[H]ard|OCP

[H]ard|OCP - The Baconnaise of Enthusiast Websites

Navigation

• [H] Forums
• Archives
• Search
• [H] Folding

• e-mail this aricle

Select Page1 - Introduction 2 - Introduction Continued 3 - Unreal Tournament 2003 4 - Unreal Tournament 2003 Continued 5 - Serious Sam 2 6 - Jedi Knight 2 7 - 3DMark03 8 - Code Creatures, Fill-Rate Tester, Dolphin Demo, SpecVIewPerf7 9 - ShaderMark v1.7 10 - Image Quality: FSAA Tests 11 - Image Quality: UT2003 12 - Image Quality: NFS HP 2, Texture Filtering 13 - Overclocking, Conclusion

Introduction

Has it been 6 months already? It was last August when ATi shocked the gaming community with the stunning 9700 pro. With a near-perfect launch and unexpectedly good driver support, this card single-handedly turned around ATi’s reputation, while throwing NVIDIA a curveball in the process. Even today, the 9700 pro is arguably the champ, even paired against NVIDIA’s latest GeForceFX. Today, ATi is out to prove that they’re not just a one-trick pony, by clearly regaining the speed crown and throwing in even more features.

Are you ready for more? That is the question ATi has popped up on their website recently. Obviously that can be interpreted as a direct slam against NVIDIA’s “Are you ready?” slogan. The difference here though is that this question is being answered directly in today’s lineup of cards that ATi is announcing. ATi is calling this the “First Hat Trick in the Graphics Industry”. (Crazy Canucks!) The three important points ATi wants you to know about this new lineup is that 1.) They are introducing a three new product series that is going to score and 2.) This will provide an improved end-user experience in all segments and 3.) This will be the most complete product lineup in the industry.

Three new products are being announced today in the Enthusiast, Performance Mainstream and Mainstream markets. Above you will also see ATI’s Product Roadmap and where they have each chip positioned.

The 9700 Pro was really the first cinematic graphics chip for the desktop platform with its DX9 VS/PS 2.0 support and higher FP color support. The GeForceFX from NVIDIA improved upon this with full 128bit FP support and even longer shader support with DX9+ VS/PS 2.0+. With this new product lineup however ATI is here to up that even higher going beyond the GeForceFX in terms of Cinematic Capability.

ATi is introducing three new cards today, the

Radeon 9800/Pro, Radeon 9600/Pro and Radeon 9200/Pro

. The Radeon 9800/Pro is based on the codenamed R350. The 9600/Pro is based on the codenamed RV350 VPU and the 9200/Pro is based on the codenamed RV280 VPU.

Before we go into the specifics of each model lets take a look at the new technology that is implemented in this new product lineup. Basically you can say ATI took the proven R300 VPU and improved upon the design. In a nutshell, that sums up the R350 and RV350’s technology. There are not startling new technologies behind the core, or at least none they are talking about.

Cinematic Rendering

The R300 supports 96bit FP (Floating Point) precision, while the GeForceFX supports up to 128bit FP precision. The 9800 Pro still supports 96bit, 64bit, and 32bit FP pixel precision. ATi’s belief is that 96bit precision is going to be all that is necessary for the current product cycle. And they pretty much hit the mark. Even today I do not know any games that completely use all of the features including high precision FP color on any of these DX9 capable video cards. Now, hopefully future games will start to utilize these features, and when they do ATi is now ready for them with the R350 and RV350. With DirectX 9 and OpenGL 2.0 support, ATi provides a platform for next generation of games that we are told we will be playing in 2003.

DX9++

Nope, that "++" is not a typo. ATi supports "DX9++" in their new VPU’s. If you recall the 9700 Pro supports DX9, the GeForceFX support DX9+ and now ATi supports DX9++. To put it simply, this marketing teams one (or two or three) upping the other as there are no such + and ++ standards. The idea here is that these DX9 class cards have support for features beyond DX9. ATi’s DX9 capabilities now surpass the GeForceFX. They now support Floating Point 3D Textures, Floating Point Cube Maps, Multiple Render Targets, Displacement Mapping (HOS), and N-Patches in DX9.

What is most unique about their DX9++ implementation however is that they now support limitless pixel shaders.

SMARTSHADER 2.1 is their official marketing name for shader support. The 9700 Pro supports SMARTSHADER 2.0. SMARTSHADER 2.1 improves upon the DX9 SMARTSHADER 2.0 in the 9700 Pro. A new F-Buffer has been introduced into the R350 and RV350 that makes it possible to execute pixel shader programs with an unlimited number of instructions without the performance overhead associated with multi-passing the rendering pipeline. Do not confuse this with the R350 not having to work harder to accomplish a longer shader though. This also fully allows for OpenGL 2.0 fragment shader support. I will quote ATi’s whitepaper on this technology:

“All current graphics processors that support shaders provide a limited amount of resources, such as a maximum number of instructions, textures, and constant registers that may be used in a single rendering pass. Rendering complex visual effects can sometimes require exceeding these resource limits. The only way to render such effects is to break them up into separate passes, such that no one pass exceeds the resource limits. Pixels are then passed through the entire rendering pipeline multiple times until all the necessary shader operations have been performed.

Unfortunately, this methodology can be slow and inefficient. The process to render a pixel in modern 3D applications is very complex and requires numerous steps, including vertex processing, backface culling, triangle setup, texture sampling, pixel shading, stencil testing, Z testing, anti-aliasing, and more. Repeating all of these steps multiple times per pixel is largely redundant, since the only thing that is changing on each pass is the pixel shader. Additionally, each pass requires the entire image to be read from, and written back to, the frame buffer, even if only a few pixels require multi-passing. Worse still, this technique does not work well with transparent or translucent surfaces, since the foreground and background colors must be blended before writing to the frame buffer, and cannot be stored separately between passes.

SMARTSHADER 2.1 solves all of these problems using new F-Buffer technology. The F-buffer, short for “Fragment-stream FIFO buffer”, is the first hardware implementation of an idea proposed in a paper by graphics researchers William Mark and Kekoa Proudfoot at Stanford University (http://graphics.stanford.edu/projects/shading/pubs/hwws2001-fbuffer). It handles complex effects by passing only the pixels that require multi-passing through the pixel shader engine multiple times, while doing all the other steps in the rendering pipeline just once per pixel. Multi-pass pixels are stored in the F-buffer between passes, rather than writing them out to the frame buffer, so transparent pixels can have foreground and background colors stored separately. This technique saves rendering time, reduces memory bandwidth requirements, and gracefully handles transparency.“

• « Previous Page
• » Next Page
• e-mail this aricle

Select Page1 - Introduction 2 - Introduction Continued 3 - Unreal Tournament 2003 4 - Unreal Tournament 2003 Continued 5 - Serious Sam 2 6 - Jedi Knight 2 7 - 3DMark03 8 - Code Creatures, Fill-Rate Tester, Dolphin Demo, SpecVIewPerf7 9 - ShaderMark v1.7 10 - Image Quality: FSAA Tests 11 - Image Quality: UT2003 12 - Image Quality: NFS HP 2, Texture Filtering 13 - Overclocking, Conclusion