On Wed, 14 Nov 2007, Julien Sylvestre wrote: > I'm not getting stellar performances with the petsc linear solver on > a 64 bit Xeon (8 CPUs with 64 Gb RAM). The machine processors are > clocked at 3 GHz, but -log_summary tells me I'm running at 1e8 flops/s > (on a single processor; I don't see a big speedup with more processors, > but that's probably due to memory bandwidth). That's 30x slower than the > clock speed. Is that normal? Have other users seen this order of > magnitude difference between clock speed and flop/s on other systems? > I'm testing a system with ~100,000 DOFs, using the cg solver. > > I'd like to know if I should invest time in tuning my libraries/system, > or just give up and buy a better computer.
The difference between clock speed and flop/s depends very much on CPU architecture, the problem you're solving, and the software you're solving it with, and from what I've heard "the software" can vary wildly depending on how you've got PETSc built. Just as a point of comparison, we've also got a 3GHz Core2 Xeon (although running in 32 bit mode), and we've made no special effort to tune PETSc. Depending on problem and solver choices, on one core with under 100,000 DOFs PETSc's log_summary reports anywhere from 1.5e8-3.5e8 flop/s. If you do put effort into tuning and see a big difference in results, would you post the options that worked best for you to the list? Thanks, --- Roy ------------------------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now >> http://get.splunk.com/ _______________________________________________ Libmesh-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/libmesh-users
