Dear PTFS team,
I don't understand "a scaling factor of 1.2 from one N to the next" Should I multiply each GFlops/s for each N by that factor?
Kind regards
Hassan Rady
Why should you multiply a Gflop/s number with 1.2? Think about the purpose of this benchmarking experiment.
This is just the usual way we set the problem size in order to get equidistant points on a logarithmic x axis. Length N = 10 * 1.2m, with going from 0 to whatever is necessary to get to N=108.
Okay. Thanks!