b10064

ggml-org/llama.cppb10064Jul 17, 2026by github-actions[bot]same day as b10063

What changed since b10063

low risk

OpenCL optimization improves q4_K memory access patterns

Improvements

  • Transpose q4_K noshuffle scales for coalesced reads
  • Improves memory access efficiency on OpenCL devices

Bug fixes

  • Fixes issue #25805

Upgrade notes

Update binaries to b10064 for OpenCL performance gains on q4_K models. No configuration changes required.

Analyzed Oct 16, 2026

Upgrading?

Release notes summary

OpenCL optimization for improved memory access patterns in q4_K quantization.

Key Highlights

  • Transpose q4_K noshuffle scales for coalesced reads
  • Improves memory access efficiency on OpenCL devices
  • Fixes issue #25805

Release Notes (excerpt)

<details open> opencl: transpose q4_K noshuffle scales for coalesced reads (#25805) </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