Radeon HD 2900 XT vs. 320MB 8800 GTS

[nicepun]

Bias doesn't have to be intentional, it comes from ignorance too.

Ignorance should not be an excuse.

[Daveb2012]

I love my 8800GTS I just bought to replace a x1950xtx, the card was only 227 after rebate + tax. it over clocks well is good for 18.5k on 3dmark05. I don't plan on changing this card any time soon.

[DilTech]

So v_rr, who did the screenshots come from?

[oohms]

Hey does anyone know when the Cat 7.6's come out?

[ColonelCain]

Hey does anyone know when the Cat 7.6's come out?

Probably near the end of the month, as usual. That is what I'm waiting for to make my GPU purchasing decision.

[GoThr3k]

still haven't seen any review of the 7.5 catalyst...

[gallag]

So v_rr, who did the screenshots come from?

Did v_rr ever respond to this? If he is proved to be misleading the community he should be banned

[XSAlliN]

Conclusion

Overall Performance Summary

In Oblivion we found the 320 MB and 640 MB GeForce 8800 GTS based video cards to perform faster than the ATI Radeon HD 2900 XT. Oddly enough they were able to handle higher grass distance settings in Oblivion despite the Radeon HD 2900 XT having much higher memory bandwidth.

Battlefield 2142 had a large difference in the gameplay experience between the ATI Radeon HD 2900 XT and both GeForce 8800 GTS video cards. Even with the much less expensive 320 MB GeForce 8800 GTS we were able to play the game smoothly at 16X Transparency Supersampling at 1600x1200 with no problems at all in intense gun fights with massive explosions. The more expensive ATI Radeon HD 2900 XT could not handle anything higher than 4X Performance Adaptive AA at 1600x1200.

S.T.A.L.K.E.R. also proved to separate these video cards by performance. The ATI Radeon HD 2900 XT was clearly the weaker performing video card. We had to lower the rendering quality to “Objects Dynamic Lighting” and run at 1280x1024 to receive playable performance. Unfortunately this does diminish the gameplay experience compared to the GeForce 8800 GTS based video cards. We were able to take the game up to full rendering quality and play at 1600x1200 with NVIDIA based cards. With the 320 MB version we had to drop the AF level to 4X and grass density to 50%.

Lost Planet is a fun game, plain and simple; we had a blast playing through the demo. If this is the future of gaming then we are very happy. There is no question that next generation titles will require fast hardware to keep up with the intense detail. This demo presented some interesting results for us. We found that the ATI Radeon HD 2900 XT really does take a large performance hit when enabling AA; to the point where it just isn’t a viable option right now. The GeForce 8800 GTS based video cards on the other hand don’t take as great a hit and some gamers may find 2X AA or more playable depending on what framerates you are comfortable with.

In Lost Planet’s outdoor areas the ATI Radeon HD 2900 XT, without AA, is slightly better performing than both GeForce 8800 GTS based video cards. However, in that one indoor area of the performance test called “Cave” we saw the framerates suffer and perform slower than the GeForce 8800 GTS based video cards. We cannot wait until the full version game is released so we can test all the levels and see how the video cards really compare throughout the entire game.

Display Resolution – 1920x1200

In our first evaluation we tested up to 1920x1200 resolution. 1920x1200 is the sweet spot for video cards with 512 MB or greater, of RAM. The 320 MB GeForce 8800 GTS does become memory capacity bottlenecked at this resolution and beyond, unlike the Radeon HD 2900 XT and 640 MB GeForce 8800 GTS. In our experiences we do have to lower in-game quality settings moving from 1600x1200 to 1920x1200 on the 320 MB GeForce 8800 GTS. However the impact is not that great in all cases.

For example in Oblivion instead of running at 2X AA at 1600x1200 we have to disable AA at 1920x1200 but can maintain all the same in-game options in the game. For S.T.A.L.K.E.R. we see the largest hit in performance at 1920x1200 and find we have to drop to Object Dynamic Lighting. Performance at this resolution trades blows with the ATI Radeon HD 2900 XT. For example in Oblivion we can run at 2X AA at 1920x1200 on the ATI Radeon HD 2900 XT but we have to disable grass completely. With the 320 MB GeForce 8800 GTS we have to disable AA but we can run at 50% grass.

S.T.A.L.K.E.R. on the other hand shows a clear advantage with the 320 MB GeForce 8800 GTS at 1920x1200. We are able to run at Objects Dynamic Lighting with maximum in-game settings except for 50% grass. However, on the ATI Radeon HD 2900 XT we have to drop to Objects Dynamic Lighting and decrease view distance, object detail, grass density, lighting and shadow quality to a “Lowest” setting.

Our conclusion from this is that at 1920x1200 the 320 MB GeForce 8800 GTS is more strained due to its memory capacity, but in some games it is still much faster than the ATI Radeon HD 2900 XT, while in others they trade blows back-and-forth.

Video Card Value

This evaluation is the embodiment of what video card value is to the gamer. We have a video card, the 320 MB GeForce 8800 GTS at around $289 providing a noticeable gameplay experience advantage compared to the ATI Radeon HD 2900 XT that costs $409. In some cases the performance gap is very wide (S.T.A.K.E.R. and BF 2142), in other games performance is closer (Oblivion, Lost Planet), but in most cases the 320 MB GeForce 8800 GTS is providing higher framerates than the ATI Radeon HD 2900 XT sometimes equating to a better gaming experience.

The 8800 GTS 320MB video card is $120 less expensive and provides a gaming experience that is equal to or superior to the HD 2900 XT. Not only that, but as our original evaluation proved the ATI Radeon HD 2900 XT uses much more power than a GeForce 8800 GTS! The 8800 GTS is simply more efficient both in terms of power and performance.

The Bottom Line

We hoped newer driver revisions would improve performance on the ATI Radeon HD 2900 XT. With the newer driver we used for this evaluation we did not see any “magic” happen when it comes to real world gaming experiences at resolutions at and above 1600x1200. The ATI Radeon HD 2900 XT is not even a match for even the much less expensive and much less power hungry 320 MB GeForce 8800 GTS.

As I mentioned in the other topic - R600 sux at the moment compared to his rival (nVidia).

PS.In the last 4 Years (even at the moment) I used/use an ATi card, so you can't say I'm a nVidia fanboy.

[Starscream]

well the extra grass the 8800GTS can show has 1 con.
if u kill a guy in Oblivion in and he drops his sword and u want the sword it can be a real pain looking for it in all of the grass

[Ubermann]

Do you see any FPS in those screens ?
If you want to show whats settings that is running at equal speed then for the love of god use fraps or whatever to show it.

Anything related to [H] should be considered a crime!
CPU's GPU's or whatever..

[AbelJemka]

What's the point of arguing about this review?

[XSAlliN]

Do you see any FPS in those screens ?
If you want to show whats settings that is running at equal speed then for the love of god use fraps or whatever to show it.

Anything related to [H] should be considered a crime!
CPU's GPU's or whatever..

Here are the FPS results:



And the FULL LINK to this review - made on 13 JUNE 2007 using Catalyst 7.5.

PS.I personally enjoy there reviews, compared to others sites that use outdated drivers and who knows what testing terms, plus a stupid conclusion about a product like:

This is a great card for a XMAS present

.

[AbelJemka]

Here are the FPS results:



And the FULL LINK to this review - made on 13 JUNE 2007 using Catalyst 7.5.

PS.I personally enjoy there reviews, compared to others sites that use outdated drivers and who knows what testing terms, plus a stupid conclusion about a product like: .

"We are going to jump straight to gaming on the next page. For system setup specifications look here. We are using the latest drivers officially supplied by ATI which are known as 8.37.4.2, these drivers have all the performance tweaks that are found in 8.38 which have been rolled up into Cat 7.5."
Have you read the thread? This bull review made on 13 june 2007, two weeks after cata 7.5 (8.38) release and one weeks after cata 7.5a (8.38.1) use 8.37.4.2 which are about 3 weeks old.
Better read previous posts before posting

[Eastcoasthandle]

Here are the FPS results:



And the FULL LINK to this review - made on 13 JUNE 2007 using Catalyst 7.5.

PS.I personally enjoy there reviews, compared to others sites that use outdated drivers and who knows what testing terms, plus a stupid conclusion about a product like: .

That review:
-did not use a retail HD 2900XT
-and 8.37.4.2 is not the current drivers as of June 13, 2007

[eXceeded]

What's the point of arguing about this review?

because their methods are suspect...

EDIT- And above info....

[Radeon117X]

Hmmm...this seems quite interesting. Are there any forum members who have an HD2900XT that want to try some of their own games so we can see how they perform? I know you cant do actual benchmark testing, but Im just looking for a comparison between the new drivers and the old for an idea of the improvement.

[AbelJemka]

Hmmm...this seems quite interesting. Are there any forum members who have an HD2900XT that want to try some of their own games so we can see how they perform? I know you cant do actual benchmark testing, but Im just looking for a comparison between the new drivers and the old for an idea of the improvement.

The best idea but need someone everyone trust.

[DilTech]

The best idea but need someone everyone trust.

Exactly... It's now going to be a lot harder to trust forum numbers after seeing the fraps screenshots v_rr posted.... He still hasn't answered on who he got those from.

[DeathReborn]

That review:
-did not use a retail HD 2900XT
-and 8.37.4.2 is not the current drivers as of June 13, 2007

Review POSTED on the 13th of June, could have been started before 7.5 came out.
Can you get the same performance from 3.37.4.2 as you can from 7.5? If yes then you lot should stop moaning about driver versions. If not then perhaps 8.37.4.2 was the latest when they started the review or they (like many others) don't like ATI very much.

I wish people would actually post something NEW instead of the same old tired moaning.

By the time I get a HD 2900XL the drivers will propbably be on 7.7/7.8 so it should have improved a fair bit by then.

[AbelJemka]

Review POSTED on the 13th of June, could have been started before 7.5 came out.

7.5 released on 31th of may. They started review like 2 weeks before posting it ans did all benchmarks before 31th, after they spent two weeks to write their review

[STEvil]

They dont give us any 1:1 settings to see how much you "lose" by using the same settings either.

What they SHOULD do is:

"The fastest card was able to play at these settings (x) resulting in these performance numbers (1). Using the slower card at these settings (x) resulted in this performance (2). To achieve similar playability results we had to use these settings (y) on the slower card which resulted in these numers (3)."

No more BS.

[CBONE]

They dont give us any 1:1 settings to see how much you "lose" by using the same settings either.

What they SHOULD do is:

"The fastest card was able to play at these settings (x) resulting in these performance numbers (1). Using the slower card at these settings (x) resulted in this performance (2). To achieve similar playability results we had to use these settings (y) on the slower card which resulted in these numers (3)."

No more BS.

Who cares about (2) when you know that settings above those used in (3) and below (1) still won't be playable? Why waste time testing or talking about it?

Why do people want to see just how far below playable the rates get? Once they are below playable, they may as well be 0. Since they are using the best settings that will remain playable, their method makes sense. If they were figuring out peak performance at settings no one will use, that would be something else.

[AbelJemka]

Showing all numbers improve credibility. Justing telling "i considere these settings unplayable" is not enough.
Pick Oblivion benchmarks. Justification for 25% grass is :

The framerates are more on par with the ATI Radeon HD 2900 XT, but notice that the Radeon HD 2900 XT had to run with lower grass detail. The GeForce 8800 GTS 320 MB and 640 MB handled grass performance faster than the Radeon HD 2900 XT in our experiences.

What it means without numbers? Nothing .

[ColonelCain]

Review POSTED on the 13th of June, could have been started before 7.5 came out.
Can you get the same performance from 3.37.4.2 as you can from 7.5? If yes then you lot should stop moaning about driver versions. If not then perhaps 8.37.4.2 was the latest when they started the review or they (like many others) don't like ATI very much.

I wish people would actually post something NEW instead of the same old tired moaning.

By the time I get a HD 2900XL the drivers will propbably be on 7.7/7.8 so it should have improved a fair bit by then.

When does the X2900XL come out?

[CBONE]

Showing all numbers improve credibility. Justing telling "i considere these settings unplayable" is not enough.
Pick Oblivion benchmarks. Justification for 25% grass is :

What it means without numbers? Nothing .

It serves no practical purpose to show every number possible. For purposes of the review, numbers below playability don't matter at all. The frame rate considered playable is clearly defined for each game.