4 ms·Go?by kelvinjps10 2y agoGo?mountainriver 2y agoI do a lot of ML, and the Go memory model is highly problematic. The Go to C FFI is too slow and doesn't interface with CUDA well.