GreenWolf13 Posted November 2, 2012 Share Posted November 2, 2012 To extend jakj's statement: Be careful with any sort of performance metric: There are so many measuring sticks out there, and not all of them measure equally on the same systems. For example, next-generation supercomputers using heterogeneous platforms (GPU/CPU pairs) can tout all the peta and zeta FLOPS they want. But that's an old metric back in the day when cpu floating point operations per unit time was the limiting factor. The current limiting factor for them is busing data between nodes and duplicating unnecessarily data in software (for 3-D problems, up to an order of magnitude worth more data and calculations). So I can have a faster FLOPS supercomputer that is actually slower because of the busing issue. Of course, if you have a computer like Watson, then FLOPS are the limiting factor. Right? Link to comment Share on other sites More sharing options...
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!Register a new account
Already have an account? Sign in here.Sign In Now