github llama.cpp · 4d ago · inference

llama.cpp b11047 fixes CUB argsort corruption on CUDA

from English

The llama.cpp project released version b11047, which includes a critical fix for CUB argsort corruption caused by in-place keys. The issue arose because `argsort_f32_i32_cuda_cub` called `DeviceRadixSort::SortPairs` with aliased input and output key buffers, causing CUB's internal double-buffer ping-pong to overwrite its own input mid-pass.

This bug resulted in corrupted permutations and intermittent garbage indices, particularly on Maxwell GPUs with CUDA 12.5 and CCCL 2.x, which subsequently triggered out-of-bounds gathers in downstream `get_rows` operations. The fix ensures that all six call sites (plain and segmented, ascending and descending, size-query and execute) use distinct keys-out buffers.

The release provides binaries for macOS, Linux, Android, Windows, and openEuler across CPU, CUDA, ROCm, Vulkan, OpenVINO, SYCL, and OpenCL backends.

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

Read original

github llama.cpp · just now · 1 view Live

llama.cpp b11122 extends MMVQ GLU fusion and adds new SYCL fusions

The llama.cpp project has released version b11122, which includes significant updates to the SYCL backend. The release extends the MMVQ GLU fusion to support mixed quantization types and introduces new fusions for rms_norm+scale and ssm_conv+silu operations.

github llama.cpp · 14h ago · 11 views

llama.cpp b11111 adds Intel Xe flash attention optimization kernels

The llama.cpp project released build b11111, which introduces Vulkan-based Intel Xe flash attention optimization kernels for split k paths on Xe-LPG Plus, Xe2, and Xe3 architectures.

github llama.cpp · 17h ago · 9 views

llama.cpp adds DFlash support for HunyuanOCR and fixes draft conversion

The llama.cpp project has added support for DFlash speculative decoding with the HunyuanOCR model by exposing layer input tensors in the target graph. This change allows the draft model to read residual streams, enabling image requests to run successfully with a draft acceptance rate of approximately 0.5 and byte-identical OCR output compared to non-speculative runs.

github llama.cpp · 20h ago · 11 views

llama.cpp b11101 release allows repeated find_package calls for llama

The llama.cpp project has released version b11101, which includes a build system fix allowing repeated find_package calls for the llama library.

lab Hugging Face Blog · 21h ago · 10 views

Transformers adds GGUF support via ggml kernels for local inference

Hugging Face's Transformers library now supports loading GGUF quantized models, leveraging the underlying ggml kernels to bring performance close to llama.cpp. This integration allows developers to run quantized checkpoints directly within the standard PyTorch-based API on supported hardware.