Skip to content

Archive LLVM 19 optimization measurements and research reports - #11

Closed
brandonros wants to merge 5 commits into
stack/ptx-experimentsfrom
stack/ptx-research-results
Closed

brandonros wants to merge 5 commits into
stack/ptx-experimentsfrom
stack/ptx-research-results

Conversation

@brandonros

Copy link
Copy Markdown
Owner

Preserves the original LLVM 19 research reports and generated evidence in a separate review layer. These measurements are historical and do not validate LLVM 21 performance or correctness. This PR can be kept as an archive without merging its generated reports into the production stack.

Part 5/5 of the split of #4; base: stack/ptx-experiments. The LLVM 21/CUDA foundation remains #3. Original #4 stays open as the unchanged integration reference.

Validation: Offline locked Cargo metadata passes for this layer. Both host shuffle-control tests pass. Both guarded-select host tests pass; Python unit tests pass (6 at cleanup, 9 with the full experiment suite). Linux CUDA compilation and GPU execution have not been run locally.

The complete five-branch stack reproduces #4 at d2104a0, except for the added pull-request CI trigger. No original branches were rewritten.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant