***8800 GTS vs 7950 GX2 benchmarks*** Inc OC results

[nib95]

Ok, well I finally got my 8800 GTS. Did a full comparison with it and my 7950 GX2 inc overclocking results from both cards. Click on the links to see screenshots, click on the image again to view it full screen.

BFG 7950 GX2 - Xtreme G 92.91 drivers

XFX 8800 GTS - Nvidia Forceware 97.02 drivers

Here are some screenies of the cards. Both are about the same length, and near enough the same height. Though in reality the GX2 is the teeniest bit longer.

The beasts

Beasts 2

Also, a little side note. On the 8800 GTS 8xQAA no sampling takes up near enough juice as 4xAA with Transparency Super Sampled AA. However imo 4xAA with TRSSAA looks a lot better and smoother, so that is the optimum setting I went with during these benchmarks. But in every comparison the GX2 and GTS are running at exactly equal settings as to be fair.

All game benchmarks were done at a resolution of 1920 x 1200


Stock speed comparison: (both cards at default speeds)


Half Life 2: Episode 1

1.)

7950 max settings (4x AA) 40fps

8800 GTS max settings (4x AA) 138fps

7950 GX2 max settings (4xTRSSAA) 29fps

GTS max settings (4xTRSSAA) 94fps

GTS max settings (8xQA TRSSAA) 46fps

2.)

GX2 max settings (4xAA) 43fps

GTS max settings (4xAA) 154fps

GX2 max settings (4xTRSSAA) 32fps

GTS max settings (4xTRSSAA) 112fps

GTS max settings (8xQA TRSSAA) 59fps

3.)

GX2 max settings (4xAA) 48fps

GTS max settings (4xAA) 123fps

GX2 max settings (4xTRSSAA) 46fps

GTS max settings (4xTRSSAA) 119fps



Eldar Scrolls IV: Oblivion

1.)

GX2 max settings HDR no AA (city) 72fps

GTS max settings HDR no AA (city) 92fps

GTS max settings HDR + 4xTRSSAA (city) 63fps

2.)

GX2 max settings HDR no AA (foilage) 43fps

GTS max settings HDR no AA (foilage) 53fps

GTS max settings HDR + 4xTRSSAA 39fps

3.)

GX2 max settings HDR no AA (outdoor landscape) 40fps

GTS max settings HDR no AA (outdoor landscape) 50fps

GTS max settings HDR + 4xTRSSAA (outdoor landscape) 34fps


Call of Duty 2 (SLI optimisations for GX2 turned ON)

GX2 max settings no AA - 26fps

GTS max settings no AA - 53fps

GX2 max settings 4xAA - 23fps

GTS max settings 4xAA - 45fps

GX2 max settings 4xTRSSAA - 20fps

GTS max settings 4xTRSSAA - 39fps


FEAR

GX2 max settings no AA no AF no Soft Shadows - 103fps

GTS max settings no AA no AF no Soft Shadows - 88fps

GX2 max settings 4xTRSSAA 16xAF Soft shadows ON - 58fps

GTS max settings 4x TRSSAA 16xAF Soft shadows ON - 49fps


Company of Heros

I believe there must have been some driver issues with my GX2 as I was getting dreadful results. I did tests on both cards with the following settings at 1690 x 1050, though I'm sure the GTS would have easily done 1920 x 1200 with max settings.

Settings

GX2 - 16.5fps average

GTS - 96.4fps average





OVERCLOCKING RESULTS

Ok, first off let me start by saying, both cards had considerable performance boosts with overclocking. So if you can, I'd recommend it to get the most out of your card. Also, I have a 92mm fan over my GPU's via a Zalman Fan bracket, so dont attempt my OC's without adequate cooling. Below are the speeds my GPU's oc'd to comfortably with decent temps and perfect stability. I'm sure I could have pushed the GTS more, but there was no need.

BFG 7950 GX2 600/1580

XFX 8800 GTS 630/2000

I am not going to be providing screenshot links with these results, as the tests were exactly the same as the shots above. The only difference was the frame rate, but image quality wise I used the same settings and thus the screenshots would have been exactly the same too! Anyway, without further a do, on to the results.


Half Life 2: Episode 1

1.)

7950 max settings (4x AA) 53fps (from 40fps)

8800 GTS max settings (4x AA) 161fps (from 138fps)

7950 GX2 max settings (4xTRSSAA) 38fps (from 29fps)

GTS max settings (4xTRSSAA) 119fps (from 94fps)

GTS max settings (8xQA TRSSAA) 58fps (from 46fps)

2.)

GX2 max settings (4xAA) 57fps (from 43fps)

GTS max settings (4xAA) 163fps (from 154fps)

GX2 max settings (4xTRSSAA) 42fps (from 32fps)

GTS max settings (4xTRSSAA) 157fps (from 112fps)

GTS max settings (8xQA TRSSAA) 75fps (from 59fps)

3.)

GX2 max settings (4xAA) 64fps (from 48fps)

GTS max settings (4xAA) 163fps (from 123fps)

GX2 max settings (4xTRSSAA) 58fps (from 46fps)

GTS max settings (4xTRSSAA) 125fps (from 119fps)


Eldar Scrolls IV: Oblivion

1.)

GX2 max settings HDR no AA (city) 73fps (from 72fps)

GTS max settings HDR no AA (city) 91fps (from 92fps)

GTS max settings HDR + 4xTRSSAA (city) 76fps (from 63fps)

2.)

GX2 max settings HDR no AA (foilage) 55fps (from 43fps)

GTS max settings HDR no AA (foilage) 63fps (from 53fps)

GTS max settings HDR + 4xTRSSAA 50fps (from 39fps)

3.)

GX2 max settings HDR no AA (outdoor landscape) 48fps (from 40fps)

GTS max settings HDR no AA (outdoor landscape) 61fps (from 50fps)

GTS max settings HDR + 4xTRSSAA (outdoor landscape) 42fps (from 34fps)


Call of Duty 2 (SLI optimisations for GX2 turned ON)

GX2 max settings no AA - 31fps (from 26fps)

GTS max settings no AA - 64fps (from 53fps)

GX2 max settings 4xAA - 27fps (from 23fps)

GTS max settings 4xAA - 54fps (from 45fps)

GX2 max settings 4xTRSSAA - 24fps (from 20fps)

GTS max settings 4xTRSSAA - 47fps (from 39fps)


FEAR

GX2 max settings no AA no AF no Soft Shadows - 119fps (from 103fps)

GTS max settings no AA no AF no Soft Shadows - 111fps (from 88fps)

GX2 max settings 4xTRSSAA 16xAF Soft shadows ON - 73fps (from 58fps)

GTS max settings 4x TRSSAA 16xAF Soft shadows ON - 59fps (from 49fps)





Well there you have it. An idea of how the 7950 GX2 and 8800 GTS perform after OC's as well as against each other. The 8800 GTS pretty much demolishes the 7950 GX2, especially after a good OC. The only game I tested where the 7950 GX2 won out was in FEAR. Most probably because FEAR is well knows to be very heavily SLI optimised. But who knows, perhaps with further driver releases the 8800 will do even better.

On the whole, I am utterly impressed with the GTS.
Considering in the Uk you can buy it for £325, £15 more then an X1950 XTX and about £30-£60 less expensive then a 7950 GX2, it really is a considerable bargain.

kudos's to Nvidia for a absolutely stonking GPU release.

Next to get my mitts on an 8800 GTX and see how an overclocked one of those compares with my GTS! I'll also try and fo other game benchmarks.


Cheers!
Nib95



***UPDATE***


Please continue reading below to see further benchmarks of 16xTRSSAA compared to 4xTRSSAA as well as a fix for using 16xTRSSAA in FEAR with driver 97.02.

 

[happy medium]

Thanks for the benchmarks.

 

[nib95]

Originally posted by: happy medium
Thanks for the benchmarks.

No problem

 

[Dethfrumbelo]

Great job. Better than most of the site reviews out there.

The FEAR results are odd indeed, but overall the 8800GTS whips the GX2 nicely (sometimes nearly twice as fast).

The overclock performance looks legit as well. Now all I need is to find one at $450 or less.

 

[imported_Shaq]

Great review!

Hard to believe that those are @ 1920x1200. You should be able to run at 16xTRSSAA for the same performance as 4xTRSSAA. A couple reviews showed 4x and 16x antialiasing as basically the same fps because of the new architecture. You may want to rerun 1 or 2 of those on the GTS side at 4xTRSSAA vs. 16xTRSSAA to be sure. However, the GX2 can't do it, but it would be cool to show that there is no frame rate loss at 4x the antialiasing.

Great job again! Will be looking forward to the GTX results.

 

[NoStateofMind]

Nice. Altho now I want to see how different it is using AMD vs Intel.

 

[m21s]

Did you patch Company of Heroes?

Unpatched runs like crap on SLI.

I average 30+ at 1920x1080 with everything maxed on my 7950GX2

 

[mazeroth]

Thanks for taking the time to do all this work and provide your results!!!

 

[Nightmare225]

Originally posted by: Dethfrumbelo
Great job. Better than most of the site reviews out there.

The FEAR results are odd indeed, but overall the 8800GTS whips the GX2 nicely (sometimes nearly twice as fast).

The overclock performance looks legit as well. Now all I need is to find one at $450 or less.

FEAR loves SLI.

 

[sisq0kidd]

In FEAR, some boxes look faded on the 8800gts.

Thanks for the review though!

 

[nib95]

Originally posted by: Shaq
Great review!

Hard to believe that those are @ 1920x1200. You should be able to run at 16xTRSSAA for the same performance as 4xTRSSAA. A couple reviews showed 4x and 16x antialiasing as basically the same fps because of the new architecture. You may want to rerun 1 or 2 of those on the GTS side at 4xTRSSAA vs. 16xTRSSAA to be sure. However, the GX2 can't do it, but it would be cool to show that there is no frame rate loss at 4x the antialiasing.

Great job again! Will be looking forward to the GTX results.

Wholey moley!
You were right!!

Just checked and it's true, you can use 16xTRSSAA for very little performance difference as compared to 4xTRSSA! Here are benchmark results.


8800 GTS 16xTRSSAA compared to 4xTRSSAA


HL2: Episode 1

1.)

GTS max settings (16xTRSSAA) 86fps (from 119fps)

2.)

GTS max settings (16xTRSSAA) 92fps (from 112fps)

3.)

GTS max settings (16xTRSSAA) 120fps (from 125fps)


Eldar Scrolls IV: Oblivion

1.)

GTS max settings HDR + 16xTRSSAA (city) 70fps (from 76fps)

2.)

GTS max settings HDR + 16xTRSSAA 44fps (from 50fps)

3.)

GTS max settings HDR + 16xTRSSAA (outdoor landscape) 35fps (from 42fps)



Call of Duty 2

GTS max settings 16xTRSSAA - 43fps (from 47fps)



FEAR for some reason just went bonkers. There were artifacts and changes of colour galore. This confirms my feelings that there is some sort of compatibility issue with these new 97.02 drivers and FEAR. That might also explain the odd results when compared to a GX2. It's definately not from overclocking, as I've ran 3dMark06 back to back for a few hours without a single glitch. Every other game also ran without a glitch. Unsual.

Anyway, again, more Kudos to Nvidia, looks like I'll be using 16xTRSSAA from now on.

One note however, 16xQ TRSSAA nearly halfs the frame rate, but to be honest shows very little difference at all. In some of the benchmarks above I used 8xQ TRSSAA which explains why 16xTRSSAA gets better frame rates then 8xQ TRSSAA. Personally I think 16xTRSSAA looks better, the fact that it performs better is a meagre bonus!

 

[schneiderguy]

great benchmarks! :thumbsup:

 

[nib95]

***UPDATE***

As I mentioned above, when i tried 16xTRSSAA on FEAR it went loopy and I started getting artifacts and colour obscurities everywhere.

I had 16xTRSSAA set on force in Nvidia control panel, and 0xAA set in FEAR.
I then tried 16xTRSSAA forced in Nvidia and 4xAA set in FEAR.

This time the benchmark ran, but I got an abysmal score of 32fps average, at max everything mind. Inc Soft shadows on.

However, I then tried 16xTRSSAA in Nvidia set to "enhance application AA" instead of "Force control panel AA" and left 4xAA in FEAR.
Well guess what! The benchmark ran super smooth and I got a similar result to what I got with 4xTRSSAA. So when playing FEAR with 97.02, use the control panel "enhance application AA" instead of force. I'm still adament that a new driver release could potentially still increase performance.

Anyway, here's a screenshot.

FEAR 1920 x 1200 16xTRSSAA max everything inc Soft shadows - 50fps average

 

[the Chase]

Thanks for all the testing nib95. You mentioned the one thing that for whatever reason almost no (or just no?) hardware sites overclock both models to their respective potential and then benchmark them against each other. It's always just OC the lesser of the 2 cards to see how close it comes to the top dog. Well how do you base a buying decision or determine value off of that?

 

[BFG10K]

As I mentioned above, when i tried 16xTRSSAA on FEAR it went loopy and I started getting artifacts and colour obscurities everywhere.

Soft shadows and AA are not compatible together in Fear.

 

[nib95]

Originally posted by: BFG10K

As I mentioned above, when i tried 16xTRSSAA on FEAR it went loopy and I started getting artifacts and colour obscurities everywhere.

Soft shadows and AA are not compatible together in Fear.

They must be lol.
Because I just ran 16xTRSAA with Soft Shadows!

See my update above!

 

[Hyperlite]

Great work! :thumbsup:

 

[nib95]

Just tried my card at 650/2000.
It crashed on its second run of 3dMark06.

So I've dropped back down to 630.

I figure if it runs fine at 640, but eventually crashes at 650, then 630 is a nice and safe option.

 

[nib95]

Bumped.

 

[Dkcode]

Unbeliveable performance from the overclocked GTS. Great work :thumbsup:

How does it perform at stock? I am not asking you to bench everything again, just curious of how much percent the overclock gave.

 

[m21s]

Originally posted by: Dkcode
Unbeliveable performance from the overclocked GTS. Great work :thumbsup:

How does it perform at stock? I am not asking you to bench everything again, just curious of how much percent the overclock gave.

His first post includes all the stock FPS compared to his 7950GX2.

 

[MBrown]

To me the IQ of the G80 is a lot better than the previous.

 

[nib95]

Originally posted by: Dkcode
Unbeliveable performance from the overclocked GTS. Great work :thumbsup:

How does it perform at stock? I am not asking you to bench everything again, just curious of how much percent the overclock gave.

Erm, read the post again
There are non OC'd and OC'd results from both cards

 

[blobbynz]

hey just a thought - toms made this mistake - I wonder if you have too.
the gx2 defaults to non sli in the 92 driver for source. could you check the sli mode for us? I'm looking at doing the gx2 -> gtx upgrade, but as I have a crossfire board and hte r600 is sampling now - well if there is not much difference...... God i hate ati drivers
I get about 80 fps on a 3gig c2d with lost coast at 19x12 all on - but only 4x aa, so your results seem a bit low. My 2.6 amd scored about what you are getting with the sli disabled ( win64 driver bug ).
Otherwise I guess I just wait for stock of the gtx without the bad resister to turn up....

 

[josh6079]

Originally posted by: MBrown
To me the IQ of the G80 is a lot better than the previous.

Exactly, it's more like ATi's.

In fact, it's even better than ATi's, and they finally got the hint that we want good visuals and playability, not one or the other. I hope this trend will continue.