github llama.cpp · 1d ago · inference

llama.cpp b10955 fixes heap corruption in ggml-cpu

from English

The llama.cpp project released version b10955, which includes a critical fix for heap corruption within the ggml-cpu backend. The update disables the precompiled header (PCH) and removes an ambiguous CACHE_LINE_SIZE definition to resolve memory safety issues.

Disabling the ggml-cpu PCH restores natural include order, ensuring ops.h is processed before hardware interference headers.
Removing the std::hardware_destructive_interference_size branch makes CACHE_LINE_SIZE deterministic and independent of include order.
The fix resolves a mismatch where C++ kernels used 256-byte cache lines while C code used 64-byte fallbacks, which previously undersized rope work buffers.
This correction prevents heap-buffer-overflow crashes in ggml_compute_forward_rope_flt caused by the buffer size discrepancy.

The release provides binaries for macOS (Apple Silicon and Intel), Linux (x64, arm64, s390x), Windows (CPU, CUDA, Vulkan, ROCm, OpenVINO, SYCL), Android, and openEuler platforms.

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

Read original

github llama.cpp · 1h ago · 1 view Live

llama.cpp b10984 adds CUDA row-contiguous SUM_ROWS support

The llama.cpp project released version b10984, introducing CUDA support for row-contiguous SUM_ROWS operations. This update also organizes the code to include GGML_OP_MEAN for handling row-contiguous tensors via a shared kernel.

github llama.cpp · 2h ago · 1 view

llama.cpp b10985 restricts RPC hash-cache to weights

The llama.cpp project released build b10985, which modifies the RPC backend to limit hash-based caching exclusively to weight tensors. This change addresses an issue where activation data was unnecessarily hashed and stored, leading to excessive disk usage.

github llama.cpp · 4h ago · 2 views

llama.cpp b10981 fixes stateful decode for Gemma-4 and optimizes GPU MoE inference

The llama.cpp b10981 release addresses critical correctness issues in the OpenVINO backend's stateful decoding path, specifically fixing support for models like Gemma-4 that use per-layer head sizes. It also introduces optimizations for GPU Mixture of Experts (MoE) inference and improves performance through KV state relayouts.

github llama.cpp · 5h ago · 1 view

llama.cpp adds Metal FA kernels for MiniCPM3's HSK=96, HSV=64

The llama.cpp project has added Flash Attention (FA) kernel support for the specific configuration used by MiniCPM3 on Apple Silicon. This change addresses a crash where the `-fa auto` flag would abort due to a missing kernel instantiation for attention key length 96 and value length 64.

github llama.cpp · 19h ago · 3 views

llama.cpp b10970 release uses fp32 accumulation on MFMA devices

The llama.cpp project has released version b10970, which includes a technical update to the fattn-mma backend. This change implements the use of fp32 accumulators specifically for CDNA architecture devices.