ASUS 2900 XT and 8800 Ultra Compared

[Comment Bot]

Please feel free to comment about our story entitled "ASUS 2900 XT and 8800 Ultra Compared"

[Time]

Hello TweakTown and folks,

Nice showing of all four GPU's and their performance levels.

I particularly liked the wattage, thingy, monitor, heh.

But the creme of the crop was the temperature monitor and the temps of each card, it proved to be an eye opener.

Now would probably be a good time to perform a review on liquid/water cooled GPU's Vs. passive and standard fan/air ventilation cooled GPU's(MSI GeForce NX8800GTX OC Liquid - Water cooling vs. the 8800 Ultra Article) and possibly even liquid/water cooling systems like Koolance and their ilk.

Keep up the good work.



"Adapt or die trying!" - [DeC]Nix - Mechwarrior4
"Win some, lose some and hopefully you've learned some."
"Let me go grind their bones... Let me go grind their bones to powder!" - Shadowbane

[digitalwanderer]

Rydermark?!? You used Rydermark as a serious benchmark? Not just that, but a beta?!?

"Flabbergasted" is a bit weak, but it comes close to describing my thoughts on the matter.

[silent-circuit]

1. Why vs. the 8800 Ultra? Why no mention of the 8800 GTX as a better price to performance option at this date and time in the conclusion section? ATi placed the standard 2900HD as a competitor for the 640MB 8800GTS (or the 320MB, dependant on who you're talking to). The 2900HD XT is one 'bump' up from the standard 2900HD, the 8800 GTX is one 'bump' up from the 640MB 8800 GTS. The 8800 Ultra would be a 2900HD XTX competitor, in theory, and these cards have yet to be seen on the market or in reviews, they're still just rumored. In the tests where the GTX is included it stays near to the 2900HD XT, and in many cases surpasses it at all but the lowest resolutions. Why even include anything below 1600x1050 anymore? To my mind, anything lower is completely unimportant when considering this level of hardware and DX9 games, as people willing to spend $400+ on a GPU are unlikely to have a small/low end monitor.

2. Where are the high-level AA/AF tests? Most have realized by this point that the R600 architecture has serious issues with AA enabled. This feels like a massive oversight, or as mentioned below, a really poorly thought out attempt at a cover-up.

3. S.T.A.L.K.E.R. was bundled with both cards. S.T.A.L.K.E.R. game art was plastered all over the 2900's HSF. S.T.A.L.K.E.R. is known to run very poorly on the 2900 architecture unless full dynamic lighting is disabled. S.T.A.L.K.E.R. was untested. Coincidence, or poorly thought out coverup?

4. Rydermark looks horrid, in my opinion, and with that in mind its performance on /any/ hardware is pretty much moot. It's heavy on shaders? Sure. But does it show real world shader performance? Not in the least. Shadermark doesn't either, but it's publically available unlike Rydermark, and it gives definite technical numbers -- as in, this card is however many FPS faster than this card running the same shaders at the same resolution -- and will test render quality as well.



Questionable at best. Fatally flawed at worst. This really needs to be revisited.

[random78s]

Nit a bad Review IMO. What I did appreciate was the sound pressure readings. I know that these are subject to the conditions in which the cards were tested but it does give one a general idea of the comparative volume of each solution.

[Peleg]

I agree with the previous poster...your benchmarks and analysis seemed a little odd. You kept comparing the prices of the 8800 Ultra and the 2900 XT, when, according to your own benchmarks, the 8800 GTX is clearly the competition for the 2900 XT. It appeared that the 8800 GTX was superior to the 2900 XT on almost all of your tests, so the fact that the 8800 Ultra is twice as expensive is beside the point.

Your comments about the differences between the cards being irrelevant because they all delivered playable framerates seemed a little strange too. Of course they deliver playable framerates on older games with all advanced options disabled. When you ran at higher resolutions and enabled AA/AF, the differences between the various cards were much more noticable. I'm assuming that the people in the market for a $400-500 graphics card don't care about how fast it runs Half-life 2 at 1280x1024 resolution with AA disabled...if they are buying cutting edge graphics cards I'm sure they are running at more demanding resolutions. You should probably base your recommendations on those more demanding tests. People interested in running old games at lower resolutions (like myself!) will be buying cheaper graphics cards anyway.

[ECH]

Originally Posted by silent-circuit

1. Why vs. the 8800 Ultra? Why no mention of the 8800 GTX as a better price to performance option at this date and time in the conclusion section? ATi placed the standard 2900HD as a competitor for the 640MB 8800GTS (or the 320MB, dependant on who you're talking to). The 2900HD XT is one 'bump' up from the standard 2900HD, the 8800 GTX is one 'bump' up from the 640MB 8800 GTS. The 8800 Ultra would be a 2900HD XTX competitor, in theory, and these cards have yet to be seen on the market or in reviews, they're still just rumored. In the tests where the GTX is included it stays near to the 2900HD XT, and in many cases surpasses it at all but the lowest resolutions. Why even include anything below 1600x1050 anymore? To my mind, anything lower is completely unimportant when considering this level of hardware and DX9 games, as people willing to spend $400+ on a GPU are unlikely to have a small/low end monitor.

2. Where are the high-level AA/AF tests? Most have realized by this point that the R600 architecture has serious issues with AA enabled. This feels like a massive oversight, or as mentioned below, a really poorly thought out attempt at a cover-up.

3. S.T.A.L.K.E.R. was bundled with both cards. S.T.A.L.K.E.R. game art was plastered all over the 2900's HSF. S.T.A.L.K.E.R. is known to run very poorly on the 2900 architecture unless full dynamic lighting is disabled. S.T.A.L.K.E.R. was untested. Coincidence, or poorly thought out coverup?

4. Rydermark looks horrid, in my opinion, and with that in mind its performance on /any/ hardware is pretty much moot. It's heavy on shaders? Sure. But does it show real world shader performance? Not in the least. Shadermark doesn't either, but it's publically available unlike Rydermark, and it gives definite technical numbers -- as in, this card is however many FPS faster than this card running the same shaders at the same resolution -- and will test render quality as well.



Questionable at best. Fatally flawed at worst. This really needs to be revisited.

I am not sure if you are simply unaware or trying to play some sort of game but from what I've seen Stalker is intended as a Nvidia game moreso then anything else. Sure the X-ray engine uses SM 3.0 however, if this post hold true it would explain a lot of why the HD 2900XT is having problems with this particular game. But let me point you to the main aspects of which I am talking about:
Optimized for Geforce2 and up
The engine does not support software rendering
Visualization optimized for hardware TnL
Optimized for T&L hardware by batching primitives in optimally sized groups
The Shader library is central to every part of the rendering pipeline
Multi-pass Rendering
Fallback Shaders
Source

To name a few. These are tall tell signs that this game was intended for Nvidia if the information in that post is accurate.

This review looks far to me since you can buy a R600CF for roughly the price of 8800 Ultra. Granted, it will depend on the location and methods available to make the purchase.

[btdvox]

I dont know but something seems odd about this Review, the 8800 GTX surpasses the Ultra in alot of cases and thats very abnormal. Ive read over 10 reviews and none of the GTX surpassed the Ultra in anyway. it was always usually 6-15% ahead.

the other thing is the 2900XT is very close to the GTX, well done for it, as again this is the first Review ive seen where it surpasses the GTS even alot. Most of the time there comparable (GTS 640mb that is)

Some of the benches seem kind of out of whack, any bottleneck issues maybe?

[wkdyff]

I went to buy a 2900 radeon at 3 stores checking out pricing.
When I noticed that those stores had the card I wanted, but they were all returns APROX 9-10 BOXES no new ones. The reasons for returning are....
"not giving the benchmarks quoted"
"Poor quality, not fast enough"
Then I looked for specs and benchmarks on pages like this one. they only show one or two games that out bench Nvidia. But Nvidia seems to out bench on dozens of other games that Ati dont benchmark with.
I wanted a Ati, but after what I have seen with the returns, I DONT WANT ONE!!
I purchased a 8800 gtx ultra, my neighbor has a 2900 Diamond, he returned it after he saw my FPS. He said his sucked. Was not getting what Radeon promised.