github llama.cpp · 11h ago · inference

llama.cpp b10756 restricts Vulkan shader_bfloat16 extension request

from English

The llama.cpp project released build b10756, which modifies Vulkan backend behavior to only request the VK_KHR_shader_bfloat16 extension when it is supported by the hardware.

This update addresses compatibility issues where requesting unsupported extensions could cause failures on certain GPUs. The release includes binaries for macOS (Apple Silicon and Intel), Linux (CPU, Vulkan, ROCm, OpenVINO, SYCL), Windows (CPU, CUDA 12/13, Vulkan, OpenCL, ROCm, SYCL), Android, and iOS.

The change ensures that Vulkan-based inference runs correctly on devices that do not support the bfloat16 shader extension.

Importance 1/3 Trust 1/3 llama.cpp Hugging Face Inference efficiency

Read original

github llama.cpp · just now Live

llama.cpp b10766 adds DeepSeek 4 vision support

llama.cpp b10762 adds DeepSeek-V4-Flash-Vision-Exp support

llama.cpp b10758 adds MUL_MAT and MUL_MAT_ID fusion for Hexagon

llama.cpp b10759 avoids KleidiAI buffer type init on dispatch

llama.cpp b10757 improves Vulkan IQ3_S mat-vec performance for larger batch sizes