Doom 3 shader mod for ++ performance

[101998]

http://www.beyond3d.com/forum/viewtopic.php?t=14874

Check that thread out. Humus is a damn genius. I have not done and benchmarks personally but the game def feels faster. Should help most of the ati cards and the n3x cards, the 6800 series really doesn't have much room for improvment in the game but I guess you could always try it if you wanted.

Please refrain from any comment that contiains the letters "G" and "T" togeather, lets try and keep a civil discussion about the custom shader. While I would like this thread not to erupt into a flame war I'm sure it will. *sigh*

 

[trungracingdev]

i'd really like to take a look...but the site seems down

 

[bobsaget]

yeah, site down or something for me too...

So how much could this help for 9800pros/xts?

 

[101998]

I picked up Doom3 today and let be begin by saying it's a kickass game so far. A few minuses like weapon reload (which I find add nothing to a game, except annoyance, so I don't know why many devs keep adding it to their games), but overall much above my expectations.

Anyway, to the fun part, exploring the technology.
I think I've found the source of why this game runs comparably slow on ATI hardware vs. nVidia at the moment, and found a solution to the problem.

First, open your doom3\base folder. Doubleclick on the pak000.pk4 file. In the "window can't open this file .. .bla bla" dialog, go on and associate the file with an app like WinRar. With this file open in WinRar, go to the glprogs directory in the file. In there you'll find the shaders. The interaction.vfp file seems to be the main rendering shader. Altering this shader to output a constant color turns most objects into that constant color, except for stuff like computer screens etc.

So doubleclick the interaction.vfp file to open it (you may have to associate the .vfp extension with a text editor like notepad or wordpad first since we're going to edit the file). Scroll down to the fragment shader. You'll find these rows:


Code:

PARAM subOne = { -1, -1, -1, -1 };
PARAM scaleTwo = { 2, 2, 2, 2 };



Add this right below them:


Code:

PARAM specExp = { 16, 0, 0, 0 };



Now scroll down to this:


Code:

# perform a dependent table read for the specular falloff
TEX R1, specular, texture[6], 2D;



Comment out that line by adding a "#" to it, and add another line that will do the same thing with math instead, so it should look like this:


Code:

# perform a dependent table read for the specular falloff
# TEX R1, specular, texture[6], 2D;
POW R1, specular.x, specExp.x;



Save the file and close your text editor. WinRar will ask if you want to update the file in the archive, select yes. Close WinRar and enjoy about 40% higher performance in Doom3. Haven't done extensive testing yet, but my performance went from 34fps in 1280x1024 to 48fps.

Conclusion and discussion:
I don't want to complain about Carmack's work, I still consider him to be the industry leader in graphics engines. Though when I read the shader it striked me how many texture accesses it did compared to the relatively short shader, even for stuff that could just as well be done with math for a small cost in instructions. Using a dependent texture lookup for POW evaluation makes a lot of sense for R200 level hardware due to instruction set limits, but for R300 and up it's much better to just spend the three cycles it takes to evaluate POW with math instead of risking texture cache trashing with a dependent texture read, which may be much more costly, especially since the access pattern in this case will be far from linear. Also, using math improves the quality too, even though it may not be very noticable in this game.

I should point out though that I'm not sure if the constant specular factor 16 that I chose is the one that the game uses, so output may be slightly different, but if this solution will be worked into the game in a future patch, then this is easily configurable by the game so that there won't be a difference, except a lot faster.

An interesting follow-up discussion may be why this dependent texture lookup is much slower on our hardware than on nVidia. Maybe there's an architectural difference that's to blame, or maybe something else? The main point here though is that this should be good enough proof that ATI hardware can run Doom3 just as good if not better than nVidia, and that we can pass on all the "ATI suck in OpenGL", "ATI's drivers suck" etc. into the trashcan where it belongs.

There you go.

 

[101998]

yeah, site down or something for me too...

So how much could this help for 9800pros/xts?

People posting benchies on beyond3d were getting 10-15% increase it looked like.

 

[^eMpTy^]

/me puts on his nvidia !!!!!! hat.

ok...humus is cool...I've played with his demos...I've read his posts...he seems to be a cool guy and his demos are sweet...but this is just horse-shit...it's a hack...it causes artifacting at medium quality and the average fps boost was about 3 or 4 fps...one guy said 10fps but didn't give much detail about his settings...

Even if it did work it's nowhere near enough to make them competitive in D3...this post by Humus is just sad...he works for ATi and has probably been tearing through the code trying to find a way to speed it up on ATi cards since the second he got the game...frankly I think it's pretty lame of him to make a post like this without fully testing it...

The main point here though is that this should be good enough proof that ATI hardware can run Doom3 just as good if not better than nVidia, and that we can pass on all the "ATI suck in OpenGL", "ATI's drivers suck" etc. into the trashcan where it belongs.

Even with his HACK ATi still sucks in OpenGL and especially sucks in Doom3...it's cool that he wants to give something to the ATi community...but this is a BS hack and his comments are irresponsible and just plain incorrect...

But hey, if you have a XTPE this 3-4 fps boost is enough to tie with the GT at some settings...

All you ATi guys will have to forgive me...I'm not trying to shit on your cards...however I'm totally shitting on Humus and his pathetic BS hacks...the part about putting comments in the trashcan is just plain sad...

 

[101998]

/ ATi still sucks in OpenGL and especially sucks in Doom3

Please explain. The performance is on par with other new d3d titles played on the same card, I have not heard of many stability issues with this game and ati cards.

Yes its a hack, your changing the shader code.
Yes the gt runs it at those speeds already, we know this so why shit on the thread?
Yes it's *only* a 15% increase in performance, but 3-4 fps increase overall can make a noticeable difference in gameplay

EDIT : Hell I think its cool that someone from ati actually browses forums and does shit like this.

 

[CrimandEvil]

Wow.....those guys are getting so excited over an average 3 frames increase. lol

 

[^eMpTy^]

Please explain. The performance is on par with other new d3d titles played on the same card, I have not heard of many stability issues with this game and ati cards.

Yes its a hack, your changing the shader code.
Yes the gt runs it at those speeds already, we know this so why shit on the thread?
Yes it's *only* a 15% increase in performance, but 3-4 fps increase overall can make a noticeable difference in gameplay

Because it causes artifacts...because he didn't test it or submit it to anyone for an official patch...he just posted it on the b3d forums to vent his own frustration with the title "How To boost your doom 3 performance by 40% on ati hardware"...which it doesn't do...and then he says "this should be enough to prove that we can run doom3 just as fast as nvidia if not faster" which again is bullshit because everyone testing it is only getting like 1 - 4 fps boost from it...

Basically this really pisses me off because the guy works for ATi and a lot of people on the b3d forums really look up to him...and he's feeding them a shit sandwich claiming it's gonna fix their doom3 woes...which it simply does not do...

He's making himself and ATi look bad by posting this bullshit...

 

[CrimandEvil]

I agree ^eMpTy^, this ATI/D3 stuff is getting fucking ridiculous. Looks like about a third of them are getting artifacting and the fact that this guy is talking as if everything is fine kind of gets me.

I would have expected him to atleast talk with Carmack about it first before he screws with his code and posts it for ATI owners to gobble up.

Yes, I believe so too. But the main point I tried to prove was that it was doable. If Carmack works this into the game, it should be fully equivalent. He can just pass the right specular exponent to the shader with constant. So while putting this in yourself isn't going to be the same thing, it will be the same thing if Carmack does it, and with the same performance boost.

So has anyone here tried it?

 

[101998]

Because it causes artifacts...because he didn't test it or submit it to anyone for an official patch...he just posted it on the b3d forums to vent his own frustration with the title "How To boost your doom 3 performance by 40% on ati hardware"...which it doesn't do...and then he says "this should be enough to prove that we can run doom3 just as fast as nvidia if not faster" which again is bullshit because everyone testing it is only getting like 1 - 4 fps boost from it...

Basically this really pisses me off because the guy works for ATi and a lot of people on the b3d forums really look up to him...and he's feeding them a shit sandwich claiming it's gonna fix their doom3 woes...which it simply does not do...

He's making himself and ATi look bad by posting this bullshit...

I didn't notice any graphical differences when I played (granted I only played for 10 min or so). As for the rest of your comments ... well hold your own opinion and I'll hold mine. Humus is a pretty smart guy maybe he got a little excited when he first posted. I know you would if you were in his place.

BTW you still have yet to respond to my opengl comments, or are you just trying to make ati look bad?

 

[^eMpTy^]

I didn't notice any graphical differences when I played (granted I only played for 10 min or so). As for the rest of your comments ... well hold your own opinion and I'll hold mine. Humus is a pretty smart guy maybe he got a little excited when he first posted. I know you would if you were in his place.

BTW you still have yet to respond to my opengl comments, or are you just trying to make ati look bad?

Look at the screenshots in that thread...just because you didn't see them doesn't mean they aren't there...

I didn't really understand what you were trying to say about OpenGL...you said that thier OpenGL was on par with their d3d performance...which didn't make sense...because it isn't on par...at this point it's pretty obvious that ATi has serious OpenGL issues...case in point, the XTPE loses to the $299 6800nu in Jedi Academy...

I'm not trying to make ATi look bad...if anything I'm pissed at Humus (who works for ATi) for making them look like whiny bitches by posting this silliness...his comments are in-excusable...

I realize he's only human and he probably was a little over-excited...but that's no excuse...he's a professional, he should act like one...and no I would not make the same mistake if I were in his place, I have more common sense than that...

 

[^eMpTy^]

I agree ^eMpTy^, this ATI/D3 stuff is getting fucking ridiculous. Looks like about a third of them are getting artifacting and the fact that this guy is talking as if everything is fine kind of gets me.

I would have expected him to atleast talk with Carmack about it first before he screws with his code and posts it for ATI owners to gobble up.




So has anyone here tried it?

If the title of the thread had been "boost your d3 performance by 10%" instead of "40%"...it wouldn't be so bad...I've slung enough mud in here already...that's all I have to say on the subject...

 

[tornadotsunamilife]

im pretty sure if it was the helpful carmack would of included it himself. however since there is been reported artifacting, then we can see that it is performance gain over visual loss. since humus aparantly works for ati he seems like he is destroying their rep from the inside. this is exactly what happened to nvidia last generation and they didn't come-up on top, i understand that ati themselves have not implemented this shader hack but the rep is still there on the line

 

[101998]

He posted the shader change as a member of the community not as an ati employee. People are posting various config changes to doom3 for better performance, this is just taking it a step further. As far as you not getting excited, look a couple posts up, maybe I intrupeted it wrong but you look rather ... upset.

The par comment is comparing opengl titles run on ati card x vs d3d titles run on the same card. You can use the same settings, ie in Far Cry the x800 can play at 1280x1024 2xaa 8xaf and get good framereates throughout the whole game and in doom3 you can play 1280x1024 2xaa/8xaf with good fps, that’s what I am getting at.

Doom 3 will run better on any nv4x based card because the game was taken into account when the chip was designed, that’s what nvidia decided to do, ati decided to go a different route. Does this mean nvidia has an unfair advantage? No, nvidia optimized ati didn't and they are now reaping the reward of their efforts. Ati choose to give their cards god like fill rate and almost 0 impact way of executing af, they reap the rewards of that with a 12pipe card having better fill rate and barely lagging behind a 16pipe card.

Maybe I started rambling a little. Hope it makes sense.

 

[gsboriqua]

Because it causes artifacts...because he didn't test it or submit it to anyone for an official patch...he just posted it on the b3d forums to vent his own frustration with the title "How To boost your doom 3 performance by 40% on ati hardware"...which it doesn't do...and then he says "this should be enough to prove that we can run doom3 just as fast as nvidia if not faster" which again is bullshit because everyone testing it is only getting like 1 - 4 fps boost from it...

Basically this really pisses me off because the guy works for ATi and a lot of people on the b3d forums really look up to him...and he's feeding them a shit sandwich claiming it's gonna fix their doom3 woes...which it simply does not do...

He's making himself and ATi look bad by posting this bullshit...

Well poking around there I noticed those with small increases had older cards. So Instead of spouting off about something I don't know of first hand I decided to load it up. I ran the timedemo and got an increase of 16 FPS on my x800XT. Thats a pretty impressive increase. I did not see any artifacts or the white pixels mentioned in that thread. Someone mentioned that if your at medium quality it shows up but not at max. I'll have to play a little more to see if I notice the white outlines mentioned...

 

[CrimandEvil]

I've noticed that those who say they got an extra XX frames are dropping the IQ quite a bit by running it without lighting and such so if people here are going to do this I hope you post screen caps and as much system info as you can

 

[tornadotsunamilife]

yes a realize he released this information as an individual. but i just don't think people should be told without knowing the consequences

 

[Xenozx]

i tried this on my 6800 ultra

in 1600x1200 with 4aa/8af on

i ran it once i first entered doom 3. then after the bench finished i ran it one more time to smooth out the cached data.

first test run was:
without the cheat i scored:
2148 frames in 67.5 sec
31.8fps

with the cheat i scored:
2148 frames in 65.5
32.8

second test run without cheat:
2148 frames in 57.5 sec
37.2 fps

second test run with cheat:
2148 frames in 60.5
35.5 fps

im sure all of that is within the margin of error.


also who ever has an ATI card and is talking 16 fps increase, I call BULL SH*T. no one but you seems to be reporting gains like this. on the beyond 3d forums a person with an x800xt was talking like 3-4 fps.

 

[wizzackr]

/me puts on his nvidia !!!!!! hat.

ok...humus is cool...I've played with his demos...I've read his posts...he seems to be a cool guy and his demos are sweet...but this is just horse-shit...it's a hack...it causes artifacting at medium quality and the average fps boost was about 3 or 4 fps...one guy said 10fps but didn't give much detail about his settings...

others reported a greater increase and NO artifacts whatsoever... did you try it?

Even if it did work it's nowhere near enough to make them competitive in D3...

excuse me, empty, but this sounds just like fan-boy comments to me - and that makes ME sad as you've presented yourself as a professional in all your other threads i've read so far.

this post by Humus is just sad...he works for ATi and has probably been tearing through the code trying to find a way to speed it up on ATi cards since the second he got the game...

so what? i say it's great! heck, hopefully nVidia-guys browse these forums, too, and hopefully they are "tearing through the code trying to find a way to speed it up on their cards". furthermore he didn't post as an ATI employee but as a private guy - i really see no problem in that. if someone posted a similar thing about nVidia HW i'd be just as excited and trust me: i wouldn't give a damn if it was an nVidia employee...

frankly I think it's pretty lame of him to make a post like this without fully testing it...

maybe he just has one of the systems that did not produce any artifacts as others reported...

Even with his HACK ATi still sucks in OpenGL and especially sucks in Doom3...it's cool that he wants to give something to the ATi community...but this is a BS hack and his comments are irresponsible and just plain incorrect...

as stated before: "Yes its a hack, your changing the shader code." - so what?!? isn't all driver optimizations changing shader code?

But hey, if you have a XTPE this 3-4 fps boost is enough to tie with the GT at some settings...

that's just being polemic - again as stated above "Yes it's *only* a 15% increase in performance, but 3-4 fps increase overall can make a noticeable difference in gameplay"

All you ATi guys will have to forgive me...I'm not trying to shit on your cards...however I'm totally shitting on Humus and his pathetic BS hacks...the part about putting comments in the trashcan is just plain sad...

empty - please give arguments for what you say. other than that i find this to be your first sad post, sorry.

 

[’m‚³‚ñ]

Ati vs Nvidia... What a waste of time. I'll go back to using my Performa with no 3d accelerator and playing solitare.

Anyway, I might try this when I get D3, but I might have a 6800 by then.

 

[gsboriqua]

also who ever has an ATI card and is talking 16 fps increase, I call BULL SH*T. no one but you seems to be reporting gains like this. on the beyond 3d forums a person with an x800xt was talking like 3-4 fps.

Actually I saw some posts with 12-13 fps increases somewhere in there. Im not going to get into an ATi vs Nvidia thing. I know what kind of increase it gave me. I have no reason to make up stuff. I assumed this thread was to help boost performance on ATi cards and I was just posting what I noted when I tried the hack/optimization/cheat whatever you want to call it.

 

[tornadotsunamilife]

but carmack would have done this himself surely if it was 100% stable no?

 

[Andrew87]

Just wondering if these people who noticed the artifacting have overclocked GPUs?

 

[zipspyder]

To the Nvidia fan boys...calling these shader hacks/cheats:

Isn't that the pot calling the kettle black...

Humus knows it may or may not cause some artifacting. If you actully read the whole thread the changes are doing the shader rendering a little different than normal so yes there may/may not be side effects. He's posting as a gamer trying to help other people with ATI cards get better performance.

If you don't like what he's doing why don't go and post over there and tell him. At least he'll have a chance to defend himself.

PS: I'm sure if ATI were to throw a bunch of money towards ID they could come up with some special "tweaks' to improve ATI's performance more also. Just ask Nvidia...

 

[MFZ]

Why are you so worked up over Humus providing a way for ATi users to get a performance boost? You sound jealous of him and insecure of your little card. Oh BTW, I got 0 artifacts with his code update. If you aren't an ATi user, piss off, it doesn't really concern you does it? If people have artifacts, they can roll back to the orig. code, no harm done. Oh yeah, get a life.

/me puts on his nvidia !!!!!! hat.

ok...humus is cool...I've played with his demos...I've read his posts...he seems to be a cool guy and his demos are sweet...but this is just horse-shit...it's a hack...it causes artifacting at medium quality and the average fps boost was about 3 or 4 fps...one guy said 10fps but didn't give much detail about his settings...

Even if it did work it's nowhere near enough to make them competitive in D3...this post by Humus is just sad...he works for ATi and has probably been tearing through the code trying to find a way to speed it up on ATi cards since the second he got the game...frankly I think it's pretty lame of him to make a post like this without fully testing it...



Even with his HACK ATi still sucks in OpenGL and especially sucks in Doom3...it's cool that he wants to give something to the ATi community...but this is a BS hack and his comments are irresponsible and just plain incorrect...

But hey, if you have a XTPE this 3-4 fps boost is enough to tie with the GT at some settings...

All you ATi guys will have to forgive me...I'm not trying to shit on your cards...however I'm totally shitting on Humus and his pathetic BS hacks...the part about putting comments in the trashcan is just plain sad...

 

[wizzackr]

but carmack would have done this himself surely if it was 100% stable no?

although i, too, think this is the case i still think this is some weired argument: you know - even einstein made some mistakes...

 

[ZenOps]

Hey, I'll take 3 frames anyday. It shouldn't affect visual quality much, it seems to pre-convert to a format that is more digestable to ATi.

Add it to the 2-3 frames that come from the 4.9 betas, and its getting pretty good. I'm *almost* thinking about Dooming 1280x1024 high quality on my Radeon 9700pro with P4C 3.2 oc'ed to 3.6 (1024x768 now). Maybe for a second time through the game.

Now that I look at it further, it does seem to make all textures "shiny". Not that thats a bad thing. I think it looks better (very slightly brighter) this way. Now, if the ambient lights could be turned up another 30 percent, it might just be a winner.

 

[YellowPeril]

i dont understand all the flaming. humus found a way to squeeze out more performance than stock. so what. empty, i notice that you have your gt overclocked, squeezing out those mhz you didnt pay for. and whats the difference between that and changing a file? also, it really does not matter who the hell he is or who he works for. and as for the artifacting issues, these as of now seem to be random and not based on anything specific, as people have been getting then with stock cards and fresh installs.

what really is pathetic here are the attitudes of people toward progress, and the immaturity in terms of brand loyalty. because you like nvidia doesnt mean that ati sucks, and that any good news for ati hardware should be spoiled by your flames.

 

[Tigerblade]

Is it just me thinking this, or is it ALWAYS an Nvidia !!!!!! who starts a flame war......as soon as a hack, mod WHATEVER comes out that actually increases performance with no loss of IQ, all of a sudden, the 2-3fps lead they did have in 1 game disappears......and its the end of the world....yet you still maintain the XT PE runs the game like shit, when it now runs it just like your GT......in this same logic, your GT now runs D3 like shit......but you wont say that will you.....

*Sighs in utter disbelief at the sad little boys*

 

[Andrew87]

I changed the "16, 0, 0, 0" to "4, 0, 0, 0" - I noticed higher numbers seem to give more white space and artifacts. I ran timedemo 1 and on the second run scored 41.5fps using Medium Quality at 1152x768 (no AA/AF and V-Sync Disabled). I think this 'tweak' and the 4.9betas (I just copied the atioglxx.dll to the Doom 3 root) seem to have gained me 1fps, which is a shame really.

 

[Andrew87]

as soon as a hack, mod WHATEVER comes out that actually increases performance with no loss of IQ

I'm no !!!!!! of either side (although I own a 9800Pro at the moment) but I believe this hack (at it "16, 0, 0, 0) does cause a loss of IQ. It whitens the textures up too much and causes artifacts, which to me ruins the game experience.

 

[Tigerblade]

I'm no !!!!!! of either side (although I own a 9800Pro at the moment) but I believe this hack (at it "16, 0, 0, 0) does cause a loss of IQ. It whitens the textures up too much and causes artifacts, which to me ruins the game experience.

Thats a bit of a generalisation don't you think? Not all ppl get artifacts......

 

[Andrew87]

More people probably experience the artifacting (snow) than have said so, but may not have noticed it due to various factors, such as brightness settings, how much attention they pay to spotting it etc. Then again, I can't be sure this is artifacting in it's proper sense of the term. The whole hack just seems to whiten things up, you especially notice it on structures which have a panel sort of structure. That's probably where the frame increase is coming from, making everything near enough the same colour (although I'm no programmer and can't really support my arguement so I'll stop here). Also those saying that they got 10-15fps increases, post some proof please. Play through Doom 3 for one minute with Fraps loaded, logging the fps to a text file without the hack, then do it again after the hack has been applied - completely exit Doom 3 and start a new instance of it to make the test fairer.

This whole 'optimization' has beeen exaggerated, my average framerate is about 40fps (see here for my ingame log.)

40fps x 1.4 = 56fps (so I should experience a 16fps increase, but I only had an increase of 1fps. It would be more accurate to claim that an upto 40% increase can be gained, although I don't think anyone has, or will manage that.

It really is poor that everyone believes these few lines of code will make ATI as competitive as NVIDIA in the current situation when it comes to OpenGL. My system even sucks at playing counter-strike using OpenGL if I compare it's performance to D3D. To see just how little progress ATI has made with OpenGL optimization, download the ATI Catalyst 3.6 drivers, run the expand command on the atioglxx.dl_ file and place it in the Doom 3 root directory. Now, run the timedemo with this 3.6 driver; there are some blue artifacts which ATI has since fixed (congratulations), but running the 4.9 driver vs the 3.6 driver shows only about an 8% increase in FPS - and the 4.9betas are supposed to be a 'Doom Hotfix'.

 

[nst6563]

I tried it...didn't notice much of a difference. I didn't show the fps, no timedemo, but the gameplay seemed the same as it was before. Also, didn't get any artifacts as were mentioned.

 

[rancor]

Actually I saw some posts with 12-13 fps increases somewhere in there. Im not going to get into an ATi vs Nvidia thing. I know what kind of increase it gave me. I have no reason to make up stuff. I assumed this thread was to help boost performance on ATi cards and I was just posting what I noted when I tried the hack/optimization/cheat whatever you want to call it.

It works on both cards in one of hte config files ya have to change the cache size settings. It helps more so on nV since as the frame rates go higher there is less stress on the cards plus more boost.

 

[Andrew87]

It works on both cards in one of hte config files ya have to change the cache size settings. It helps more so on nV since as the frame rates go higher there is less stress on the cards plus more boost.

What's even worse, is people are posting about frame increases which are UNRELATED to this hack. I wonder if those 10fps and above increases were from this hack or the cache setting. hmm....

 

[Tigerblade]

At the end of the day, it still looks hit hot on ATI's top hardware and is VERY playable....the over exaggerated 'D3 runs shit on ATI' is just plain and utter crap.....

I wouldn't worry about CS for much longer anyways mate, HL 2 will be blowing D3 to the side soon Then you'll have Source CS WOOT!!!

In fact I'd bet that most of the ppl saying it runs like shit on ATI are playing D3 on GF2's and GF3's..........again, playable, but why talk shit???

 

[gordon151]

It works on both cards in one of hte config files ya have to change the cache size settings. It helps more so on nV since as the frame rates go higher there is less stress on the cards plus more boost.

Uh, no it's not noted anywhere that you are required to change the cache size settings and no one with an nVidia card has reported any change in their fps as of yet.

 

[Francois_C_N]

With this mod, I got a 4fps increase in 1024 768, high quality, no AA on my 9800 pro. On the other hand, the 4.9betas didnt do anything to my doom...