Overall 3DMark05 score was:
5197 ; as above
, when it ran under WinXP x64 a weeks or so ago. (Check link on ORB for date).
Just ran it in Win32 and only got:
4078 ; but underclocking the card to 95% of stock (380/465) as it has issues at (400/490) stock speeds, possibly after a overclocking misadventure it was damaged a little. CPU Score ('05) just then was
5895 under Windows XP Pro (x86 - 32 bit). But 3DMark05 only used 2 cores anyway
SiSoft SANDRA shows 4-way systems in their real light.
Little surprised it scores so lowly, as only 95% of stock speeds, would expect to score: 4937 ; give or take.
CPU usage during the test was: 1 core at 100% for first half (0% - 50%), then 1 and a half cores for 50% - 75% of the test, then just shy of 2 cores for 75% - 100% completion of test. This would be reflected the in CPU Score no doubt.
Under
3DMark06:
3DMarks:
1271
SM2.0 Score:
593
CPU Score:
2649
Compare URL:
http://service.futuremark.com/compare?3dm06=84438
Bear in mind my Radeon X800XL was clocked at 380/465 (not 400/490)
Test was performed in WinXP Pro
First 80% of the test used only 1 core, the last 2 tests used all 4 cores at about 97% load each, the CPU Score above for 3DMark06 will most likely reflect this (I would hope anyway).
I think some of my earlier tests on ORB (back in 400/490) where with Quality texture filtering, instead of High Quality texture filtering forced in the driver. This would have inflated the score back then perhaps. (?). Could always run with High Performance filtering forced in the driver and see what it scores.
Update: Also only running Catalyst 5.8 in WinXP 32, so I'll update some stuff and check if results improve.
I am not expert on 3DMark btw, haven't used it seriously since 3DMark 2003SE. More into SiSoft SANDRA myself, and trying to develop software for 4-way to 32-way systems. (ie: The concept of FutureProof software, with seperate code paths for various systems, aswell as seperate EM64T vs AMD64 paths, each compiler optimized
).