llama.cpp Released Build b12020 with OpenCL FWHT Support
Release b12020 of llama.cpp introduces OpenCL support for Fast Walsh-Hadamard Transform (FWHT), along with standard prebuilt binaries across multiple platforms. This update expands hardware acceleration capabilities for devices utilizing OpenCL, improving performance consistency across various hardware backends. The release includes prebuilt binaries for macOS, Linux (supporting various CUDA, Vulkan, and SYCL options), Android, and Windows, alongside specific Snapdragon device optimization assets.
## BACKGROUND
llama.cpp is an open-source software library created by Georgi Gerganov that performs local inference on large language models using the GGML tensor library. OpenCL is an open, royalty-free standard for cross-platform parallel programming of diverse accelerators found in CPUs, GPUs, and other processors.