4 ms·
I really like these set of lecture notes for optimizing matrix multiplication: https://ppc.cs.aalto.fi/ch2/v7/ https://ppc.cs.aalto.fi/ch2/v7/ (The transpose tr
by gfd 5y ago
I really like these set of lecture notes for optimizing matrix multiplication: https://ppc.cs.aalto.fi/ch2/v7/ https://ppc.cs.aalto.fi/ch2/v7/ (The transpose trick is used in v1)
- progbits 5y agoThis deserves it's own submission, wonderful resource!
- eigenvalue 5y agoI find it surprising that, even after using all those tricks, they are still only to achieve around 50% of the theoretical peak performance of the chip in terms of GFLOPS. And that's for matrix multiplication, which is a nearly ideal case for these techniques.
- gfd 5y agoHmm where did you see that? In the graph at the bottom of v7, it got "93% of the theoretical maximum performance"