Overclock.net banner
1 - 20 of 29 Posts

·
Premium Member
Joined
·
25,202 Posts
Discussion Starter · #1 ·
Synthetics and demos are done, sans Heaven with the CF set up, having issues getting it to complete.

I'll have in game runs completed tomorrow.
Couldn't get my LG monitor to work consistently so all tests are done at 1600*1200.

SLI clocked @ 850/2000/1700
CFX clocked @ 1000/1250

SLI score are top and left images, all settings the same.

Couple of facts:

SLI peak load was 1148w during Vantage Furmark, that's at 1.138v.
CFX peak load during Vantage Furmark was 988w @ 1.2875v.

I will graph the game play benchmarks, I'll only use fixed routes when possible.
BBC2 High Value target level
M2033 Trolley fight level.
Still deciding on the other games.























*Note Lost the RE5 shot, but CFX was 20FPS > SLI
 

·
Premium Member
Joined
·
25,202 Posts
Discussion Starter · #2 ·
Some game runs done.



Single card play, the 480 is head and shoulders above the single 5870, but in SLI not so much.
Game play is much smoother with CFX, maybe a driver issues for NV.



BC2 is a wash SLI vs CFX, single I give the slight edge to the 5870.



SLI owns M2033, the numbers may look close, but man the gameplay is infinitely better with SLI.









Any other requests, just let me know.
I don't plan on keeping the 480 much longer.
 

·
Registered
Joined
·
1,227 Posts
Nice results. Although I thought SLI 480s would perform much better than 5870s in CFX. Looking at these figures you would have to be stupid not to go for the 5870s
 

·
Registered
Joined
·
237 Posts
There is no such thing as a single decided number you can use to compare how fast many different video cards are at different games. The graphic card performance may vary wildly from one game engine to another. A graphicard can be faster than another at one game and the positions can be reversed with another game. Game performance do often vary between different versions of graphic card drivers. Graphic card creators tend to improve weak spots in their drivers over time so some cards get very different scores depending on whether it is tested just after release or six months later. Some cards are better at pixel shaders while some are being better at multitexturing. Some are better with full screen anti-aliasing while others are better with none.

You should really test these graphic cards with a wide vary of display settings, I am dissapointing right now. This is really not telling me which one that is performing the best.
 

·
Registered
Joined
·
1,832 Posts
Quote:


Originally Posted by HighOC
View Post

SLI peak load was 1048w during Vantage, that's at 1.138v.
CFX peak load during Vantage was 988w @ 1.2875v.

Its not a Big deal of power consumption as ppl say!!

So if it reaches 988w for CFX then does that mean I can't run 5870 CF?
 

·
Registered
Joined
·
2,626 Posts
wow looks like with OCing and all the driver updates lately xfire 5870's are finally catching up to sli 480s, didnt think that would happen! but yay!!

edit: quick question semi on topic for this thread... would a 5870+5970 scale better than 3 5870's?
 

·
OG AMD
Joined
·
8,966 Posts
Excellent work.Pretty consistant with single card results,no?
 

·
Registered
Joined
·
237 Posts
Sorry about the dissapointment, I can tell that these tests is only showing wich cards that is best in only anti-aliasing.

The results may differ or reverse dramatically if turning it off between these graphic cards.
 

·
Premium Member
Joined
·
32,942 Posts
Quote:


Originally Posted by NKNK
View Post

Sorry about the dissapointment, I can tell that these tests is only showing wich cards that is best in only anti-aliasing.

The results may differ or reverse dramatically if turning it off between these graphic cards.

If you happen to notice, ALL of the tests were done with Anti-Aliasing. And the two different setups traded blows with each other. The results will likely be identical (except with obviously slightly higher framerates) with AA turned off.

And I'm having a VERY hard time believing those power numbers
. Is that at the wall? If those are figures from the wall, then we also have to figure in power supply efficiency. You're likely not really drawing more than 700w or so from either setup from the PSU.
 

·
Registered
Joined
·
237 Posts
The reason why is that all was tested with anti-aliasing.

These graphic cards got too much vary on all chipsets.

You cannot know if the ''slowest'' graphic card on these tests is speeding away without anti-aliasing or not. '' It could be different settings than anti-aliasing only''

We simply don't know wich one that is the best, even when doing the same tests all over again.

We do only know wich one of these that is best performing with anti-aliasing with the current drivers.

But great information this far
 

·
I.T. Specialist III
Joined
·
4,679 Posts
How someone can look at this and not see that the 5870 is a better buy, is beyond me.
 

·
Registered
Joined
·
1,712 Posts
Quote:


Originally Posted by NKNK
View Post

There is no such thing as a single decided number you can use to compare how fast many different video cards are at different games. The graphic card performance may vary wildly from one game engine to another. A graphicard can be faster than another at one game and the positions can be reversed with another game. Game performance do often vary between different versions of graphic card drivers. Graphic card creators tend to improve weak spots in their drivers over time so some cards get very different scores depending on whether it is tested just after release or six months later. Some cards are better at pixel shaders while some are being better at multitexturing. Some are better with full screen anti-aliasing while others are better with none.

You should really test these graphic cards with a wide vary of display settings, I am dissapointing right now. This is really not telling me which one that is performing the best.


Way to tell us what we all already know
 

·
Premium Member
Joined
·
25,202 Posts
Discussion Starter · #19 ·
Quote:

Originally Posted by Giac View Post
Nice results. Although I thought SLI 480s would perform much better than 5870s in CFX. Looking at these figures you would have to be stupid not to go for the 5870s
There are a couple of where fermi will really flex its muscles, but as it stands now GTAIV is the only game where fermi actually provides a very noticeable improvement over Cypress.

Quote:

Originally Posted by NKNK View Post
There is no such thing as a single decided number you can use to compare how fast many different video cards are at different games. The graphic card performance may vary wildly from one game engine to another. A graphicard can be faster than another at one game and the positions can be reversed with another game. Game performance do often vary between different versions of graphic card drivers. Graphic card creators tend to improve weak spots in their drivers over time so some cards get very different scores depending on whether it is tested just after release or six months later. Some cards are better at pixel shaders while some are being better at multitexturing. Some are better with full screen anti-aliasing while others are better with none.

You should really test these graphic cards with a wide vary of display settings, I am dissapointing right now. This is really not telling me which one that is performing the best.
Sorry
I didn't buy $1700 worth of GFX cards to not run AA.
But I appreciate your input, and if there is a certain game/demo you'd like to see without AA, just let me know.

Quote:

Originally Posted by squall325 View Post
So if it reaches 988w for CFX then does that mean I can't run 5870 CF?

Be close, depends on your overclocks and voltages.

Quote:

Originally Posted by twich12 View Post
wow looks like with OCing and all the driver updates lately xfire 5870's are finally catching up to sli 480s, didnt think that would happen! but yay!!

edit: quick question semi on topic for this thread... would a 5870+5970 scale better than 3 5870's?
My 4870X2/4870 scaled nicely, but I've never ran 3 card config.

Quote:

Originally Posted by Redwoodz View Post
Excellent work.Pretty consistant with single card results,no?
Yes, save for a couple of instances I can't tell the difference in game play between cards.

GTAIV favors the 480, BBC2 favors the 5870

Quote:

Originally Posted by WildEast View Post
whats the CPU used?

EDIT: btw, 200$ difference between the 2 builds? weird lol
It shows in a couple of the screen shots, QX9650 @ 4250

Quote:

Originally Posted by pioneerisloud View Post
If you happen to notice, ALL of the tests were done with Anti-Aliasing. And the two different setups traded blows with each other. The results will likely be identical (except with obviously slightly higher framerates) with AA turned off.

And I'm having a VERY hard time believing those power numbers
. Is that at the wall? If those are figures from the wall, then we also have to figure in power supply efficiency. You're likely not really drawing more than 700w or so from either setup from the PSU.
Correct at the wall, I've seen the SLI hit 1200w+ on my Kill a watt, system shutdown of course.
 
1 - 20 of 29 Posts
Top