b9984

ggml-org/llama.cppb9984Jul 13, 2026by github-actions[bot]same day as b9982

What changed since b9982

low risk

Adds Q2_K quantization support to the SYCL DMMV reorder path for Intel oneAPI hardware.

New features

  • Q2_K quantized models can now use the SYCL DMMV reorder path on Intel oneAPI hardware.

Improvements

  • Expands the SYCL DMMV reorder path to cover the Q2_K format, improving performance and compatibility for Intel GPUs.

Upgrade notes

This is a narrow backend change affecting only SYCL builds; no config or API changes are required. Users on Intel oneAPI hardware running Q2_K models should see the benefit automatically after rebuilding or pulling the new binaries.

Analyzed Oct 24, 2026

Upgrading?

Release notes summary

Adds support for Q2_K quantization format to the SYCL (Intel oneAPI) DMMV (Dynamic Multi-Vector) reordering path, improving performance and compatibility for Intel hardware.

Key Highlights

  • SYCL Q2_K support
  • Intel oneAPI backend improvement
  • DMMV reordering path expansion

Release Notes (excerpt)

<details open> sycl: add Q2_K to DMMV reorder path (#25064) Signed-off-by: Todd Malsbary <todd.malsbary@intel.com> </details> **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