On 4/13/10 3:59 PM, Erik Paulson wrote:I initially benchmarked the two machines using GeekBench. The old Opteron-based machine scored 3323, the new i7 machine scored 7971.
Is your benchmark:
A) The same program as your Condor job
B) Operating on the realistic data, ideally actual data?
There should be virtually no difference between running a program
through Condor on a machine and logging into that machine and running
it by hand. Anything other than that may or may not be a useful
predictor of performance.
After I put the new node into service, I had one of my users submit four identical jobs...two to the new node, two to an old node. The i7-based machine finished in ~100 hours, the Opteron-based machine finished in ~120 hours. This is obviously not the result I was expecting.
The Core i7 processor has a number of new technologies in it, and my initial thought was that one of these technologies may be causing problems. That's why I asked if anybody else is using i7's, just to see if others are getting similar results.
Chris
_____________________________________________
Chris Freemesser, University of Rochester
Department of Brain & Cognitive Sciences /
Center for Visual Science
Meliora Hall, Room 244
E-Mail: chris@xxxxxxxxxxxxxxxxx
Phone: (585)275-0786
_____________________________________________
_______________________________________________
Condor-users mailing list
To unsubscribe, send a message to condor-users-request@xxxxxxxxxxx with a
subject: Unsubscribe
You can also unsubscribe by visiting
https://lists.cs.wisc.edu/mailman/listinfo/condor-users
The archives can be found at:
https://lists.cs.wisc.edu/archive/condor-users/