Sign in with
Sign up | Sign in
Your question

5870 Matrix not pushing potential?

Last response: in Graphics & Displays
Share
July 28, 2010 4:04:08 AM

I just built my new rig today;

i7 930
6GB Corsair Dominator
2 X-25 RAID 0 for OS
640GB Caviar Black

and this;

http://www.newegg.com/Product/Product.aspx?Item=N82E168...


I was somewhat displeased when I ran a benchmark on the Heaven Unigine and didn't get the results I expected or I could just be plain old wrong. I installed the newest drivers off of ATI's site and my FPS doesn't seem to be that great. Of course i'm trying to push running the benchmark on max everything but it still seems somewhat low. Any suggestions?
July 28, 2010 4:13:32 AM

notrace said:
I just built my new rig today;

i7 930
6GB Corsair Dominator
2 X-25 RAID 0 for OS
640GB Caviar Black

and this;

http://www.newegg.com/Product/Product.aspx?Item=N82E168...


I was somewhat displeased when I ran a benchmark on the Heaven Unigine and didn't get the results I expected or I could just be plain old wrong. I installed the newest drivers off of ATI's site and my FPS doesn't seem to be that great. Of course i'm trying to push running the benchmark on max everything but it still seems somewhat low. Any suggestions?



Well one problem straight out I can think of would be that the uniengine runs tesselation like crazy, its really showcasing what tesselation is capable of. Now the fermi's were built around this and were made to perform well with tesselation. Now correct me if im wrong but this isnt exactly ati's strongsuit. Plus i dont know if uniengine is optimized for dual gpu cards such as your new mars.. I could be wrong but i think those are the main reasons for reduced performance. I bet your system will be great in actual games though!!! :)  :)  :)  :) 
m
0
l
a c 153 U Graphics card
July 28, 2010 4:13:36 AM

What kind of results where expectings? What was your score and settings you used?

Hes doesn't have a Mars by the way, what were you talking about?



I found a score from someone not me, who has the same GPU.


FPS: 27.4
Scores: 689

Q6600 @ 3.2
XFX 5870
Asus P5K-e
Vista x64
4GB 800mhz Ram

Render: direct3d11
Mode: 1920x1200 4xAA fullscreen
Shaders: high
Textures: high
Filter: trilinear
Anisotropy: 16x
Occlusion: enabled
Refraction: enabled
Volumetric: enabled
m
0
l
Related resources
a b U Graphics card
July 28, 2010 5:03:06 AM

Quote:
Try some games.ati 5800 series card are very poor in tessletation.U should have brought a nvidia card if u want good performance in tessletation


so, he should have bought Nvidia just to get better results in a single synthetic benchmark?

dont sweat it, Unigine is heavy on the tess, more so than any game currently on the market (or likely to be for a long while)

though it would help if you share your scores with us, so we can tell you if there is indeed a problem. im sure if you run a benchmark like COP, crysis or vantage you will get much more impressive scores.

m
0
l
a c 153 U Graphics card
July 28, 2010 5:30:37 AM

Quote:
Huh fanboy alert.


How is that in any way shape or form a fanboy alert? It was very informative in fact.

Nvidia will perform better on tesselation benches, but tesselation isn't used to that degree in todays games.

To tell someone they should have bought Nvidia becuase of that (and the results of one benchtest), now that fanboying.
m
0
l
a b U Graphics card
July 28, 2010 5:32:09 AM

+1

thanks helltech. It's not often i get called a fanboy, but it usually makes me laugh :lol: 

m
0
l
a b U Graphics card
July 28, 2010 7:29:29 AM

I'm pretty sure Nvidia can make a GTS250 look better in some benchmarks than a HD5970.
m
0
l
July 28, 2010 4:27:30 PM

I did two tests this morning;

Unigine

Heaven Benchmark v2.1

FPS:
15.2
Scores:
383
Min FPS:
2.6
Max FPS:
37.0
Hardware

Binary:
Windows 32bit Visual C++ 1500 Release May 21 2010
Operating system:
Windows 7 (build 7600) 64bit
CPU model:
Intel(R) Core(TM) i7 CPU 930 @ 2.80GHz
CPU flags:
2806MHz MMX SSE SSE2 SSE3 SSSE3 SSE41 SSE42 HTT
GPU model:
ASUS EAH5870 Series 8.702.0.0 2048Mb
Settings

Render:
direct3d11
Mode:
1920x1200 8xAA fullscreen
Shaders:
high
Textures:
high
Filter:
trilinear
Anisotropy:
16x
Occlusion:
enabled
Refraction:
enabled
Volumetric:
enabled
Replication:
disabled
Tessellation:
extreme

----------------------------

Unigine

Heaven Benchmark v2.1

FPS:
23.1
Scores:
583
Min FPS:
9.4
Max FPS:
41.3
Hardware

Binary:
Windows 32bit Visual C++ 1500 Release May 21 2010
Operating system:
Windows 7 (build 7600) 64bit
CPU model:
Intel(R) Core(TM) i7 CPU 930 @ 2.80GHz
CPU flags:
2806MHz MMX SSE SSE2 SSE3 SSSE3 SSE41 SSE42 HTT
GPU model:
ASUS EAH5870 Series 8.702.0.0 2048Mb
Settings

Render:
direct3d11
Mode:
1920x1200 8xAA fullscreen
Shaders:
high
Textures:
high
Filter:
trilinear
Anisotropy:
16x
Occlusion:
enabled
Refraction:
enabled
Volumetric:
enabled
Replication:
disabled
Tessellation:
normal

------------------------

Indeed, drastic frame rate changes with lower Tessellation, but I'm still throughly satisfied. Any other suggestions?
m
0
l
July 28, 2010 6:49:01 PM

Helltech said:
What kind of results where expectings? What was your score and settings you used?

Hes doesn't have a Mars by the way, what were you talking about?

I meant matrix, i was thinking about the mars for some reason haha!



I found a score from someone not me, who has the same GPU.


FPS: 27.4
Scores: 689

Q6600 @ 3.2
XFX 5870
Asus P5K-e
Vista x64
4GB 800mhz Ram

Render: direct3d11
Mode: 1920x1200 4xAA fullscreen
Shaders: high
Textures: high
Filter: trilinear
Anisotropy: 16x
Occlusion: enabled
Refraction: enabled
Volumetric: enabled

m
0
l
a b U Graphics card
July 28, 2010 7:09:51 PM

when the heaven benchmark came out, it was much less tessellation intensive compared to the new 2.xxx or great version i'm sure in part because of the fermi release, which it's tessellation power is far greater the the 5xxx series

but it's power doesn't really matter, tessellation is a rare feature, and so far in the game that use it is only used in a few limited instances, metro 2033 uses it somewhat, but personally the visual difference is very little, in other titles like dirt 2 and Avp it's really just a marketing stunt to draw people to games more then they maybe normally would, because they can say look unleash your new dx 11 card and use tessellation

in any game in your collection unless you have unreasonable expectation you should be more than satisfied
m
0
l
a b U Graphics card
July 28, 2010 7:48:40 PM

Quote:
Huh fanboy alert.


not fanboy alert, just a user correcting your false information ;) 
m
0
l
a b U Graphics card
July 28, 2010 8:09:52 PM

Here is what I came up with :

x4 @ 4Ghz / 5850 @ 1000/1300 w/ Catalyst 10.4's

Avg FPS = 33
Min FPS = 14.1
Max FPS = 67.2

Score = 937

Render:
direct3d11
Mode:
1680x 1050 8xAA fullscreen
Shaders:
high
Textures:
high
Filter:
trilinear
Anisotropy:
16x
Occlusion:
enabled
Refraction:
enabled
Volumetric:
enabled
Replication:
disabled
Tessellation:
normal

@ OP, if you want a higher score you will need to bump up your CPU clock, a stock clocked CPU is not going to give you the results you are looking for when it comes to this specific benchmark.
m
0
l
July 28, 2010 8:28:05 PM

OvrClkr said:
Here is what I came up with :

x4 @ 4Ghz / 5850 @ 1000/1300 w/ Catalyst 10.4's

Avg FPS = 33
Min FPS = 14.1
Max FPS = 67.2

Score = 937

Render:
direct3d11
Mode:
1680x 1050 8xAA fullscreen
Shaders:
high
Textures:
high
Filter:
trilinear
Anisotropy:
16x
Occlusion:
enabled
Refraction:
enabled
Volumetric:
enabled
Replication:
disabled
Tessellation:
normal

@ OP, if you want a higher score you will need to bump up your CPU clock, a stock clocked CPU is not going to give you the results you are looking for when it comes to this specific benchmark.


Thanks for the info mate.

I heard you were able to OC a 930 up to around 3.2 on the stock cooler, is that true?
m
0
l
July 28, 2010 8:31:12 PM

jonnyboyC said:
when the heaven benchmark came out, it was much less tessellation intensive compared to the new 2.xxx or great version i'm sure in part because of the fermi release, which it's tessellation power is far greater the the 5xxx series

but it's power doesn't really matter, tessellation is a rare feature, and so far in the game that use it is only used in a few limited instances, metro 2033 uses it somewhat, but personally the visual difference is very little, in other titles like dirt 2 and Avp it's really just a marketing stunt to draw people to games more then they maybe normally would, because they can say look unleash your new dx 11 card and use tessellation

in any game in your collection unless you have unreasonable expectation you should be more than satisfied


After playing a bunch of other games I'm still satisfied. But i still believed this card was faster than what it's running at. Or maybe im being over ambitious.

If you have any more ideas I'd love to hear em :-) and thanks for the aid
m
0
l
July 28, 2010 8:33:06 PM

This particular benchmark does rely on your CPU speed a good bit, OC and you will see an increase.

You should also check what filter your AA is using. If you've got it on Edge-Detect at 8x, that's actually doing 24x. That would cause a huge performance hit -- try using different filters and see what your results are.

The original benchmark, version 1.0 I believe had the equivalent of the "normal" tessellation setting. The high/extreme settings were added in 2.0 specifically to showcase what Nvidia's Fermi could do. And truly, they will achieve better scores using those methods -- however, it's completely useless since no games utilize tessellation even to the "normal" level (No, not even Metro 2033).
m
0
l
a b U Graphics card
July 28, 2010 8:33:48 PM

notrace said:
Thanks for the info mate.

I heard you were able to OC a 930 up to around 3.2 on the stock cooler, is that true?


well it mostly depends on your ambient temp but for the most part yes.

get an aftermarket cooler and crank that up that CPU, the 930 is a monster at high clocks ;) 
m
0
l
a b U Graphics card
July 29, 2010 4:03:48 AM

actually, your results do seem low.

here are my results from a 5870:



i would certainly expect your GPU do be keeping up with it, not that far behind. Could be a CPU bottleneck, but erven at stock an i7 should be sufficient.

you have the latest cat drivers? a good PSU?
m
0
l
a b U Graphics card
July 29, 2010 6:35:18 AM

huh?

I lowered the AA to 2x and got a better overall numbers?



this does not make sense.



m
0
l
a b U Graphics card
July 29, 2010 7:13:34 AM

bear in mind, i did this benchmark a while back. probably cat 10.3, if not 9.4.

mainly posted it to illustrate that the OP should be getting better scores.
m
0
l
a b U Graphics card
July 29, 2010 8:02:12 AM

welshmousepk said:
bear in mind, i did this benchmark a while back. probably cat 10.3, if not 9.4.

mainly posted it to illustrate that the OP should be getting better scores.


i was referring to my previous bench ;) 

i raised the AA and managed a better score, normally when you raise the AA it gives you a lower score, not a higher one.
m
0
l
a b U Graphics card
July 29, 2010 8:35:46 AM

oh i see, ignore me then.

did you update the drivers in the interim?

m
0
l
a b U Graphics card
July 29, 2010 3:44:01 PM

welshmousepk said:
oh i see, ignore me then.

did you update the drivers in the interim?


nope, im going to try the 10.7's again today and re-test.
m
0
l
!