Releases: mitsuharu/llama.cpp-windows-gpu-builder
Releases · mitsuharu/llama.cpp-windows-gpu-builder
Release list
llama-b10680-win-x64-rocm-10.0.0-gfx1201-no-peer-copy
Windows x64 ROCm 10.0.0 (gfx1201) binaries built from llama.cpp b10680. The ROCm runtime and device kernels are bundled; peer copy is disabled.
llama-b10680-win-x64-rocm-7.14.0-gfx1201-no-peer-copy
Windows x64 ROCm 7.14.0 (gfx1201) binaries built from llama.cpp b10680. The ROCm runtime and device kernels are bundled; peer copy is disabled.
llama-b10664-win-x64-rocm-7.14.0-gfx1201-no-peer-copy
Windows x64 ROCm 7.14.0 (gfx1201) binaries built from llama.cpp b10664. The ROCm runtime and device kernels are bundled; peer copy is disabled.
llama-b10615-win-x64-rocm-7.14.0-gfx1201-no-peer-copy
Windows GPU binaries built from llama.cpp b10615. CUDA: 12.4; ROCm: 7.14.0; Vulkan SDK: 1.4.357.0. Peer copy is disabled in CUDA/ROCm builds.
llama-b10615-win-x64-rocm-7.14.0-gfx1200-no-peer-copy
Windows x64 ROCm 7.14.0 (gfx1200) binaries built from llama.cpp b10615. The ROCm runtime and device kernels are bundled; peer copy is disabled.