85.3 GFlops: Optimizing FP32 Matrix Multiplication on a Single AMD Zen 3 Coregithub.com·93 pts·houslast·22