tldr: The bottleneck is in the linear algebra libraries. Unoptimized Rust is 2x faster than unoptimized Python. Optimized Python is 2x faster than unoptimized

tldr: The bottleneck is in the linear algebra libraries. Unoptimized Rust is 2x faster than unoptimized Python. Optimized Python is 2x faster than unoptimized Rust because it can vectorize matrix multiplications, but Rust can't. My takeaway: stick to a GPU where everything is more parallel.