How to boost your Doom3 performance by 40% on ATI hardware

doesnt matter when HL2 comes out its all over. gg ATI !

 

And you are speaking with the full authority of [H] I take it?

 

Or maybe

A) the game uses tons of shaders

 

I want to know how they get the nvidia card to work with out drivers ?

I mean after all they are just hacks and tweaks built into the damn things .

 

I dunno, but for example HardOCP continued to use the 60.72 drivers for nearly every review (save one I think) until official Nvidia drivers were out. Also, they wouldn't bench FarCry 1.2 until the patch was publically released despite reviewers having it a week or two early. OTOH, they did bench Doom3 with 4.9 beta drivers.

 

well the patch was taken back. SO i think that was a good move by h . Same with nvidia drivers.

Better than anand using 4.5 drivers when 4.7s are out

 

Was ATI getting recent builds to do such tweaks
I know some people also playing with low end systems, but they are not enjoying it much.
Has it come to the point where IHV's have to babysit developers..even have them write code for them, after emails with STALKER and seeing the lack of effort put into making modern cards run better (something Carmack always is preaching..better hardware)..it seems that way.
This isn't even about just ATI cards, the current generation of Nvidia cards could benefit from such tweaks.

 

Yeh, Nvidia is affected by the same problem, but to a lesser extent (~10% instead of 15-20%). That is, using Application AF is 10% faster than CP-AF - your tweak is ~10% slower on Nvidia hardware. However, if ATI was able to fix the IQ bug when using Application AF (looks like bilinear filtering instead of trilinear or something - whatever it is its not there with Nvidia cards), ATI users would be able to use Doom3 with its original shaders with similar performance to your "tweak" and the HardOCP benchmarks by using Application AF. That's probably worth looking into, as its likely something that can be fixed more easily by ATI instead of having id software write a new path for ATI cards. Likewise, fixing this IQ bug would likely benefit other games that use Application AF in the future in a similar way to Doom3. Overall sounds like the best fix.

 

It was application controlled AF. Notice how they used 8xAF in particular instead of 16xAF. 8xAF is automatically enabled in-game when you switch to High Quality mode and this was thoroughly talked about in both D3 reviews.

ATI and nVidia representatives were also both present for the benchmarks and if there was a major IQ issue on the ATI cards with 8xAF enabled in-game you can be damn sure that one of them or the HardOCP reviewers would of said something about it.

Does anyone actually have some side by side screenies of an X800 using in-game AF vs CP AF?

There is a difference between a blatant HACK to the game code done by someone in their "free time" (yea right lol) on a forum and official drivers released by the cards manufacturer.

If there is something that needs to be fixed it will be addressed in a DOOM 3 patch or ATI driver update soon.

 

LOL you are busting me up! [H] posted DOOM 3 benchmarks months before the game shipped, yet they balk at using a FarCry patch before public release? That's rich!

Maybe you should find a better source of info?

-FUDie

 

MONTHS before it was released? lmao

Try about TWO WEEKS.

http://www2.hardocp.com/article.html?art=NjQy

And that was the first DOOM 3 article.

That was when the second one was posted.

And these were OFFICIAL benchmarks done at ID Software with nVidia and ATI reps present.

You guys really crack me up lol.

 

First one huh?

http://www.hardocp.com/article.html?art=NDc0LDI=

right...

 

Sit down, fool.
http://www.hardocp.com/article.html?art=NDc0LDE=

Forgive me, that was over a year ago, which certainly implies that months went by.

-FUDie

 

LOL, no. That means the POW function I tested causes a hit (10 cycles in NVShaderPerf vs 9 cycles with the default specular lookup). It's actually pretty funny reading your posts because they're so dumb.

The later Humus replacement (MAD_SAT/MUL) has a very small hit or gain, +/- 2fps. NVShaderPerf gives that replacement 6 cycles, but at a higher GPU utilization.

Please keep entertaining me. You're a riot.

 

Nothing more then a demo provided by ID Software tested with a 5900 Ultra and a 9800 pro.

That review has absolutely nothing to do with the two official reviews that were done at ID Software with the nVidia and ATI reps present with a finished copy of the game using the latest drivers and hardware.

 

nevermind, you're talkng about the performance preview id gave.

 

Exactly, yet it was done by [H].