A follow-up pull request in the llama.cpp repository has optimized low-level CPU dot product operations for the q1_0 ...
Ordering coffee is easy. Besting the Starbucks app with AI chat is going to be very, very hard. Ordering coffee is easy.