockIndexing
+.. autoclass:: pystencils.gpu.BlockIndexing
    :members:
 
-.. autoclass:: pystencils.gpucuda.LineIndexing
+.. autoclass:: pystencils.gpu.LineIndexing
    :members:

pyproject.toml

+[project]
+name = "pystencils"
+description = "Speeding up stencil computations on CPUs and GPUs"
+dynamic = ["version"]
+readme = "README.md"
+authors = [
+    { name = "Martin Bauer" },
+    { name = "Jan Hönig " },
+    { name = "Markus Holzer" },
+    { name = "Frederik Hennig" },
+    { email = "cs10-codegen@fau.de" },
+]
+license = { file = "COPYING.txt" }
+requires-python = ">=3.10"
+dependencies = ["sympy>=1.9,<=1.12.1", "numpy>=1.8.0", "appdirs", "joblib", "pyyaml", "fasteners"]
+classifiers = [
+    "Development Status :: 4 - Beta",
+    "Framework :: Jupyter",
+    "Topic :: Software Development :: Code Generators",
+    "Topic :: Scientific/Engineering :: Physics",
+    "Intended Audience :: Developers",
+    "Intended Audience :: Science/Research",
+    "License :: OSI Approved :: GNU Affero General Public License v3 or later (AGPLv3+)",
+]
+
+[project.urls]
+"Bug Tracker" = "https://i10git.cs.fau.de/pycodegen/pystencils/-/issues"
+"Documentation" = "https://pycodegen.pages.i10git.cs.fau.de/pystencils/"
+"Source Code" = "https://i10git.cs.fau.de/pycodegen/pystencils"
+
+[project.optional-dependencies]
+gpu = ['cupy']
+alltrafos = ['islpy', 'py-cpuinfo']
+bench_db = ['blitzdb', 'pymongo', 'pandas']
+interactive = [
+    'matplotlib',
+    'ipy_table',
+    'imageio',
+    'jupyter',
+    'pyevtk',
+    'rich',
+    'graphviz',
+]
+use_cython = [
+    'Cython'
+]
+doc = [
+    'sphinx',
+    'sphinx_rtd_theme',
+    'nbsphinx',
+    'sphinxcontrib-bibtex',
+    'sphinx_autodoc_typehints',
+    'pandoc',
+]
+tests = [
+    'pytest',
+    'pytest-cov',
+    'pytest-html',
+    'ansi2html',
+    'pytest-xdist',
+    'flake8',
+    'nbformat',
+    'nbconvert',
+    'ipython',
+    'matplotlib',
+    'py-cpuinfo',
+    'randomgen>=1.18',
+]
+
+[build-system]
+requires = [
+    "setuptools>=61",
+    "versioneer[toml]>=0.29",
+    # 'Cython'
+]
+build-backend = "setuptools.build_meta"
+
+[tool.setuptools.package-data]
+pystencils = [
+    "include/*.h",
+    "boundaries/createindexlistcython.pyx"
+]
+
+[tool.setuptools.packages.find]
+where = ["src"]
+include = ["pystencils", "pystencils.*"]
+namespaces = false
+
+[tool.versioneer]
+# See the docstring in versioneer.py for instructions. Note that you must
+# re-run 'versioneer.py setup' after changing this section, and commit the
+# resulting files.
+VCS = "git"
+style = "pep440"
+versionfile_source = "src/pystencils/_version.py"
+versionfile_build = "pystencils/_version.py"
+tag_prefix = "release/"
+parentdir_prefix = "pystencils-"

pystencils/datahandling/pycuda.py

-try:
-    import pycuda.gpuarray as gpuarray
-except ImportError:
-    gpuarray = None
-import numpy as np
-
-import pystencils
-
-
-class PyCudaArrayHandler:
-
-    def __init__(self):
-        import pycuda.autoinit  # NOQA
-
-    def zeros(self, shape, dtype=np.float64, order='C'):
-        cpu_array = np.zeros(shape=shape, dtype=dtype, order=order)
-        return self.to_gpu(cpu_array)
-
-    def ones(self, shape, dtype=np.float64, order='C'):
-        cpu_array = np.ones(shape=shape, dtype=dtype, order=order)
-        return self.to_gpu(cpu_array)
-
-    def empty(self, shape, dtype=np.float64, layout=None):
-        if layout:
-            cpu_array = pystencils.field.create_numpy_array_with_layout(shape=shape, dtype=dtype, layout=layout)
-            return self.to_gpu(cpu_array)
-        else:
-            return gpuarray.empty(shape, dtype)
-
-    @staticmethod
-    def to_gpu(array):
-        return gpuarray.to_gpu(array)
-
-    @staticmethod
-    def upload(array, numpy_array):
-        array.set(numpy_array)
-
-    @staticmethod
-    def download(array, numpy_array):
-        array.get(numpy_array)
-
-    def randn(self, shape, dtype=np.float64):
-        cpu_array = np.random.randn(*shape).astype(dtype)
-        return self.to_gpu(cpu_array)
-
-    from_numpy = to_gpu
-
-
-class PyCudaNotAvailableHandler:
-    def __getattribute__(self, name):
-        raise NotImplementedError("Unable to initiaize PyCuda! "
-                                  "Try to run `import pycuda.autoinit` to check whether PyCuda is working correctly!")

pystencils/include/PyStencilsField.h

-#pragma once
-
-extern "C++" {
-#ifdef __CUDA_ARCH__
-template <typename DTYPE_T, std::size_t DIMENSION> struct PyStencilsField {
-  DTYPE_T *data;
-  DTYPE_T shape[DIMENSION];
-  DTYPE_T stride[DIMENSION];
-};
-#else
-#include <array>
-
-template <typename DTYPE_T, std::size_t DIMENSION> struct PyStencilsField {
-  DTYPE_T *data;
-  std::array<DTYPE_T, DIMENSION> shape;
-  std::array<DTYPE_T, DIMENSION> stride;
-};
-#endif
-}