b9852

ggml-org/llama.cppb9852Jul 1, 2026by github-actions[bot]same day as b9851

What changed since b9851

medium risk

b9852 adds initial q1_0 quantization support to the OpenCL backend, with Adreno-specific GEMM/GEMV kernels.

New features

  • OpenCL backend now supports the q1_0 quantization format for the first time.
  • Dedicated Adreno GEMM and GEMV kernels accelerate q1_0 inference on Qualcomm GPUs.
  • General (non-Adreno) OpenCL q1_0 path added for broader device coverage.

Improvements

  • Extends quantization coverage of the OpenCL backend beyond previously supported formats.
  • Adreno-targeted kernels should improve throughput for low-bit models on mobile GPUs.

Upgrade notes

No configuration or model-format migration is required; the change is additive and scoped to the OpenCL backend. q1_0 models will only benefit when running on OpenCL, and Adreno optimizations apply solely to Qualcomm GPUs. Other backends and existing quantization formats are unaffected.

Analyzed Oct 31, 2026

Upgrading?

Release notes summary

Introduces support for the q1_0 quantization format in the OpenCL backend, including specific optimizations for Adreno GPUs to accelerate GEMM and GEMV operations.

Key Highlights

  • Initial q1_0 support for OpenCL
  • Adreno GEMM/GEMV optimizations for q1_0
  • General OpenCL quantization support

Release Notes (excerpt)

<details open> opencl: initial q1_0 support (#25160) * opencl: general q1_0 support * opencl: add Adreno GEMM/GEMV for q1_0 </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