Please register or login. There are 0 registered and 1231 anonymous users currently online. Current bandwidth usage: 326.30 kbit/s December 04 - 06:01am EST 
Hardware Analysis
Forums Product Prices

  Latest Topics 

More >>


  Radeon 9700 Pro, ATi Flexes Some Muscle 
  Oct 11, 2002, 07:30am EDT 

R300 Overview Continued

By: Dan Mepham

In keeping with the DX9 specification, the R300 also supports Vertex Shader 2.0 and Pixel Shader 2.0 specifications. The 2.0 specifications of the Vertex and Pixel Shader units essentially calls for a much higher degree of flexibility in terms of the operations software designers are able to implement. While there currently isnít much software that takes advantage of this programmability, you can bet that now that the hardware is in place, software developers will take full advantage. The R300 actually features 4 discrete vertex shader pipelines, designed around a 32-bit scalar and 128-bit vector processor, which provides for extremely impressive and accurate vertex throughput.

The R300 core on our Radeon 9700 Pro is clocked at 325 MHz, which is extremely impressive considering the size of the core. Clearly ATi has done its homework in assuring that it will be able to produce acceptable yields of a product this size.

By and large, though, the limiting factor with respect to the performance of todayís video cards has not been the core logic itself, but the available memory bandwidth (or lack thereof). The cards are limited not by the coreís ability to process data, but by the memoryís ability to feed the core with data as fast as it demands. While there have been some clever approaches to dealing with this problem, the most inventive of which is ST Microís Tile-based renderer, the usual method of dealing with it has simply been to increase memory bandwidth, typically by using faster DRAMs. This is, however, an increasingly expensive approach, and ultimately will fail as faster DRAMs simply canít be made fast enough.

To increase bandwidth, ATi has increased the memory bus width to 256-bits, up from 128-bits on the Radeon 8500. The Radeon 9700 Pro tested here today is equipped with 310 MHz DDR memory (effectively 620 MHz), yielding an available bandwidth of almost 20 GB/sec. The R300s memory controller actually consists of four 64-bit load-balanced controllers, as opposed to a single 256-bit controller, which helps to prevent the wasting of bandwidth as the four load-balanced controllers keep the bus fully saturated.

In addition, the R300 also features a slightly enhanced version of ATiís HyperZ technology called HyperZ III. HyperZ III essentially reduces wasted memory bandwidth by not rendering pixels that are determined to be 'behind' other pixels based on checks of the current pixels stored in the z-buffer (in a 3-dimensional environment, like those of 3D games, Ďzí is used as the depth axis, and extends directly into and out of the monitor or scene). Overall the R300 core benefits from the most available bandwidth of any commercial level processor yet. Lastly, the R300 features two integrated 400 MHz RAMDACs, as well as native dual monitor support.

1. Introduction
2. R300, A Basic Overview
3. R300 Overview Continued
4. The Card
5. The Drivers
6. Test Disclosure
7. Performance - Antialiasing
8. Performance - Antialiasing Cont'd
9. Performance - Scaling
10. Performance - Anisotropic Filtering & Professional
11. Summary

Discuss This Article (1 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 

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.