Comment Comment by pseudopersonal With a unified 192GB memory, could this challenge nvidia's lock on ML training? Or is CUDA still insurmountable? parent pseudopersonal 3 years ago
