Low 3dMark06 Score? (Radeon HD 5770)

[happy medium]

Good god man its just your CPU. Other people are mucking it up in here, but if you're running at 1280x1024, the standard "free" res, your 5770 is going to be bottlenecked not only in the CPU tests, but in the SM tests as well, returning abnormally low scores in all 3 subsections.

I mean, really, the CPU isn't only used in the "CPU" test, its used in all of them and at such a low resolution its going to severely cripple the card's overall performance.

That dosen't explain his low sm2 ,sm3 scores.

 

[GodisanAtheist]

Sure it does. When you run an actual game at a low resolution with a fast GPU, what happens? You get frame capped by your processor. Same thing is happening here. All 3dMark does is look at the frame rate stats of a particular run to pump out a score. If his CPU isn't feeding his GPU during SM 2 & 3 runs, his FPS is going to suffer and so is his score. 3dMark effectively thinks that his GPU is crap despite the fact that its being held back by his processor.

Trust me on this one. When I upgraded from my 8800 GTS 320 to my Radeon HD4850 on an A64 3800x2 @ 2.7ghz system, I ran into the EXACT SAME PROBLEM.

The 8800 would score a rather comfortable 8000 3d marks, which wasn't bad for the card. My 4850 ALSO scored 8000 3d marks, which was REALLY bad for that card. After a short "lol ATI drivers" and "I should've stayed green" tard phase I upgraded to an e7300 system which I clocked to 3.2 Ghz and immediately got a MUCH more respectable 12,000 score, with both my SM scores increasing by 1500 points in addition to a sweet 1000 point increase in my CPU score.

Its his CPU.

 

[happy medium]

OK, I will down clock my cpu and confirm this for the op.
I think I can get her down under 2.0 ghz.
I will run 3d Mark and check sm2,sm3 scores.

Edit: 3d mark 06,

q9950 @ 3.6
overall, 16,947
Sm2, 6,545
Sm3, 7725
Cpu 5,284

http://service.futuremark.com/home.action;jsessionid=4AF5D55CD302759821A0D1AB15E02523

Edit 2: q9550 @ stock 2.8

 

[toyota]

OK, I will down clock my cpu and confirm this for the op.
I think I can get her down under 2.0 ghz.
I will run 3d Mark and check sm2,sm3 scores.

Edit: 3d mark 06,

q9950 @ 3.6
overall, 16,947
Sm2, 6,545
Sm3, 7725
Cpu 5,284

http://service.futuremark.com/home.action;jsessionid=4AF5D55CD302759821A0D1AB15E02523

Edit 2: q9550 @ stock 2.8

lower your cpu to 1.8 and disable two cores and your cpu will be about like a 5000 X2.

 

[happy medium]

lower your cpu to 1.8 and disable two cores and your cpu will be about like a 5000 X2.

How do I disable 2 cores?

 

[toyota]

How do I disable 2 cores?

easiest way is by setting the affinity in task manager to just 2 processors for 3DMark.

 

[happy medium]

easiest way is by setting the affinity in task manager to just 2 processors for 3DMark.

ok I got it.

But I could only set the multy down to 333 x 6, so I'm running at 2.0 with 2 cores.

BRB gonna run it now.

 

[happy medium]

Drum roll please.................

Q9550 @ 3.6 5750 @ 900, 1350

16,947
Sm2 6545
Sm3 7725
cpu 5284

Q9550 @ 2.0 5750 @ 900, 1350

10,061
Sm2, 4341
Sm3, 5809
Cpu 1852

http://service.futuremark.com/home.action;jsessionid=E34051C0EA8FBC4A1B45756776BA7C09

Holy crap! thats the same score as the OP.
I guess we got our answer.

 

[jaggerwild]

You can't compare a Quad to a X2 core? Also an Intel will clock the snot out of MOST AMD duals.

At the Original poster 3DMARK06 is HEAVY CPU DEPENDENT, so unless you can over clock that CPU. Also the GPU scores will suffer for the same reason.

 

[happy medium]

You can't compare a Quad to a X2 core? Also an Intel will clock the snot out of MOST AMD duals.

At the Original poster 3DMARK06 is HEAVY CPU DEPENDENT, so unless you can over clock that CPU. Also the GPU scores will suffer for the same reason.

I disabled 2 cores and downclocked q9550 to 2.0 and got the same cpu score as the op.
It works.

 

[GodisanAtheist]

Thanks for the confirmation Happy. It always bugged the crap out of me seeing how my 3dMark scores never really matched those that reviewers got until it dawned on me that my value cores were really dragging performance down compared to the average reviewer's uber processor.

 

[happy medium]

Thanks for the confirmation Happy. It always bugged the crap out of me seeing how my 3dMark scores never really matched those that reviewers got until it dawned on me that my value cores were really dragging performance down compared to the average reviewer's uber processor.

And thank you, If you didn't post, I would not have ran it.
I don't like messing with my overclock.:twisted:

I can say one thing.... I could not wait to get my overclock back..Soooooo slow! hehehe.

 

[toyota]

And thank you, If you didn't post, I would not have ran it.
I don't like messing with my overclock.:twisted:

I can say one thing.... I could not wait to get my overclock back..Soooooo slow! hehehe.

remember that a core 2 duo at 2.0 is not a bottleneck for even a 5870 according to BFG10K.

 

[Hyperlite]

<<<---- gloating, a little.

OP should know though that older games (pre-2009 or so, crysis and stalker aside) should be able to make decent use of the 5750 despite the CPU. New stuff is gonna bring it to it's knees and put a metaphorical bullet in it's head though. OC that thing!!

 

[happy medium]

<<<---- gloating, a little.

OP should know though that older games (pre-2009 or so, crysis and stalker aside) should be able to make decent use of the 5750 despite the CPU. New stuff is gonna bring it to it's knees and put a metaphorical bullet in it's head though. OC that thing!!

I don't think he can its a Hp machine.
Mabe his board will take a better dual core?
I don't know much about HP motherboards.

 

[toyota]

I don't think he can its a Hp machine.
Mabe his board will take a better dual core?
I don't know much about HP motherboards.

yeah no way to safely oc that cpu. also it probably would not be worth the expense and trouble to fool with getting another cpu even if it would work.

oh well he at least he didnt get a really expensive gpu as that would have just been wasted with that cpu. I remember when I moved to this pc from a 5000 X2 and even with a wimpy 4670 I picked up 30% more performance at 1280 in Far Cry 2. lol

 

[BlackGS01]

Thanks fellas for going above and beyond. Guess it's my CPU after all. Time for a new Mobo and CPU when I get the cash.

 

[BlackGS01]

I don't think he can its a Hp machine.
Mabe his board will take a better dual core?
I don't know much about HP motherboards.

Max I can get up to with this mobo is a 5600+. Not really worth it I don't think.^_^

<<<---- gloating, a little.

OP should know though that older games (pre-2009 or so, crysis and stalker aside) should be able to make decent use of the 5750 despite the CPU. New stuff is gonna bring it to it's knees and put a metaphorical bullet in it's head though. OC that thing!!

It's actually a 5770

 

[toyota]

Max I can get up to with this mobo is a 5600+. Not really worth it I don't think.^_^



It's actually a 5770

yeah but with your cpu it might as well be a 5750.

 

[arredondo]

This thread is informative. My new 5770 card (auto-OC'd to 950/1210) in my Quad Core Q6600 rig:

3DMark Score 12237
SM 2.0 Score 4678
SM 3.0 Score 5968
CPU Score 3363

So I guess it follows the pattern mentioned here. A better processor using the same card is giving up some better scores.

 

[betasub]

So to re-iterate, 3dMark06 at default 1280x1024 is seriously CPU-dependent for most modern enthusiast systems. I have an official version with full options, and only by moving to higher resolutions and better AA does it shift to being less CPU-dependent, and more of a true graphics benchmark.