b10175

ggml-org/llama.cppb10175Jul 29, 2026by github-actions[bot]same day as b10174

What changed since b10174

low risk

Adds RDNA3.5 and RDNA3 to the mmq configuration sets so AMD GPU kernels can be tuned independently.

New features

  • RDNA3.5 is now a distinct mmq config entry, enabling architecture-specific kernel tuning.
  • RDNA3 is now a distinct mmq config entry, enabling architecture-specific kernel tuning.

Improvements

  • AMD RDNA3/RDNA3.5 GPUs no longer share a single mmq tuning profile, allowing per-architecture parameter selection.

Upgrade notes

This is a build-time tuning change with no config or CLI impact; simply rebuild or pull the new binaries. Users on RDNA3/RDNA3.5 hardware may see slightly different quantization kernel selection, but no migration is required.

Analyzed Nov 1, 2026

Upgrading?

Release notes summary

Added RDNA3.5 and RDNA3 to the multi-matrix quantization (mmq) configuration sets, allowing for independent tuning of these AMD architectures.

Key Highlights

  • Added RDNA3.5 and RDNA3 to mmq configs
  • Enabled independent tuning for these architectures

Release Notes (excerpt)

<details open> add rdna3.5, and 3 to mmq configs so they can be tuned independently. (#26199) </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)…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories