b10357

ggml-org/llama.cppb10357Aug 11, 2026by github-actions[bot]same day as b10356

What changed since b10356

low risk

b10357 adds an OpenCL Flash Attention prefill kernel optimization by transposing the K tile in local memory.

Improvements

  • OpenCL Flash Attention prefill kernels now transpose the K tile in local memory, improving throughput on OpenCL backends.

Upgrade notes

This is a single-commit delta over b10356 with no API, config, or build-system changes; OpenCL users should see faster FA prefill without any action. Users on other backends (CUDA, Metal, Vulkan, CPU) are unaffected and can upgrade purely for version alignment.

Analyzed Oct 30, 2026

Upgrading?

Release notes summary

Optimized OpenCL Flash Attention prefill kernels by transposing the K tile in local memory to improve performance.

Key Highlights

  • Optimized OpenCL FA prefill kernels
  • Transposed K tile in local memory

New Features

  • Performance improvement for OpenCL Flash Attention

Release Notes (excerpt)

<details open> opencl: transpose the K tile in local memory for FA prefill kernels (#26428) </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 7.14) - Ubuntu x64 (OpenVINO) - Ubuntu x64 (SYCL FP32) - Ubuntu x64 (SYCL FP16)…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories