docs(fold): gates 4–6 on gfx1151 — the image, preflight, and gate 6 so far - #383
Merged
glennneuber merged 1 commit intoSep 25, 2026
Merged
Conversation
…o far The ROCm host's record, beside the CUDA section: - Gate 4: the fold image, with a b11081 payload whose structure is unchanged against 0.34.3's. - Gate 5: PASS=20 SKIP=12 with rocm7-0-34-4-dynres (#378). - Gate 6, think off and OCRBench: equal to production in every scored cell and item. - Gate 6, think on, under the aligned protocol (so far): - the loop probe's thinking is byte-identical between flows; - gemma4:31b shows no payload effect and flow noise only; - gemma4:26b has one pre-existing loop, byte-identical on b10969. Generator outputs are pasted verbatim. The status table's rows 4–6 and open item 5 now say where gfx1151 stands. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This PR adds the gfx1151 record of gates 4–6 to #375's task doc. It targets #375's branch, not
main. Everything in it was measured on gfx1151 only, and it touches no other platform's section.What
A new section, "Gates 4–6 on gfx1151 (2026-09-25)", next to the CUDA one in
docs/maxusai/tasks/upstream-sync-0.34.4.md. It also fills in the status table's rows 4–6 for gfx1151 and narrows open item 5 to what is still running.maxusai-ollama:0.34.3-dynres-5-g29ae523-rocm7-gfx1151: b11081, payload structure unchanged against 0.34.3's, native payload byte-identical to the ROCm cross-check. The ROCm 10 lane builds toorocm7-0-34-4-dynres(#378)The think-off, OCRBench and loop-probe tables are generator output pasted verbatim (
compare.sh,summarize_extbench.py,probe_readout.py).Verification
check_source_paths.py --changed-sincethe fold head: every referenced file resolves.check_no_names.pywith a local deny-list file, over the tree and this branch's commit: no match.amd-server/rocm-gfx1151🤖 Generated with Claude Code