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

sorry, this just sounds really really funny.

 

I would guess on jvd's behalf that he meant to explore what shaders are being replaced. Just a look and see.

 

I think jvd was just trying to derail the point at hand, which was the Humus tweak giving only very small (1-2fps) performance improvements over App-controlled AF, as used by the HardOCP benchmarks.

In other words, the Humus hack offers no significant performance improvement or "gap closing" as digitalwanderer speculated, over the HardOCP benchmark figures since they ran with AppAF. All the Humus hack did was allow people who were running with ATI CP-forced AF to run at the same rate as app-controlled AF without the AF bug.

Humus' hack didn't improve performance significantly over what we've seen from the HardOCP benchmarks.

 

That would make more sense. It's kind of confusing though, given that Doom was designed from the ground up with nvidia hardware in mind the shader replacement shouldnt be that extensive. And another thing, if these replacement shaders are more nvidia friendly but mathematically equivalent then maybe ati should consider a similar approach but that's opening a whole new can of worms I guess.

 

No. I'm saying if you want to run ati cards with out shader replacements then you should run nvidia cards with out shader replacements .

Users in this thread are showing decreases in performance because of this minior changes. It obviously breaks nvidia's shader replacements .

I have a 6800gt and I want to run doom3 as all other cards will run it. Not as nvidia wants me to run it so they look good on a benchmark

 

a 1.5 fps difference "obviously breaks nvidia's shader replacements"??? Ok there

 

I'm not talking about your benchmarks. Ohter users on about page 23 i think report 10fps loss with a 6800 because of this tweak.


We should have a fair lvled playing ground. Both ati and nvidia should be tested with out shader replacements .

What I want to know is why aren't websites doing this .

After all a whole bunch of sites dug into ati and disabled things to try to get nvidia style af done on the x800s.

Why aren't htey in favor of a fair playing ground here ?

Why aren't you ?

 

Nobody is saying that they want to run without replacements on ati hardware. The point is that it may not be necessary if the application-controlled AF issue is resolved.

Yes and this is completely irrelevant to the ATI/AF issue and just adds noise to this discussion.

Last time I checked people wanted the same output from their videocard. No card will run doom3 'as all other cards'. Why do you care how nvidia or ati generate an image if the images are equivalent. Now you're just making shit up to bitch about.

 

Since when is "artifacting" not an IQ issue?... Others in this thread have said such artifacting stems from D3 Application AF control not treating enough texture layers/stages with AF. If that's not your position, then what is your position relative to the cause of the artifacts?

Right--which plainly indicates the problem is with D3 application controlled AF, and not with Humus' tweak... We already know the Catalysts can do the proper AF texture-stage treatment in D3 because of the good results we get from Cpanel AF override control--which proves that the drivers are capable of proper AF treatment in D3. This brings us back full circle again to the prospect that the fault lies in the D3 engine with respect to the instructions it issues to the Catalysts for AF treatment when the Catalysts hand over control of AF to the D3 engine, which happens when you set the Cats to App Preference.

As far as your constant reliance on a particular set of "benchmarks" goes, you might care to entertain the prospect that during his testing Humus has engaged his own set of benchmarks, which provide results which have led him to his own conclusions about these matters. Considering the source of both sets of benchmarks, and considering my own positive experience with Humus' tweaks thus far, I have no doubt as to which of the two sets I find most credible...

In fact, your sole purpose seems to be to attempt to explain away or obfuscate the impact of Humus' tweaks, whereas the purpose for Humus seems clearly to be that of helping R3/4x0 owners get the kinds of performance and IQ possible from D3 with their existing drivers and hardware but not directly supported by the shipping D3 configuration settings. It's very difficult to find any sort of fault with Humus' efforts to date. On the contrary, I think he is to be applauded as opposed to trivialized.

 

No, that was me posting that result, trying a different shader mod on interaction.vfp that resulted in the 6800 having do do an extra pass to complete, hence slower performance - about 7fps. When using the Humus hack, I get around a 2fps decrease.

Better test with NV3X then, because this is where most of these took place to compensate for removing the NV3x path from Doom3

roffels. the "nvidia style of AF"? ATI fans like yourself should thank their lucky stars HardOCP used application-controlled AF for their benchmarks because by doing so they were able to get the same results as using the humus hack. If they didn't, the scores would be even lower. If anything, HardOCP was being kind to ATI, as their app-controlled AF benchmarks are representative of what people are getting with the "Humus hack" as evidenced by the benchmarks here.

 

Not necessarily. We don't have a direct comparison of the performance hit of forced AF on 6800/x800. Who knows, maybe the x800's would've gained some ground.

 

Does Nvidia have the some problems with AF where if you set it in the game the AF isn't as good as if you set it in the control panel?

 

Ditto!

 

Nope.

 

No it was not you. I will go search for it to shut you up .

I guess you can't read then big guy.

I'm asking why there isn't an investigation on nvidia hardware performance and shader replacements like there were with ati's af .

Learn to read please .

I think nvidia fans should thank thier lucky stars no one has done that.

 

Nah I've tried it briefly, app AF does't give me any significant fps increase with my 6800GT, though I should probably try again since someone reported an FPS increase. We know ATI loses ~15-20% performance with CP-forced AF. I'll try it now and let you know.

 

arrrr trylinear

 

I, for one, would like to know the shaders that are being replaced and what they are being replaced by. It'd give a better picture as to where the NV3x/NV4x are weak...good to know things..

 

We do?

 

but they gain back that performance loss with the humus tweak..

I assume you have a 6800... does the game look the same when AF is set in-game and when AF is set in the CP? (screenshot comparison)