2 ms·

Ampere has native acceleration (Ampere Optimised Frameworks) for AI workloads

• brucethemoose2 3y ago
  I dont think llama.cpp integrates this though. Accelerate is all but depreciated because of the Metal backend anyway.