Matrix Multiplications on GPUs Run Faster When Given Predictable Data

This was a surprisingly easy read for a non-computer science person. Thanks for sharing.