X-bit labs

Started by: Wilfred | Date 08/25/06

Comments: 14 | Last Comment:  12/13/07

[1-10]

1. Performance will probably continue to improve with new drivers considering the flexible memory controller the R580 has. It's probably still optimized toward the lower bandwidth conditions of GDDR3 so there should be a boost once they reconfigure and optimize it toward the higher bandwidth GDDR4.

Still, I'm disappointed that ATI didn't try to bin higher and launch the X1950XTX at 675MHz core. With all that extra bandwidth, the core could have easily been made more hungry. I guess your difficulties in overclocking suggest higher stock core clocks would be difficult although people have been offering 700MHz core clock models. A 80nm shrink would have helped, but c'est la vie.

2. In the HL2 videos you see that little hitch that Source games are infamous for. I just wish review sites would have the BALLS to complain about the little stutters and hitching instead of kissing Valves ass everytime they make a game.

3. why do you guys bechmark in QUality :*( it just stupid since most people know Quality setting on Nvidia has shocking IQ.

4. Did you have to remove Warhammer Dawn of War from the benchmarking lineup? I was curious as to whether ATI would ever fix its lacklustre performance on it. I would of thought its more popular than the other two RTS you benchmark (not that I'm suggesting removing either of them)

5. Thanks for providing a few videos comparing the IQ between ATI and Nvidia's default driver settings. Xbitlabs is probably the first American hardware reviewer site to cover that issue. Hopefully that will be the beginning of the exploitation of Nvidia's lackluster default brilinear IQ settings. There was hardly any pressure on Nvidia to fix that problem, until now (hopefully).

With attention brought to this issue, let us begin comparing them against each other with High Quality settings. We could also test them against each other in default settings to see what the difference is.

Hope it's not too much for me to ask of you, but may you kindly update the review with Geforce cards benchmarked in High Quality settings, as a side-to-side comparison with default Quality settings? It will make for a very interesting review--and perhaps one of the most popular GPU reviews before the G80 is released. At least, it is guaranteed to be the most popular X1950XTX review.

Gamers who pay $400 for high-end Nvidia cards certainly do not want to play in the horrible default Quality setting that is actually low-quality in mipmap shimmering/noise.

"Xbitlabs is probably the first American hardware reviewer site..."

American? Try www.ixbt.com, www.ixbt.ru, www.digit-life, www.ferra.ru... All the same Russian publishing group. One of the best technically analytical sources on the web.

I Agree, when reviewing the top end models, only highest image quality settings should be used (i.e: adaptive aa vs. super sampling aa), for i will only use these settings if i get one of those premium models.

6. "the game runs on a modified, yet still obsolete engine from Quake 3" (on Call of Duty 2, p11)

Wasn't the first Call of Duty based on Quake 3 and the second a new engine?

"Infinity Ward have moved away from the Quake [3] engine they used for the first Call of Duty, and have coded a new engine (borrowing a range of elements from the Doom 3 engine) for Call of Duty 2. This engine appears to deliver the goods and when run at maximum eye candy really does look as good, if not better, than the Doom 3 engine."

http://www.tweakguides.com/COD2_1.html

7. Test

8. After doing some research with the current 90 series drivers by Nivida, it seems that you do not have to use High Quality image settngs after all. All you have to do is disable Anisostropic Sample Optimization, which still causes shimmering. Leaving Trilinear Optimization on does not seem to cause any shimmering in the newer drivers. Just make sure that Anisostroipic Mip Filter Optimization is also off.

ATI drivers do not use optimization with Anisostropic Filtering after all, at least it does not cause shimmering.

That way, there is as close to perfect match as possible between Nvidia and ATI cards in image quality when testing or benchmarking.

Further testing is needed for confirmation with other cards, since I have only tested it with 6800 Ultra and the 7900 series.

Whoops! Please never mind the above. I was trying out modded drivers. The official driver releases still cause shimmering even with ONLY Trilinear Optimization enabled. The only solution is to just use High Quality, which requires that all of the optimizations are disabled.

9. The article's spec sheet was wrong.

1. The 'adaptive antialiasing' on ATI cards nclude support of mixing multsampling with supersampling on transparent texture.

2. Vertex texture fetch is can be done Radeon's vertex shader, except texture formats are not exposed. Vertex texturing can still be done, but using pixel shader.[1]

3. Tessellation on all cards in the list are supported by loading differ vertex shader programs[2]. Such function can be offloaded to pixel shader as well.

4. NVIDIA cards have supported rotated grid supersampling since GeForce 6[3].

[1] http://www.bit-tech.net/hardware/2005/10/17/sapphire_radeon_x1800xl/3.html

[2] ftp://download.nvidia.com/developer/Papers/2004/Vertex_Textures/Vertex_Textures.pdf

[3] http://www.nvidia.com/object/IO_12394.html

10. please re-upload these vids!
http://www.xbitlabs.com/articles/video/display/ati-x1950xtx_8.html

[1-10]