Greg Edwards wrote:
I have access to a wide variety of MP Itanium2 machines, I'd be interested to trial a test.
I think the peak performace will be achieved if you run two (or more, depending on the number of CPUs) separate independent mprime/prime95 tasks on each machine. There is no point to run multi-threaded and/or multi-processor versions unless you want to get a quick answer for a particular number (as it was done for verifying M42). Running independent tests in parallel will result in better overall performance.
Max _______________________________________________ Prime mailing list [email protected] http://hogranch.com/mailman/listinfo/prime
