Repository navigation
docs/specs: sync with current status, record benchmark references - #6
Merged
Merged
Conversation
Bring the docs and living OpenSpec specs back in line with the shipped code and capture the GPU benchmark results as an in-repo reference baseline. Benchmark references (new docs/benchmarks.md): - Methodology, host specs, and recorded GLUPS for the D3Q19 wind tunnel across CPU / CUDA / OpenCL, at two grid sizes, on the RTX 5060 + i9-12900K box (~1.15 GLUPS on GPU, ~30x the 24-thread CPU; ~78% of the card's DRAM peak). - Indicative cross-machine row vs the Apple M2 Max (~1.35 GLUPS). - Linked from README, docs/features.md, and docs/backends.md. Drift fixes: - docs/features.md: CUDA/OpenCL were still labelled "stubs" / "planned"; mark them validated (they ship and pass ~1e-5 vs the fp64 oracle). - README: soften the SciPP claim — it is wired into the build but not yet linked by any target (matches features.md's "wired, unused"). - build-and-packaging spec: replace the FetchContent auto-fetch requirement with the actual bootstrap_deps.sh + find_package(NumPP CONFIG) mechanism; fill the TBD Purpose; add the macOS CI job and the -Wl,--no-as-needed -ltbb detail to the CI requirement; add a "throughput benchmark with recorded references" requirement. - oracle-validation and external-fields specs: fill their TBD Purpose placeholders. openspec validate --all --strict passes (14/14).
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Answers the "are README and OpenSpec in sync?" audit by fixing the drift it surfaced, and captures the GPU benchmark results as an in-repo reference baseline (rather than living only in prose).
New:
docs/benchmarks.md(reference baseline)Records methodology, host specs, and measured GLUPS for the D3Q19 wind tunnel across CPU / CUDA / OpenCL at two grid sizes, so future runs have a known baseline to compare against.
(Full table incl. 160×80×80 in the doc.) Host: i9-12900K (24 threads) + RTX 5060. Plus an indicative cross-machine row vs the Apple M2 Max (~1.35 GLUPS). Linked from README,
docs/features.md, anddocs/backends.md.Drift fixes
docs/features.mdREADME.mdfeatures.md)build-and-packagingspecbootstrap_deps.sh+find_package(NumPP CONFIG)mechanismbuild-and-packagingspecTBD … update after archivebuild-and-packagingspec-Wl,--no-as-needed -ltbbdetailbuild-and-packagingspecoracle-validation,external-fieldsspecsTBD …placeholderVerification
openspec validate --all --strict→ 14/14 passed.just benchrun on the reference box before recording.TBD/FetchContent/ "stub" drift in living specs or docs.