Make Home Page
Add to Favorites
1.
*Excellent* article. Nice, balanced, very informative. I will use this for reference now from now on when I'm comparing GPUs.
[Posted by: Olop | Date: 08/18/06 07:46:43 PM]
Discussion
Discussion on Article:
Seven Games and One Week: Games vs. GPUs
Started by: Olop | Date 08/18/06 07:46:43 PM
Comments: 20 | Last Comment: 01/12/07 02:45:27 PM
Expand all threads | Collapse all threads
[1-14]
2.
thanks Mr.stepin, you're the best reviewer throughout the web!
but cant you use a more graphical chart?(like anandtech and Guru3D)
i'm sure a pro reviewer needs a pro Chart!
Thanks again!
Moein, Iran
but cant you use a more graphical chart?(like anandtech and Guru3D)
i'm sure a pro reviewer needs a pro Chart!
Thanks again!
Moein, Iran
[Posted by: Moein | Date: 08/19/06 01:42:58 AM]
3.
I dun'no something wrong here folks?
I have a 1600pro agp Athlon64 3200 and it runs Halflife EP1,100- 85 fps never drops below 45fps, 1024x768 everything on high but no HDR or FSAA. IMHO HDR is horrible in this game. As for FSAA= fuzzy.
For $165 sure it's not the best card but it does the job. Why would I pay 2 or 3 times that amount to enable some lame ass fetures that marketing hype says I should have. Remember before you spend folks, your $500 card today is tomorrows old bargin bin $75 news.
Keep in mind most LCD panels do not refresh more than 75 fps, something you won't EVER here ATI or nVidia talk about.
Just let'n off some steam here people as these skewd "perfomance reviews" are tiresome and misleading.
I have a 1600pro agp Athlon64 3200 and it runs Halflife EP1,100- 85 fps never drops below 45fps, 1024x768 everything on high but no HDR or FSAA. IMHO HDR is horrible in this game. As for FSAA= fuzzy.
For $165 sure it's not the best card but it does the job. Why would I pay 2 or 3 times that amount to enable some lame ass fetures that marketing hype says I should have. Remember before you spend folks, your $500 card today is tomorrows old bargin bin $75 news.
Keep in mind most LCD panels do not refresh more than 75 fps, something you won't EVER here ATI or nVidia talk about.
Just let'n off some steam here people as these skewd "perfomance reviews" are tiresome and misleading.
[Posted by: Dennis_1 | Date: 08/19/06 06:51:51 AM]
4.
Two comments: First, page 21, the 1920x1200 graph, the scores for the 7900's and 1900XT are not in agreement with the bar graph on page 4 - the 7900's scores are way high, while the 1900XT is low...
Second, am rather curious why the 7900GTX is included in both the premium and high end categories, when it's MSRP is $100 more than the 1900XTX? It's nice to have the reference point in the high end chart, as it's indicative of what an overclocked 7900GT would perform like, but stating it to be in the same price range as the 1900XT even at the time of publication is a bit of a stretch.
Second, am rather curious why the 7900GTX is included in both the premium and high end categories, when it's MSRP is $100 more than the 1900XTX? It's nice to have the reference point in the high end chart, as it's indicative of what an overclocked 7900GT would perform like, but stating it to be in the same price range as the 1900XT even at the time of publication is a bit of a stretch.
[Posted by: Khato | Date: 08/19/06 10:50:52 AM]
5.
Totally bias review. 7600gt are less noisy?...oh, come on. 7600gt sux...no VIVO, 128bit and are clocked 2 high for their own good. HIS 1800gto ICEQ 3 at 700/800 with 16pipes for $189 kills it.
[Posted by: Bias review everywhere | Date: 08/20/06 02:52:31 AM]
7.
these cards are very popular in market, but very few reviews
Owing to the terribly difference between 7300GT GDDR2 and GDDR3, the referral performance 350/666 7300GT GDDR2 is pointless
Basically most 7300GT GDDR3 ranged from 500/1300 to 560/1600
That would be a good competitor to 7600GS...
[Posted by: Is that possible to include 7300GT GDDR3? | Date: 08/20/06 10:35:59 PM]
8.
Hello guys, my first time posting on Xbit forums...
Here is a serious quality issue concerned here: The settings used for testing of Nvidia cards give it anywhere from 10% to 40% performance boost--namely the 'Quality' image setting. This default setting requires that Trilinear optimization and Anisotropic sample optimization are on. This degrades the image quality to unacceptable levels that cannot be seen on ATI cards--which is the flickering of mipmaps due to excessive trilinear and AF optimizations (heavy texture shimmering that cannot be seen in captured snapshots--only in motion). When turning these optimizations off, it can degrade performance by as much as 40% in UT2004.
Nvidia's default Quality setting offers much worse texture filtering quality than ATI's default Catalyst AI Normal, Quality setting.
For more of apples-to-apples comparison, please use High Quality for testing Nvidia cards while forcing Full Trilinear Filtering to be on in ATI's drivers (although only for Direct3D games). This is as close to exact match as it can be between Nvidia and ATI in image quality.
For ensuring that Full Trilinear Filtering is on in OpenGL games with ATI cards, simply enable HQ Anisostropic Filtering. ATI's High Quality AF setting forces full trilinear due to the implementation of such maximum AF algorithm. If you feel that enbaling High Quality in Nvidia's drivers would give ATI an unfair advantage, then simply enable ATI's High Quality AF while keeping the Catalyst AI at normal. There is no way to turn off Nvidia's hidden optimizations for certain games, so as you already know, ATI's Catalyst AI should not be turned off either. Keep in mind that the AF setting should be left in the Quality setting, instead of Performance, so that trilinear filtering will be allowed.
Why give Nvidia a 10%-40% advantage with its default Quality setting that offers much worse shimmering that cannot be done on ATI's cards, if you can give ATI a measly 2%-5% advantage by changing Nvidia's setting to High Quality, and leaving ATI's in default? If you are really on Nvidia's side, or if Nvidia has "bribed" you, then simply give Nvidia a 5%-15% advantage by turning on ATI's High Quality AF while keeping Nvidia's setting on High Quality--that way, you can be ensured that FULL trilinear is implemented in virtually all of the filtered textures for both ATI and Nvidia. That would be unfair to ATI, since ATI would be offering unprecedented IQ over Nvidia's.
The solution, as you and I know it, is to change Nvidia's setting to High Quality while leaving ATI's at default (and forcing full trilinear on in the registry or with ATITool program for Direct3D games). This will remove ATI's well-known advantage in some Direct3D games by up to 5%, BUT removing Nvidia's shimmering optimizations will penalize the GeForces by up to 40% in games that showcase lots of distant textures.
The evidence is here, although by ATI:
http://img320.imageshack.us/my.php?image=nvdesperatestand134 d.jpg
It does not mention UT2004, which does indeed show a 40% penalty, as tested by independent users.
For the videos on Nvidia's shimmering, check:
http://www.3dcenter.org/artikel/g70_flimmern/index2_e.php
Even enabling the two hidden optimizations in ATI's drivers (Trilinear Optimizations and AF Optimizations) AND setting Catalyst AI to High still does not cause the shimmering effects of Nvidia's undersampled AF in Quality setting.
If you, Xbit Labs, do not agree with me, please rest assured that 3DCenter of Germany has the evidence in videos along with the suggestion: "All benchmarks using the standard setting for NV40 and G70 against the Radeon are invalid, because the Nvidia cards are using general undersampling which can (and does) result in texture shimmering."
It means that only the benchmarks using High Quality setting for Nvidia cards are valid, within the +/-5% margin instead of the 15-40% margin concerning such unfair (and ugly) advantage over ATI.
Here is a serious quality issue concerned here: The settings used for testing of Nvidia cards give it anywhere from 10% to 40% performance boost--namely the 'Quality' image setting. This default setting requires that Trilinear optimization and Anisotropic sample optimization are on. This degrades the image quality to unacceptable levels that cannot be seen on ATI cards--which is the flickering of mipmaps due to excessive trilinear and AF optimizations (heavy texture shimmering that cannot be seen in captured snapshots--only in motion). When turning these optimizations off, it can degrade performance by as much as 40% in UT2004.
Nvidia's default Quality setting offers much worse texture filtering quality than ATI's default Catalyst AI Normal, Quality setting.
For more of apples-to-apples comparison, please use High Quality for testing Nvidia cards while forcing Full Trilinear Filtering to be on in ATI's drivers (although only for Direct3D games). This is as close to exact match as it can be between Nvidia and ATI in image quality.
For ensuring that Full Trilinear Filtering is on in OpenGL games with ATI cards, simply enable HQ Anisostropic Filtering. ATI's High Quality AF setting forces full trilinear due to the implementation of such maximum AF algorithm. If you feel that enbaling High Quality in Nvidia's drivers would give ATI an unfair advantage, then simply enable ATI's High Quality AF while keeping the Catalyst AI at normal. There is no way to turn off Nvidia's hidden optimizations for certain games, so as you already know, ATI's Catalyst AI should not be turned off either. Keep in mind that the AF setting should be left in the Quality setting, instead of Performance, so that trilinear filtering will be allowed.
Why give Nvidia a 10%-40% advantage with its default Quality setting that offers much worse shimmering that cannot be done on ATI's cards, if you can give ATI a measly 2%-5% advantage by changing Nvidia's setting to High Quality, and leaving ATI's in default? If you are really on Nvidia's side, or if Nvidia has "bribed" you, then simply give Nvidia a 5%-15% advantage by turning on ATI's High Quality AF while keeping Nvidia's setting on High Quality--that way, you can be ensured that FULL trilinear is implemented in virtually all of the filtered textures for both ATI and Nvidia. That would be unfair to ATI, since ATI would be offering unprecedented IQ over Nvidia's.
The solution, as you and I know it, is to change Nvidia's setting to High Quality while leaving ATI's at default (and forcing full trilinear on in the registry or with ATITool program for Direct3D games). This will remove ATI's well-known advantage in some Direct3D games by up to 5%, BUT removing Nvidia's shimmering optimizations will penalize the GeForces by up to 40% in games that showcase lots of distant textures.
The evidence is here, although by ATI:
http://img320.imageshack.us/my.php?image=nvdesperatestand134 d.jpg
It does not mention UT2004, which does indeed show a 40% penalty, as tested by independent users.
For the videos on Nvidia's shimmering, check:
http://www.3dcenter.org/artikel/g70_flimmern/index2_e.php
Even enabling the two hidden optimizations in ATI's drivers (Trilinear Optimizations and AF Optimizations) AND setting Catalyst AI to High still does not cause the shimmering effects of Nvidia's undersampled AF in Quality setting.
If you, Xbit Labs, do not agree with me, please rest assured that 3DCenter of Germany has the evidence in videos along with the suggestion: "All benchmarks using the standard setting for NV40 and G70 against the Radeon are invalid, because the Nvidia cards are using general undersampling which can (and does) result in texture shimmering."
It means that only the benchmarks using High Quality setting for Nvidia cards are valid, within the +/-5% margin instead of the 15-40% margin concerning such unfair (and ugly) advantage over ATI.
[Posted by: Bo_Fox | Date: 08/20/06 10:44:48 PM]
9.
QUOTE:
"The GeForce 7900 GTX falls into the same price range as the Radeon X1900 XT, so both these cards are present in the high-end category. But while the Radeon X1900 XTX looks competitive against the GeForce 7900 GTX in the premium class, the Radeon X1900 XT with lower GPU and memory clock rates has no chance against such a formidable opponent."
Retail pricing for the X1900XT is around $320 on Newegg.com while the pricing for 7900GTX is around $470. (I checked them both after I read this article). How can they belong to the same price range having a price difference of $150? I don't think it's fair comparing these two cards. It's more like x1900xt vs 7900GT. Thanks.
"The GeForce 7900 GTX falls into the same price range as the Radeon X1900 XT, so both these cards are present in the high-end category. But while the Radeon X1900 XTX looks competitive against the GeForce 7900 GTX in the premium class, the Radeon X1900 XT with lower GPU and memory clock rates has no chance against such a formidable opponent."
Retail pricing for the X1900XT is around $320 on Newegg.com while the pricing for 7900GTX is around $470. (I checked them both after I read this article). How can they belong to the same price range having a price difference of $150? I don't think it's fair comparing these two cards. It's more like x1900xt vs 7900GT. Thanks.
[Posted by: Loc | Date: 08/21/06 10:01:48 AM]
10.
Nvidia-biased? The prices say it all.
Look at Bo_Fox's article. It makes so much sense! If the 7900GTX was tested at High Quality settings, it would actually be just a little slower than the X1900XT. The 7950GX2 would hardly be any faster than the X1900XTX in most games.
Yeah, I can see the shimmering with my 6800 Ultra when I set it to default "Quality". Very distracting in UT2004, Battlefield 1942, BF2, etc... Looks horrible when Nvidia "cheats" in its undersampling with mainly bilinear 2X AF texture layers, switching with 16X AF textures based on which mipmap gets the priority in display. It even looks like Z-fighting, but appears ALL over the textures.
Look at Bo_Fox's article. It makes so much sense! If the 7900GTX was tested at High Quality settings, it would actually be just a little slower than the X1900XT. The 7950GX2 would hardly be any faster than the X1900XTX in most games.
Yeah, I can see the shimmering with my 6800 Ultra when I set it to default "Quality". Very distracting in UT2004, Battlefield 1942, BF2, etc... Looks horrible when Nvidia "cheats" in its undersampling with mainly bilinear 2X AF texture layers, switching with 16X AF textures based on which mipmap gets the priority in display. It even looks like Z-fighting, but appears ALL over the textures.
[Posted by: Wow! | Date: 08/21/06 12:10:46 PM]
11.
This review is very comprehensive and definitely useful in considering such an investment. However I do not think that the X1900XT (High End) is not equal to the 7900GTX in a number of games such as Morrowind where the min. frames are essential. It is also stated they are similar in price, this is ludicrous. A 512MB X1900XT can be had for just over $500 whereas the 7900GTX will cost at least $650. Not exactly the same pricepoint...
[Posted by: Martin | Date: 08/22/06 03:27:12 AM]
12.
Why don't you bench with high quality driver settings?
[Posted by: honkflow | Date: 08/24/06 04:24:40 AM]
13.
Yup, bench all the cards with high setting, those shimmering mips reported by others are the reason I allways keep the setting at high in drivers, be it ATI or Nvidia.
[Posted by: STB | Date: 08/25/06 07:00:29 AM]
14.
what is the pass for the game
[Posted by: adi-m | Date: 01/12/07 02:45:27 PM]
[1-14]
Our Sponsors
