Sign in with
Sign up | Sign in
Your question

5870s low 3dmark06 scores

Last response: in Graphics & Displays
Share
February 19, 2010 4:36:42 PM

Ok, so I have 3 5870s in Crossfire with a PII 965 @ 4.01GHZ my score in 3dmark06 (all regular default settings) is 19437, with just 2 in crossfire was 19234, and with crossfire disabled, I was getting like 18450. THe rest of my system is a M3A79-T delux MOBO 8GB DDR2 Ram. Win7 64bit ultimate 4 OS paritions (xp, vista, 2 Win7) HArd drive on this parition is a 250GB ST2350310AS Seagate Barracuda.
February 19, 2010 4:52:00 PM

That's probably an error of the software itself. Are you running the paid version of the 3D Mark? I'm not sure if the freeware version detects muilti-GPUs.

I'd would not worry about that though, it's just a syntetich benchmark anyway, it means very little in real-gaming perfomance.
m
0
l
a b U Graphics card
February 19, 2010 4:55:58 PM

Freeware works just fine with multi GPU. I call CPU bottleneck on this one. My 3dmark06 score is nearly entirely dependent on my CPU clock, and with a powerful graphics setup (such as yours), you won't see much if any gain in 3dmark from the additional GPU horsepower. You should see benefits in games though.
m
0
l
Related resources
Can't find your answer ? Ask !
February 19, 2010 8:32:23 PM

cjl said:
Freeware works just fine with multi GPU. I call CPU bottleneck on this one. My 3dmark06 score is nearly entirely dependent on my CPU clock, and with a powerful graphics setup (such as yours), you won't see much if any gain in 3dmark from the additional GPU horsepower. You should see benefits in games though.


Do you think my Q9650@3.6Ghz is bottlenecking my results? I have a single 5870.

Thanks in advance.
m
0
l
a b U Graphics card
February 19, 2010 8:48:41 PM

Agree with whats been said, my experiences.
3dmark 06 tests at 1280x1024 -the test results seem to favor cpu hp. Its where my system with a i5750 and a 'intermediate' gpu, ati 4770 seems to score above average.
Where 3dmark Vantage , scores more relevant to current gaming resolutions/environment. IMO. Vantage should give you the higher score VS others with less graphic power.

http://www.futuremark.com/benchmarks/

3dmarkvantage/download/
http://www.futuremark.com/benchmarks/3dmarkvantage/down...
m
0
l
a b U Graphics card
February 19, 2010 10:10:48 PM

I've never managed to break 20k with any cards and my Phenom II at 3.8ghz. Best was 19700 and that was with crossfired 4770's. My 5850 got to about 19200 and my crossfired 5770's get to about 19500 also.

In summary, 3dmark06 sucks as an actual test of your graphics capability.
m
0
l
a b U Graphics card
February 20, 2010 12:31:58 AM

Absolutely agreed. For me, all of the tests (not just the CPU tests) scale perfectly with CPU clockspeed, and at stock clocks for my i7, I saw no gain from 1 4870x2 to 2 in 3dmark, despite noticeable gains in games.
m
0
l
a b U Graphics card
February 20, 2010 12:32:38 AM

MARSOC_Operator said:
Do you think my Q9650@3.6Ghz is bottlenecking my results? I have a single 5870.

Thanks in advance.

It may well be bottlenecking you in 3dmark. It's almost definitely not bottlenecking you in games.
m
0
l
February 20, 2010 3:11:48 AM

I did try it under Windows XP, and then Vista. got in the 22K with both (even though it shows under GPUZ that it's only using 2 cards in crossfire in XP) I think I'm going to try to uninstall directX and reinstall it in my win7 partitions
m
0
l
a b U Graphics card
February 20, 2010 7:39:12 AM

That score is fine. I wouldn't worry about it. You're almost definitely CPU limited in 3dmark - try actual game benchmarks (or at least try 3dmark vantage, which is less constrained by CPU).
m
0
l
February 20, 2010 11:30:33 AM

I'm getting this:

Overall: 20025
SM 2.0: 7597
SM 3.0: 9946
CPU: 5369
m
0
l
February 20, 2010 11:37:18 AM

Aren't the GPU tests on 06 single threaded also? When running I (personally) only see 1 thread being fully utilized with software monitoring. If this is true, I would not use 06 to benchmark.
m
0
l
February 20, 2010 12:30:56 PM

I plan on doing some comparison testing between game benchmarks in the next couple days on the different OSes to see what the difference is. The 2 issues I have is trying to figure out if there really is no benefit in having the 3rd card (might as well try to sell it and get some money out of it while it's worth something) and second, trying to find where the difference is between the OSes performance is at.
m
0
l
!