Update the pinned llama.cpp build to b10507 - #13892
Merged
Merged
Conversation
Bumps LlamaCppDownloadService.ReleaseTag from b10310 and refreshes all seventeen DownloadHashManager.LlamaCpp lists (twelve archives plus the five unpacked llama-server keys). The two cudart redistributables are byte-identical to the previous releases, so their entries repeat the known hash. Archive hashes come from the release assets' SHA-256 digests, spot-checked against locally downloaded copies; the llama-server hashes are taken from the extracted binaries, and the method was verified by reproducing b10310's committed Windows and macOS ARM64 hashes first. The Windows llama-server.exe is still identical across CPU/Vulkan builds, as is the Linux one, so the single-key-per-OS assumption in ResolveLlamaCppExecutableKey still holds. Smoke-tested on macOS ARM64 with the flags LlamaCppServerManager passes: TranslateGemma 4B translation (--no-jinja --chat-template gemma) and the multimodal OCR path (LightOnOCR + mmproj) both work. Co-Authored-By: Claude Opus 5 <[email protected]>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Bumps the pinned llama.cpp release from b10310 to b10507 (latest, published 2026-08-20).
LlamaCppDownloadService.ReleaseTag→b10507DownloadHashManager.LlamaCpplists refreshed: twelve archives plus the five unpackedllama-serverkeys. The twocudart-*.zipruntimes are byte-identical to the previous releases, so their new entries repeat the known hash.Verification
llama-serverhashes are from the extracted binaries. The extraction method was validated first by reproducing b10310's committed Windows and macOS ARM64 hashes.ResolveLlamaCppExecutableKeystill holds: the b10507llama-server.exeis identical in the CPU and Vulkan zips, and the Linuxllama-serveris identical in the CPU and Vulkan tarballs (x64 and arm64).libggml*.0.18.1→0.20.2,libllama.0.0.10310→0.1.2).Smoke test (macOS ARM64, real b10507 binary, flags
LlamaCppServerManagerpasses)-ngl 99 -c 8192 -np 1 --swa-full --no-jinja --chat-template gemma→ correct Danish output.--mmproj→ read the text off a subtitle image correctly.dotnet testfilter~LlamaCpp: 17 passed.🤖 Generated with Claude Code