b10196

ggml-org/llama.cppb10196Jul 30, 2026by github-actions[bot]same day as b10195

What changed since b10195

low risk

Single-commit delta adds a synchronization barrier before clearing embedding sequences in llama-context.

Improvements

  • llama-context now waits for pending asynchronous copies to finish before clearing the embedding sequence state, tightening internal ordering guarantees.

Bug fixes

  • Prevents a race where in-flight async copies could touch embedding sequence data after it was cleared.

Upgrade notes

This is a one-commit increment over b10195 with no API, config, or file-format changes; a drop-in binary swap is sufficient. Rebuild from source only if you vendor llama.cpp, since the change is internal to llama-context.

Analyzed Oct 31, 2026

Upgrading?

Release notes summary

This release addresses a context synchronization issue by ensuring pending asynchronous copies are completed before clearing embedding sequences.

Key Highlights

  • Synchronized pending async copies in the context before clearing embedding sequences

New Features

  • Context management fix for async operations

Release Notes (excerpt)

<details open> llama-context : sync pending async copies before clearing embd_seq (#25676) </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - Ubuntu x64 (CPU) - Ubuntu arm64 (CPU) - Ubuntu s390x (CPU) - Ubuntu x64 (Vulkan) - Ubuntu arm64 (Vulkan) - Ubuntu x64 (ROCm 7.2) - Ubuntu x64 (OpenVINO) - Ubuntu x64 (SYCL FP32) - Ubuntu x64 (SYCL FP16)…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories