1. Introduction | 2. Specification | 3. Drivers | 4. Benchmarking (pt.1) | 5. Benchmarking (pt.2) | 6. OpenGL | 7. Demos | 8. Conclusion  


RADEON 9800 Pro review


Reviewed by Allan "Zardon" Campbell

Its been a great six months for ATI, since the release of the 9700 Pro they have dominated the high end mainstream market, superseeding Nvidia's 4600 by a long shot. The combined value for money and the outright performance, specifically with FSAA and aniso enabled, shot it to the number one position with all die hard gamers the world over. Its no surprise that ATI recently released news that they shipped their millionth Direct X 9 class graphics card. As we all know Nvidia's FX really did fail to hit the big time, with suprisingly average performance for such a long delayed product, the shock was only reinforced by the horrendous noise problems.

So what have ATI got in store for us gamers with the new top of the line 9800 pro, the much awaited R350 core? Is this set to be the nail in FX's coffin or is it just basically an overclocked 9700 pro? In this review I will not only be detailing the performance of this new card but I will also be giving my opinion as to whether its a worthwhile upgrade for all of you who currently have the good fortune to own a 9700 based card. I would also have liked to test it against the FX product that I had for a few days, but unfortunately I had to return this card long before ATI sent me this 9800 pro.

Product Overview and feature list
Over the last six months, the designers of the RADEON 9700 pro looked back at all areas of the chip to identify areas where timings and signal integrity could be improved. The result of this work, the RADEON 9800 PRO, features a highly optimised design that is able to run at significantly higher clock speeds than its predecessor, without the need for large, noisy cooling solutions. erm, FX anyone?

There is currently only one configuration of 9800 available, that being 128 meg of DDR with VGA, DVI-I and TVOUT. There is still the additional power supply connection needed, but this has changed from the smaller connection on the 9700 to a standard 4 pin molex. The pictures below will show this connection. ATI have stated that this requires a 300 Watt PSU or greater to ensure normal system operation, especially when there are a number of other internal devices installed.


Images above: 9800 pro (top) and Sapphire 9700 pro.
You can clearly see in the above image the different power connector sizes.
click for larger image.

As you can see from the images there is not a vast difference in the card sizing, the 9800 pro being just slighty longer, so it shouldnt prove to be a problem for most systems when fitting. Core and memory speeds on the new card are 380/680 respectively.


Firing up the latest Powerstrip V3.30 (below) verifies these figures

Alot of the text in this section will already be known to the more experienced users but I feel its worth recapping over the various technologies for those coming into this article without any prior knowledge, if you know it already or get a sore head from reading it, then skip to page 2.

System requirements are Windows XP, Windows 2000 or Windows ME, with a Pentium III or IV or AMD K7/Athlon XP, AGP 4x (1.5V), AGP 8x (0.8V) or Universal AGP 3.0 - 4x/8x slot. 128 Meg of system memory is needed along with a CD drive to install the software.

SMARTSHADER™ sets a new standard for visual realism with its 128 bit floating point colour precision, enabling hyper-realistc imagery to be displayed with billions of colour variations and allows users to experiance complex, movie quality effects in next generation games and applications.

  • Full support for Microsoft Direct X 9.0 programmable vertex and pixel shaders in hardware
  • 2.0 Vertex Shaders support vertex programs up to 65,280 instructions with flow control
  • 2.0 Pixel Shaders support up to 16 textures per rendering pass
  • New F-Buffer technology supports fragment shader programs of unlimited length
  • 128-bit, 64-bit and 32-bit per pixel floating point colour formats
  • Multiple Render Target (MRT) support
  • Shadow volume rendering acceleration
  • Complete feature set also supported in
  • OpenGL via extensions
  • High dimension floating point textures

SMOOTHVISION™ 2.1 technology enhances image quality by removing jagged edges and bringing out fine texture detail, without compromising performance. This 2nd generation technology provides new levels of image quality with advanced full-scene anti-aliasing (FSAA) and anisotropic filtering.

VIDEOSHADER™ RADEON 9800 PRO VPU continues to provide industry leading video playback with VIDEOSHADER technology: a seamless integration of pixel shaders with video. the RADEON 9800 PRO VPU also offers FULLSTREAM video de-blocking technology that removes blocky artifacts (noise removal filtering) for captured video and provides sharper image quality.

FSAA and Aniso: The Radeon 9800 like its older brother has 2x/6x/8x full scene anti-alaising modes and a maximum 16x aniso setting. To go into a little more detail on these: Firstly AA: This is a sampling technique that creates more detailed and realistic looking images, by removing the stair stepping effect seen on the edges of objects within computer generated images. High quality anti-aliased graphics are achieved with sub-pixel edge detection and colour compression for greatly improved performance.
Anisotropic filtering enhances visual quality by rendering sharp, detailed textures. As more texture samples are filtered, the image quality improves. The RADEON 9800 PRO VPU filters more samples than "the competition", with minimal performance degradation. 2x/4x/8x/16 modes are available.

Further details on the card, with some excellent (as usual) ATI in house artwork
Technology Features of the 9800 Pro
Click images above for full size

HYPER Z III+
Hyper Z was introduced in the RADEON 9700 PRO and included the most advanced technologies developed to date for improving memory bandwidth efficiency. These technologies focused on minimizing data transfer to and from the Z-buffer or depth buffer, which typically accounts for a majority of the memory bandwidth consumed by advanced graphics processors and can be a major performance bottleneck. Once important component of HYPER Z III is a high performance Z cache thats helps hide latency associated with Z-buffer reads and writes.

HYPER Z III+ takes this technology a step further with an enhanced Z-cache that is more flexible and better optimised to work with stencil buffer data. The stencil buffer co-exists with the Z-buffer and behaves similarly, in that an application can set a pixel#s stencil value and compare it against the value stored i nthe stencil buffer to determine if the pixel gets rendered or not. The main difference is that the Z-values in the Z-buffer represent the "depth" of a pixel, while the values in the stencil buffer can represent anything the programmer wants them to.

One of the most common uses for the stencil buffer is for rendering real-time shadow volumes. In this case, the application calculates which parts of the image fall in the shadow of other objects that have already been rendered. As long as all objects are rendered in the correct order, this technique makes it possible to generate accurate shadows for any moving object and/or light sources in a scene.

This process requires alot of extra computation, so it has been used sparingly (if at all) in most existing games. Future game engines, however, such as the Doom 3 engine, are expected to use it heavily to create very realistic environments. The enhanced Z-cache feature of HYPER Z III+ increases the performance of stencil shadow volumes and will help to delliver a superior experience when playing the next generation of 3d games.

1. Introduction | 2. Specification | 3. Drivers | 4. Benchmarking (pt.1) | 5. Benchmarking (pt.2) | 6. OpenGL | 7. Demos | 8. Conclusion