llama.cpp

Change Windows AMD example to release build to make inference much faster.

#6525

Merged