3 ms·
It's utilizing Apple ANE and probably other optimization tools provided by Apple's framework. Not sure if llama.cpp uses them, but if they're not then the bench
by worldsavior 1y ago
It's utilizing Apple ANE and probably other optimization tools provided by Apple's framework. Not sure if llama.cpp uses them, but if they're not then the benchmark on GitHub says it all.