Repository navigation
gpu-checks: GPU check and D1-D4 diagnostic harness for PRs #26-#28 - #29
Merged
Merged
Conversation
I built this harness to check the three outside PRs (#26-#28) on a GPU device host without merging them. It includes the unattended G1-G5 runner, the D1-D4 long-form diagnostic mode, a pinned-corpus preflight, the 88 GiB pilot memory admission (rechecked before every full-profile launch), sealed output directories, payload-validated INCONCLUSIVE verdicts, wrapper exit mapping (memory WAIT 75 -> 0) and CPU-only selftests. Host-local values come only from GPU_CHECK_* environment settings; see gpu-checks/README.md "Device configuration". Provenance: one squashed snapshot of 12 locally reviewed commits (final tree identical to local commit 2fecb88).
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
I'm adding the harness I use to check the outside PRs #26-#28 on a GPU device host without merging them. This PR contains only
gpu-checks/; it does not include or merge #26-#28.What it does:
--out-dir.completeonly for coherent D1-D4 evidence; anything missing, failed or malformed renders INCONCLUSIVE with no comparison.run-diag.shmaps to 0; genuine failures stay non-zero.GPU_CHECK_*environment settings (README "Device configuration").My checks on this exact head (7eb0f7e):
py_compile runner.py;bash -non all five scripts;runner.py --diag-selftestPASS;run-diag.sh --planandrun-all.sh --planbyte-identical to the committed outputs; corpus sha256 equals the pin;selftest.shPASS (dummy systemd unit pause/resume, wrapper exit mapping 75->0, 1->1, 2->2, 0->0, 143->143);bench/stream_check.py --self-testPASS;git diff --checkclean. The repository has no hosted CI workflow.