Repository navigation
chore(vendor): pin llama.cpp to the PrismML fork at 9a9394a8 - #25
Closed
ykhrustalev wants to merge 2 commits into
Closed
ykhrustalev wants to merge 2 commits into
ykhrustalev wants to merge 2 commits into
Conversation
Lets a plan target llamacpp_ios_pipette at github.com/PrismML-Eng/llama.cpp 9a9394a895b96003ca842a6041cb28ac49a108f7. The fork has no bNNNN tag, so the build stamps and submits the full sha, and descriptors name the fork's URL.
A 5.95 GB GGUF needed ~3.4 MB/s sustained to land inside 30 minutes, and a timed-out foreground pull keeps nothing, so every retry restarted from zero.
Member
Author
|
Closing: this branch existed to hand-sign and install a PrismML-fork iOS build on boston-iphone-17pro-1..5 (internal-pipette-farm#100, now closed). Those phones now run behind pipette-ios-proxy in version mode, which installs only builds the internal releases bucket publishes, so a branch archive cannot reach them. Reopen if fork builds get a publishing path. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem
Plans need an iOS build to target
llamacpp_ios_pipetteatgithub.com/PrismML-Eng/llama.cpp9a9394a895b96003ca842a6041cb28ac49a108f7withprivate_thermal = true. The submodule is shared, but Android still reports the upstream URL, so collect results only from the iOS archive this branch produces.Solution
vendor/llama.cppat the PrismML fork, pinned to9a9394a8vendor/llama.cpp.tag, because the fork'sprism-b*tags aren'tbNNNN, so the build reports as untaggedRuntime.thisBuildinstead of the upstream defaultTesting
ios/patches/001-ggml-metal-oom-nullcheck.patchapplies cleanly to the fork./ios/build-llama.sh simbuilds and stampsuntagged(commit: "9a9394a8…"); 911 iOS unit tests pass