Check out this thread

[Keysplayr]

Oh Wow!! Yeah your right. That does look like A$$.

 

[Jeff7181]

I'm going to go through all this again one by one cause as I did it this time it was a bit unorganized... I'll be posting more screen shots after I'm all done testing the stuff. I'm thinking that heavy pixelization may be because I'm running medium detail and it using compression for specular.

 

[kami]

[qWell, the tweak "works" in such that it gives you free AF. Whether it is actually doing AF or not is another story[/quote]
But this is with the card forced to AF in the control panel, I don't think Doom can override that, can it? Plus it looks a lot better than when I used no AF, but I haven't done any comparison screenies yet.

 

[Jeff7181]

Ok... ignore my previous posts, lol.

Here's some screenshots...

High Quality, Normal Shader, 42.9 FPS
High Quality, Tweaked Shader, PARAM specExp = { 16, 0, 0, 0 }; , 45.0 FPS
High Quality, Tweaked Shader, PARAM specExp = { 8, 0, 0, 0 }; , 45.0 FPS
Medium Quality, Normal Shader, 47.1 FPS

When I took those other screenshots before, I forgot that using Medium detail settings would compress the "specular maps" or whatever you call it.

Now that that's out of the way... if you compare the first three I have posted here, you can see that with that value set to 8, the image quality is almost exactly the same, and it runs about 2 FPS higher, which is a 4.6% increase in performance.
With the setting at 16, it was a little too bright I believe. And "Humus" in the Beyond3D forums mentioned that as well, that he normally uses 16, but Doom 3 being so dark, they may have used 12 or 8... looks like 8 to me.

 

[Turkish]

This is gonna sound stupid but my "base" folder withing Doom 3 directory doesn't have a folder named "glprogs"? Am i supposed to create it?

 

[CaiNaM]

CATALYST 4.7's ----- CATALYST 4.9's ---- w/ Humus' Mod
2xAA 4xAF

1024 ------ 51.1fps ------------ 55.8fps ------------- 63.5fps
1280 ------ 38.4fps ------------ 39.8fps ------------- 47.9fps

4xAA 8xAF

1024 ------ 40.6fps ------------ 42.2fps ------------- 48.4fps
1280 ------ 29.1fps ------------ 29.8fps ------------- 34.6fps

All gfx options enabled; Settings on HighQ

 

[Noid]

EDIT _____________ this post refects an improperly applied tweak

SO IGNORE THIS ... (read my next post)

(I did see good results)

Same here.

Re-tested .. no gain. (r9700pro)
Since I cant play using HQ in 10x7, I tested using 8x6 HQ.

8x6 HQ no AA, vSync ON

CP - app pref AF = 51.3 fps
CP - forced 2 AF = 51.7 fps
CP - forced 4 AF = 48.5 fps
CP - forced 8 AF = 45.3 fps
CP - forced 16 AF = 43.4 fps

 

[PhoenixOrion]

Originally posted by: XietyCOM
This is gonna sound stupid but my "base" folder withing Doom 3 directory doesn't have a folder named "glprogs"? Am i supposed to create it?

Inside base folder is a pak000 pk4 file. Open that and there is a glprogs folder. backup your entire pak000 pk4 file before replacing the interaction.vfp file within the glprogs folder.

Now my x800pro runs with elevated min fraps.

Both x800pro and my 6800gt still has that microsecond stutter whenever multiple monsters spawn in a room. hmmm...even with the cachemeg tweak inside the doomconfig plus humus vfp tweak.

 

[Terranboy]

Don't judge it by the timedemo. I show only a minor increase there. But in the actual game I a running highQ (instead of medium) and 4xAA (none before) with no trouble at all, it actaully runs better than on the lower settings. Very pleased. ( again, AthlonXP2600 Radeon 9800Pro)

 

[Jeff7181]

I just tested the last modified shader posted by Humus and got 44.6 FPS, but the image is reproduced most exact.

Here's a screenshot... "Humus Final"

Compare it with the Normal one and there's very little difference... look at the texture inside the elevators since they're not effected by the lighting outside the elevator.

*EDIT* With 8XAF forced in the Control Panel I got 44.0 FPS.

 

[Noid]

Originally posted by: PhoenixOrion

Originally posted by: XietyCOM
This is gonna sound stupid but my "base" folder withing Doom 3 directory doesn't have a folder named "glprogs"? Am i supposed to create it?

Inside base folder is a pak000 pk4 file. Open that and there is a glprogs folder. backup your entire pak000 pk4 file before replacing the interaction.vfp file within the glprogs folder.

Now my x800pro runs with elevated min fraps.

Both x800pro and my 6800gt still has that microsecond stutter whenever multiple monsters spawn in a room. hmmm...even with the cachemeg tweak inside the doomconfig plus humus vfp tweak.

Umm ,,, maybe that's why its not working for me ...

So, we're supposed to :::: ???

1. backup pak000.pk4
2. unpack the pak000.pk4 file
3. replace the interaction.vfp file
4. Re-pak the pak000.pk4 file

Then test ?

 

[Jeff7181]

Originally posted by: gururu
the post tweak no POW does look darker, but not any worse than the no tweak with specular, particularly when gamma is bumped up just a notch..
the 'with POW' looks like crap. So I don't understand why you are tweaking the paths. Why aren't you using the 'fixed patch' vfp file?

What's the "fixed patch" vfp file?

 

[gururu]

Originally posted by: Jeff7181

Originally posted by: gururu
the post tweak no POW does look darker, but not any worse than the no tweak with specular, particularly when gamma is bumped up just a notch..
the 'with POW' looks like crap. So I don't understand why you are tweaking the paths. Why aren't you using the 'fixed patch' vfp file?

What's the "fixed patch" vfp file?

Hey Jeff7181,

I don't know which you were using, you just confised me because you were playing around with the POW settings.

the one I used

gururu

 

[Jeff7181]

Originally posted by: gururu

Originally posted by: Jeff7181

Originally posted by: gururu
the post tweak no POW does look darker, but not any worse than the no tweak with specular, particularly when gamma is bumped up just a notch..
the 'with POW' looks like crap. So I don't understand why you are tweaking the paths. Why aren't you using the 'fixed patch' vfp file?

What's the "fixed patch" vfp file?

Hey Jeff7181,

I don't know which you were using, you just confised me because you were playing around with the POW settings.

the one I used

gururu

Yeah, I started messing with it before I got to the end of the 31 page thread =)

 

[gururu]

the man speaks.

The lookup table is constant in Doom, so there isn't any real strong argument against replacing it with code. The lookup table was faster than doing the exact sequence of math ops that the table encodes, but I can certainly believe that a single power function is faster than the table lookup.

my interpretation is that he doesn't dispute the claims. That is, instead of performing a 'Lookup' , hardware can employ a 'power' function (or mathematical calculation) to draw out certain elements of a scene. If the hardware does this fast enough, it can end up being faster than performing the 'lookup'.

whatever the heck any of this means beats me.

 

[Jeff7181]

Originally posted by: gururu
the man speaks.

The lookup table is constant in Doom, so there isn't any real strong argument against replacing it with code. The lookup table was faster than doing the exact sequence of math ops that the table encodes, but I can certainly believe that a single power function is faster than the table lookup.

my interpretation is that he doesn't dispute the claims. That is, instead of performing a 'Lookup' , hardware can employ a 'power' function (or mathematical calculation) to draw out certain elements of a scene. If the hardware does this fast enough, it can end up being faster than performing the 'lookup'.

whatever the heck any of this means beats me.

The way I understand it is the original shader has instructions to look up a value on a table rather than do the math itself. While that would be faster on slow hardware, the R3xx and R4xx cores have the power to actually do all the math faster than it can look up the value in a table. So... in a VERY simplified explanation, if it needed the result of 4x4, the original shader would look up the value on a multiplication table and see that it's 16. This modified shader would actually make the GPU do the math and ad 4+4+4+4 and get 16.

*EDIT* So in effect, Carmack took the need for nVidia to optimize away, he bypassed the weakness of nVidia's GPU, it's number/pixel processing power... many people are saying using this tweak on nVidia hardware makes it slow, which makes perfect sense since NV30 and NV40 don't have the raw computational power that the R3xx and R4xx have (respectively). NV30 and NV40 aren't doing the math... they're looking up the answers... using a cheat sheet so to speak. There's nothing wrong with that... I bet Carmack did that to make D3 run on such a wide variety of hardware... you don't need the raw number/shader crunching power of the R3xx and R4xx because it's not constantly doing that where as Far Cry relies heavily on shaders... which apparently doesn't use those same type of lookup tables, which is why if you don't have at least a 9800 or FX5900 you can pretty much forget about Far Cry with PS2.0.

 

[Noid]

Originally posted by: Noid

Originally posted by: gururu

Originally posted by: Avalon

Well, the tweak "works" in such that it gives you free AF. Whether it is actually doing AF or not is another story

actually, in my case, the results with AF(tweak) are higher than no AF (no tweak).

Same here ...

Re-tested .. no gain. (r9700pro)
Since I cant play using HQ in 10x7, I tested using 8x6 HQ.

8x6 HQ no AA, vSync ON

CP - app pref AF = 51.3 fps
CP - forced 2 AF = 51.7 fps
CP - forced 4 AF = 48.5 fps
CP - forced 8 AF = 45.3 fps
CP - forced 16 AF = 43.4 fps

I did get a increase ! :~
(after properly applying tweak)

_________before ________ | __after __
CP - app pref AF = 51.3 fps - - 52.2 fps
CP - forced 2 AF = 51.7 fps - - 53.3 fps
CP - forced 4 AF = 48.5 fps - - 52.5 fps
CP - forced 8 AF = 45.3 fps - - 52.0 fps
CP - forced 16 AF = 43.4 fps - 51.9 fps

Soooo .... I can also run forced 16 AF WITHOUT fps loss.
(or I may use forced 2 AF)

 

[Jeff7181]

Now if only Carmack would work on the surround sound issue with a lot of Soundstorm and Audigy cards...

 

[CaiNaM]

Originally posted by: Jeff7181
Now if only Carmack would work on the surround sound issue with a lot of Soundstorm and Audigy cards...

using "surround", my sound quits after a few min.. is this a widespread "bug"?

 

[Brian48]

Originally posted by: Jeff7181
Now if only Carmack would work on the surround sound issue with a lot of Soundstorm and Audigy cards...

Just out of curiosity, what issue is this? I only have a 4.1 so I don't actually have a "surround" setup per say, but I'm currently D3 with sound coming out of all four speakers (sans CMSS). It's probably only 2xstereo and not actually true 4.1, but it doesn't bad at all.

 

[Noid]

Surround Sound ON + soundstorm = loss of sound during gameplay or selecting menu
(clicking surround sound button ON and OFF brings it back)

So, I dont use 'surround sound ON'

Back on subject...

Maybe those not getting results,,, didnt apply the tweak properly. (like me)
OR ,,, are comparing using settings lower than HQ in game.
(I noticed no gain using medium settings)

 

[Jeff7181]

Originally posted by: Noid
Surround Sound ON + soundstorm = loss of sound during gameplay or selecting menu
(clicking surround sound button ON and OFF brings it back)

So, I dont use 'surround sound ON'

Back on subject...

Maybe those not getting results,,, didnt apply the tweak properly. (like me)
OR ,,, are comparing using settings lower than HQ in game.
(I noticed no gain using medium settings)

It's not specific to Soundstorm... I'm using an Audigy 2 ZS and the sound cuts out just like CaiNaM.

It seems as though it's very widespead... I've seen lots of discussion about it, so far the only solution I know of is to turn off surround, or turn audio acceleration in the control panel down to the 2nd knotch, which basically does the same thing as turning off surround sound.

Oh... and for the record... Far Cry's surround doesn't work as it should either... Sometimes I can't even hear my gun being fired.

(seems like developers are so worried about pretty graphics they don't spend enough time on the audio)

 

[PhoenixOrion]

Lol. b3d thread is now up to 33 pages.