Build compiled CPU wheels with C++ bindings on PyPI#248
Merged
jameslehoux merged 2 commits intomasterfrom Apr 9, 2026
Merged
Conversation
Only section 3 (plotfile visualisation) truly requires openimpala.core. All other sections now work on CPU: - Import core conditionally; set oic = None on failure with one message - Sections 1-2 (tortuosity, anisotropy) use oi.* only — unaffected - Section 3 (cells 10-11) skips with a one-line message when oic is None - Section 4 (BPX export) uses build_bpx_json() helper that falls back to pure Python when the C++ ResultsJSON is unavailable - Section 5 (PyBaMM) works unchanged — just reads the JSON file https://claude.ai/code/session_01BXGy2knFzA2bgVnuNRNVVQ
Replace the pure-Python wheel build (pypi-wheels-cpu.yml) with a cibuildwheel build that compiles the C++ backend (pybind11 _core module) against CPU-only HYPRE and AMReX (GPU_BACKEND=NONE). This makes `from openimpala import core` work on CPU — no CUDA required. Tutorial 2 is restored to its original clean form with no fallback code; `oic.TortuosityHypre`, `oic.ResultsJSON`, etc. all work on CPU now. Also return the 3-D solution field from the pure-Python solver path and expose it as TortuosityResult.solution_field. https://claude.ai/code/session_01BXGy2knFzA2bgVnuNRNVVQ
Performance Benchmark Results
Fastest solver: bicgstab at 64³ (0.3875s) Benchmark: uniform block (analytical τ = (N-1)/N) |
Code Coverage ReportGenerated by CI — coverage data from gcovr |
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Replace the pure-Python wheel build (pypi-wheels-cpu.yml) with a
cibuildwheel build that compiles the C++ backend (pybind11 _core module)
against CPU-only HYPRE and AMReX (GPU_BACKEND=NONE). This makes
from openimpala import corework on CPU — no CUDA required.Tutorial 2 is restored to its original clean form with no fallback code;
oic.TortuosityHypre,oic.ResultsJSON, etc. all work on CPU now.Also return the 3-D solution field from the pure-Python solver path and
expose it as TortuosityResult.solution_field.