b9925
ggml-org/llama.cppb9925Jul 8, 2026by github-actions[bot]same day as b9924
What changed since b9924
Adds CUDA f16->f16 support for GGML_OP_SET_ROWS; otherwise identical to b9924.
New features
- CUDA backend now supports the f16->f16 variant of GGML_OP_SET_ROWS, enabling that operation to run on GPU instead of falling back.
Upgrade notes
Drop-in replacement for b9924 with no config or API changes. CUDA users working with f16 tensors and SET_ROWS may see improved throughput or fewer CPU fallbacks; other backends are unaffected.
Analyzed Oct 30, 2026
Upgrading?
Release notes summary
Adds support for f16->f16 GGML_OP_SET_ROWS operation on CUDA.
Key Highlights
- CUDA support for f16->f16 GGML_OP_SET_ROWS
Release Notes (excerpt)
<details open> cuda : add support for f16->f16 GGML_OP_SET_ROWS (#25367) </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) **Android:** - Android arm64 (CPU) **Windows:**…
Published by ggml-org on GitHub.
Read the full release notesRelated Repositories
:electron: Build cross-platform desktop apps with JavaScript, HTML, and CSS
Godot Engine – Multi-platform 2D and 3D game engine
ClickHouse® is a real-time analytics database management system
Self-hosted game stream host for Moonlight.
DuckDB is a high-performance analytical database system