AquaMark3 GeForce FX 5900 Ultra Vs. Radeon 9800 Pro

Article Index

AquaMark3 GeForce FX 5900 Ultra Vs. Radeon 9800 Pro - Page 5


AquaMark3 - GeForce FX 5900 Ultra Vs. Radeon 9800 Pro
DX8/DX9 Pixel and Vertex Shader Benchmarking

By: Dave Atlavilla
September 15, 2003


AquaMark3 Performance Measurement Test
Lighting, Shading  and Fill Rate

The final AquaMark3 test we ran proved to be the most interesting to us.  Below is a screen capture from the AquaMark3 "SVIST" Shader Visualization Technique test.  Again, red areas denote DX9 shaders, yellow areas are DX8 and blue have no shader effects.  This test is one of the tests in the Pixel Performance Measurement test we've graphed for you below.

Shader Visualization Technique

Here's what the folks at Massive have to say about the Pixel Performance test.

The result of this test is the average number of drawn pixels per second (which is written into the result files). It?s kind of a ?real life fill rate test?, with the difference that this fill rate test contains complex pixel shaders as well as different texture and blending modes. The test consists out of two individual runs.

? The first run creates the overdraw of each frame. Each frame will be measured and a sum will be created of how many overall pixel have been drawn in this frame.

? The second run is a normal benchmark run and the results of the first run will be taken to calculate the drawn pixel per second.


Oddly enough, this test liked NVIDIA's Detonator 45.23 drivers a bit better than the Det 51.75 drivers.  We could speculate that this has occurred perhaps as a result of the recent effort, that went into optimizing drivers for Half Life 2 DX9 performance.  However, that's pure speculation and we like to work more in the factual here, so feel free to interpret your own perspective.


What can we say about AquaMark3 and our quest for a solid DX9 benchmark, to complement what is soon to be released by the team at Valve Software?  Well, unfortunately AquaMark3, in our opinion is not the answer.  Now before folks start drafting a rant-mail to the editor here, allow us to put a few things into perspective.  There is no question AquaMark3 is a solid and useful benchmarking tool.  As a matter of fact, with its basis on a real game engine, and liberal use of DX9 shader effects, it's a pretty good indication of what current generation mixed mode DirectX gaming performance will be like on a given graphics solution.  However, if you want to really measure what "pure" DX9 performance metrics could be for a video card, Half Life 2, with its resounding Microsoft endorsement as a perfect DX9 specimen, is much closer to reality in our view.  On the other hand, we'll most likely include AquaMark3 as a general DirectX benchmark, for future graphics articles here at HotHardware.  AQM3 is a highly configurable and stable benchmarking platform for a current DirectX gaming experience.  However, as we all know, DX9 shader technology in next generation game engines, will become even more pervasive in the months ahead, so there's a lot more to the NVIDIA and ATi story that needs to be told.  Not to mention what that slightly influential Carmack fellow is cooking up in OpenGL.

Comments, questions, rants?
Let us know in the HotHardware Forum!



Tags:  Radeon, GeForce, Ultra, force, fx, 980, pro, ULT, AMA, AR, AM, K

Related content