I am trying to use PAPI to measure FLOPs of my code. Unfortunately, all of my available resources
are based on Intel Sandy Bridge.
What I learn from here (http://icl.cs.utk.edu/projects/papi/wik ... SandyFlops) is that
the flop counts from PAPI on Intel Sandy Bridge have serious overcounting problem and thus I cannot
use flop counts and rates on Sandy Bridge as an absolute measure of the amount of work done.
So my question is will this problem (counting flops on Sandy Bridge) be fixed in the coming updates?
I think PAPI 5.3 still has this problem. Or is there any other way to measure the flops on Sandy Bridge?
Now I can only rely on my manually calculated results:-(