Please register or login. There are 0 registered and 1133 anonymous users currently online. Current bandwidth usage: 326.30 kbit/s December 10 - 10:53pm EST 
Hardware Analysis
      
Forums Product Prices
  Contents 
 
 

  Latest Topics 
 

More >>
 

    
 
 

  ATi Radeon X800 XT, the new King of the Hill? 
  May 04, 2004, 09:00am EDT 
 

Pixel and vertex shaders


By: Sander Sassen

As mentioned, with the R420 architecture ATi opted not to include as of yet unsupported features; that’s clearly a different approach from their previous architectures. ATi in the past didn’t only take a lead over Nvidia in terms of performance but also in features; they were the first to ship a fully DirectX 9.0 compatible part. This meant that buying an ATi graphics card was a relatively safe buy, as it was more future-proof than Nvidia’s. ATi’ products were able to take advantage of game features that were to be used in games up till a year down the line. However, you can’t always plan too far ahead, as the games industry might move towards something else. And in fact it did, multiple, powerful, pixel and vertex shaders are what is needed today, and although older architectures have these shaders on board, they’re not powerful enough, or there’s too few of them to matter.

Radeon 9800XT, Radeon X800 XT

Fig 2. The Radeon X800 XT heatsink is even smaller than that of the Radeon 9800 XT.

If we look at the old R360 architecture we see that it only supports up to four vertex shader pipelines and eight pixel shader pipelines with one texture unit per pipeline. Nvidia’s NV38 architecture has to make do with even less, four vertex shader and four pixel shader pipelines, although with two texture units per pixel pipeline. The R420 architecture however has no less than six vertex shaders and sixteen pixel pipelines, or rather four groups of four, with one texture unit per pipeline. Nvidia has the same number of vertex and pixel shader pipelines. One noteworthy difference is though that ATi keeps with the DirectX 9.0 specification and thus uses 24-bit precision throughout the pipeline, Nvidia uses the, as of yet unreleased, DirectX 9.0c specification and employs 32-bit precision.

This means that the type of shader supported is different. Due to their DirectX 9.0 compliance, ATi supports up to shader model 2.0, which besides offering ‘just’ 24-bits precision also allows for no more than 256 vertex shader and 512 pixel shader instructions per vertex- or pixel shader program. Nvidia supports shader model 3.0 which enables 65.535 individual instructions per program, both for the vertex and pixel shaders. But ATi also has a feature that Nvidia lacks, 3Dc compression, which is a step up from DXTC, S3TC and is an open standard introduced with the R420 architecture and the X800 series of ATi graphics cards. 3Dc compression is used on normal maps which are created to give a rendered image texture, it allows for a 4:1 texture compression ratio and is hardware accelerated in the R420 architecture.



1. Introduction
2. Pixel and vertex shaders
3. Benchmarks, what and how?
4. 3Dmark03, 2001 SE and AquaMark3
5. UT2003, AA and AF
6. UT2004, AA and AF
7. Halo, AA and AF
8. Far Cry, AA and AF
9. Benchmarks, evaluation
10. Conclusion
11. Pricing Information

Discuss This Article (18 Comments) - If you have any questions, comments or suggestions about the article and/or its contents please leave your comments here and we'll do our best to address any concerns.


Rate This Product - If you have first hand experience with this product and would like to share your experience with others please leave your comments here.

 

    
 
 

  Related Articles 
 
 

  Newsletter 
 
A weekly newsletter featuring an editorial and a roundup of the latest articles, news and other interesting topics.

Please enter your email address below and click Subscribe.