llama.cpp

855f5440 - Change Windows AMD example to release build to make inference much faster. (#6525)