Occus benchmark performance between 32 and 64 bit OS...
I was running some benchmark numbers on a new HP Z600 workstation specification we are looking at to replace some units coming off lease. The system has an i7 Xeon X5550, 6 gig of ram, and Nvidia Quadro FX3500 graphics card. The system has no software other than the OS and Wildfire 4 and all the specific direction for the Occus benchmark were followed. Initially, we installed a 32 bit Windows XP operating system and ran the 32 bit benchmark on a 32 bit installation of Wildfire 4 build M110, with the following results: cpu = 497 graphics = 665 disk = 76 total = 1163 THEN, we rebuilt it with a 64 bit Windows XP operating system, downloading the appropriate latest 64 bit graphics driver and ran the 32 bit benchmark again on a 64 bit installation of Wildfire 4 build M130, with the following results: cpu = 518 (4% worse) graphics = 1116 (68% worse) disk = 78 (2% worse) total = 1635 (41% worse) This definitely surprised us. We would have expected roughly equal performance. The graphics is the real culprit here, but I have no idea why???? COULD it be the difference in build dates; M110 vs M130? Seems very odd... Any thoughts? Comments? Suggestions? For lack of anything better, I will download the 64 bit WF4 M110 build just to see if that makes a difference... Thanks... Paul Korenkiewicz FEV, Inc. 4554 Glenmeade Auburn Hills, MI., 48326
Could it be that you're running a 32-bit benchmark on a 64-bit system? Is there a 64-bit version of the benchmark that you can run?
I've noticed that 64-bit OS and drivers are not as mature or well maintained as the 32-bit counterparts. My Dell Precision M4400 running Windows XP Pro 64-bit has been quite unstable compared to my prior Dell Precision M65 running Windows XP Pro 32-bit. I believe that the only reason to go with 64-bit is to break through the 4GB RAM barrier if you have large/heavy assemblies open. Otherwise, you're probably better of going with a 32-bit system if possible for better support and compatibility.