Halo Tweak Guide
Posted: 2003-11-08 by Achilles |
The biggest jump in performance is to drop from PS 2.0 down to PS 1.4 / 1.1. Going down to 1.4 loses very little visually but will yield a large perfomance difference on some cards while going down to 1.1 loses a bit more but Halo still looks very good. Here are some results from in house testing with a few different cards
| Card | PS 2.0 | PS 1.4 | Gain |
| 9500 Pro | 34 | 39 | +15% |
| 9600 XT | 27.40 | 29.46 | +7.5% |
| 5950 | 50.87 | 49.12 | -1% |
All the cards were benchmarked at 1024 x 768 on a P4 2.8. The performance gains vary quite a bit with the 9500 Pro having the largest benefits and the 5950 actually running a tad slower. A drop down to the fixed function rendering path results in huge drops in visual appeal and is not recommended unless it is absolutely essential. Next to go would be shadows and specular. Shadows are nice but in darker areas (and there are many) they are very subtle and do not add much. Lower quality textures are acceptable too; the texturing in Halo is done well and does not suffer too much although the effects are noticable with grainier textures especially up close. Particles are the one thing that should stay on because the effects are very apparent and add a lot of visual flair. Grenades just are not as menacing when the explosion is not there.
|
The differences between PS 2.0 and 1.4 are not too striking. We were had pressed to find much difference between the scene at all. |
Halo may not be the game that everyone expected when it was first announced years ago but it is a quality title and there is no doubt that Halo will be a cornerstone of video card benchmarking. As a game, it does have its shortcomings but it offers a bevy of tweaking and benchmarking features for the enthusiast to play with. The timedemo feature allows a consistent scene to be used across all systems and yields a lot more information than just a simple average framerate and will be useful in analyzing the strengths of different cards as well as comparisons between the efficiencies of various cards under different rendering pipelines.


