Re:Improving Supercomputing Accuracy by Sacrificing Precision
by privong on Wednesday November 09, 2016 @08:36PM
(#12916894)
What about doing something like the inverse? Top500 for algorithms on a fixed hardware configuration. You have an input dataset and an answer (perhaps something with an analytic solution, for sake of comparison). That would let you compare the algorithm more directly.
[ Parent
| on HN ]
There are no comments on this comment yet.