Matrix Multiplication Gets Faster

Enter matrix size N=1024 roll_your_own_multiply(). Elapsed time = 3.67529 sec DdotMultiply(). Elapsed time = 3.59084 sec Dgemv_multiply(). Elapsed time = 0.272707 sec Dgemm multiply(). Elapsed time = 0.0264843 sec

A terminal-style screenshot compares the runtime of several matrix multiplication methods for a 1024-sized matrix. Each successive method is dramatically faster, making the joke about an impressive acceleration in a programming lab or benchmark. It is shared as a reaction to unexpectedly effective optimization or rapidly improving performance. EN: A terminal-style screenshot compares the runtime of several matrix multiplication methods for a 1024-sized matrix. Each successive method is dramatically faster, making the joke about an impressive acceleration in a programming lab or benchmark. It is shared as a reaction to unexpectedly effective optimization or rapidly improving performance.

• @ithueti / 989 2019-06-06