b10042

ggml-org/llama.cppb10042Jul 16, 2026by github-actions[bot]same day as b10041

What changed since b10041

low risk

b10042 enables CUDA graph acceleration for Volta and Turing GPUs, a single performance-focused change over b10041.

New features

  • CUDA graph execution is now supported on Volta and Turing architecture GPUs, extending graph-based acceleration to older NVIDIA hardware.

Improvements

  • Faster CUDA inference on Volta/Turing cards thanks to graph capture reducing per-kernel launch overhead.

Upgrade notes

No configuration or migration steps are required; the change is automatic for users on Volta or Turing GPUs. If you observe instability or incorrect output on those architectures, you can fall back to b10041 or disable CUDA graphs. The CORS log-noise fix from b10041 remains included.

Analyzed Oct 31, 2026

Upgrading?

Release notes summary

Activates CUDA graph support for Volta and Turing architecture GPUs to improve performance.

Key Highlights

  • Enable CUDA graphs on Volta+Turing architectures
  • Performance improvement for supported GPUs

Release Notes (excerpt)

<details open> Enable CUDA graphs on volta+turing (#25749) </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) **Android:**…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories