mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2026-07-08 04:00:22 +00:00
f549b0007d
This lets the copy to the destination device use the host-visible vidmem optimization.