• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Doom 3 shader mod for ++ performance

Status
Not open for further replies.
SnakEyez187 said:
Yay somebody gets it :p
Click to expand...

Who the hell cares about OEM's?

I was talking about the performance crown not how many cards got stuck in OEM machines.

There were probably plenty of FX 5200's stuck in OEM machines the last couple of years lol.
 
tranCendenZ said:
Only legit scores are ones used with official id software code, not shader iq/performance hacks.
Click to expand...

Considering John Carmack himself has responded to this saying why it works and even giving a clue as to how to improve the performance even more, I'd say that you have no idea what you're talking about. Making these changes is simply something Carmack chose NOT to do. His reasons are his, whatever they may be, and I'm not going to read anything into that. Software developers in general want to do as little additional work as possible, so in this case, Carmack made the rendering paths the same whenever he could. ALL software can be optimized for better performance on any hardware, it's just a matter of whether you want to spend the time doing it or not. To do it, nor not to do it, it doesn't make this optimization any less valid. It simply causes the paths to be even more divergent, and adds complexity when it comes to debugging.

Carmack's response also suggests that he knew this could be done, and chose not to. Again, his reasons are his reasons, and I'm not judging him.

Even further, all software are hacks. Finding elegant ways to make hardware do something is the very definition of a hack. Dismissing this as invalid because it's a "hack" or optimization is very ignorant.
 
Just an update for those of you that haven't been keeping up with the original B3D thread, Carmack responded:

Our specular function isn't a POW function, it matches the bias / scale calculations done on the NV10/NV20/R200 paths.
Click to expand...

With a response from humus as well:

This of course explains why it doesn't look exactly the same. This is also a good thing. Now we don't even need to use a POW function, but we'll be fine with a MAD_SAT, which make it even faster. Got 21% boost in the timedemo with this. Now I still don't know exactly what exponent he's approximating, or if he's using different lookup tables for different materials, but I've tried this and it seems to look the same, but our fellow artifact hunters may want to verify that.
Click to expand...

And then goes on to give a new bit of code to use instead of the old one:

MAD_SAT R1, specular, 4.0, -3.0;
Click to expand...

So replace the pow line with this new mad_sat one, and it's supposed to be even faster. :)
 
burningrave101 said:
Who the hell cares about OEM's?

I was talking about the performance crown not how many cards got stuck in OEM machines.

There were probably plenty of FX 5200's stuck in OEM machines the last couple of years lol.
Click to expand...

Who cares about OEM? Considering it's one of the largest markets, I'd say.. everybody.
 
NV10/NV20/R200 specular paths? Talk about coding for the widest chipset support.

I would think there would be plenty of room for improvement if that statement is true, for both Nvidia and ATi. There have been tons of advancements since those chipsets.
 
There would be tons of space for improvement if Carmack started using nV specific ogl extenstion. too bad ATi really doesn't have much in that catagory.
 
I was talking about the performance crown not how many cards got stuck in OEM machines.
Click to expand...

Maybe for one post, then you started fighting the fact that ATI was bigger, made more money, and had more marketshare back then (again due to its huge OEM presence, being basically the only market 8-15 years ago) , can't seem to find the post where you brought up the term "performance crown" before this one though

There would be tons of space for improvement if Carmack started using nV specific ogl extenstion
Click to expand...

I'm sure if he went down this road he would have had a seperate path just for NV cards, which he did, and then he threw out. Not sure if the ARB2 path contains any vendor specific code though
 
GeniusInABottle said:
Who cares about OEM? Considering it's one of the largest markets, I'd say.. everybody.
Click to expand...

Um no.

People that build their own machines dont care about OEM so why should something that gets stuck in OEM machines mean anything at all on a site like HardOCP?
 
ZenOps said:
NV10/NV20/R200 specular paths? Talk about coding for the widest chipset support.

I would think there would be plenty of room for improvement if that statement is true, for both Nvidia and ATi. There have been tons of advancements since those chipsets.
Click to expand...

rancor said:
There would be tons of space for improvement if Carmack started using nV specific ogl extenstion. too bad ATi really doesn't have much in that catagory.
Click to expand...

I think you guys just answered your own questions. The more advanced, or specific, he made his general ARB2 path, the less compatible it would have been with legacy hardware.
 
burningrave101 said:
Um no.

People that build their own machines dont care about OEM so why should something that gets stuck in OEM machines mean anything at all on a site like HardOCP?
Click to expand...

I was refering to nVidia and ATi caring. Which, believe me, they do.
 
People that build their own machines dont care about OEM so why should something that gets stuck in OEM machines mean anything at all on a site like HardOCP?
Click to expand...

How many people and companies were building their own computers 8-15 years ago?
 
SnakEyez187 said:
Maybe for one post, then you started fighting the fact that ATI was bigger, made more money, and had more marketshare back then (again due to its huge OEM presence, being basically the only market 8-15 years ago) , can't seem to find the post where you brought up the term "performance crown" before this one though
Click to expand...

burningrave101 said:
ATI got kicked around by other GPU manufacturers for over 15 years and then out of nowhere they had ONE good round which was the R300's and then now nVidia is back in the running again.
Click to expand...

That was the first thing i said and i was making reference to the performance of the cards.

Why the hell would i care about how many ATI cards were in OEM machines? That has absolutely nothing to do with them getting kicked around by the other GPU vendors.

HardOCP is an enthusiast site and you hardly ever see anyone recommend buying an HP, Dell, Gateway OEM machine because of that fact.
 
SnakEyez187 said:
How many people and companies were building their own computers 8-15 years ago?
Click to expand...

Once again, that has NOTHING to do with the performance crown. Your trying very hard to make an extremely moot point.

It wouldn't matter if there was 500 ATI cards in OEM machines and ONE 3DFX card in a custom built. If the 3DFX card was faster it was better then all those 500 ATI cards lol.
 
burningrave101 said:
Once again, that has NOTHING to do with the performance crown. Your trying very hard to make an extremely moot point.
Click to expand...

I'm not the one who brought up ATI's entire history into question rather than just the 3d portion, which would have made atleast a tiny bit of sense. You never once said "performance crown" in any of your posts, but jumped instantly on the *fact* that ATI was the bigger company and made more money back then, and those statements you made clear as day, wait forgot something, lol.

If you're going to suddenly change the subject to who had the performance crown, that's great, but that's not what you argued in answer to my first post, and that's certainly not what i'm arguing either

ATI came nowhere close to making more money and getting bigger then everybody else.
Click to expand...

ATI is a small canadian company that just just now managed to catch up with nVidia in market share the last couple of years because of their R300 cores. ATI couldn't touch nVidia before that.
Click to expand...

None of those bolded subjects are majorly relevent to having the performance crown
 
The Batman said:
Someone mind linking me to Carmack's response?
Click to expand...

"Our specular function isn't a POW function, it matches the bias / scale calculations done on the NV10/NV20/R200 paths."

Basically Carmack's intention was to make things look the same on all rendering paths, to make the engine as flexible and accurate as possible for programmers and artists across the entire spectrum of 3d cards. The new Humus/ATI optimization is at ends with this method - an approximation instead of mathematical equivalency - but it is faster for ATI owners.
 
SnakEyez187 said:
I'm not the one who brought up ATI's entire history into question rather than just the 3d portion, which would have made atleast a tiny bit of sense. You never once said "performance crown" in any of your posts, but jumped instantly on the *fact* that ATI was the bigger company and made more money back then, and those statements you made clear as day, wait forgot something, lol.

If you're going to suddenly change the subject to who had the performance crown, that's great, but that's not what you argued in answer to my first post
Click to expand...

SnakEyez187 said:
Wow, you really need to brush up on some ATI history, where is every other GPU company from 15 years ago, and please indulge me on how they got "kicked around" yet still made more money and got bigger then everybody else back then, now, if you had said maybe 5 years, then you would be making some sense, but don't even try to tell me ATI was crap it's entire lifespan
Click to expand...

I never changed the subject. You did. I was talking about ATI getting kicked around by other GPU manufacturers and anyone could easily see i was talking about the performance area because of how i made mention of ATI making their first real run with the R300 cores. Why would i start talking about who had more cards in OEM machines or who had more market share because of that back then?

And it wasn't just the 3D portion ither. They had the slower 2D cards as well.

Your whole OEM idea wasn't even relevant to the discussion as noone with any brains has ever bought anything according to market share or the size of the company.

PERFORMANCE is the ONLY thing that matters.

And ATI was not bigger nor did they make more money then nVidia before the R300 cores. They might of 15 years ago but that wasn't what i was talking about anyways. I was talking about about the more recent years up to the release of the R300 cores.

Where is all of ATI's money now if they were so big then? nVidia owns the marketing realm right now and has for several years.
 
The Batman said:
Someone mind linking me to Carmack's response?
Click to expand...

http://www.beyond3d.com/forum/viewtopic.php?t=14874&postdays=0&postorder=asc&start=220

Our specular function isn't a POW function, it matches the bias / scale calculations done on the NV10/NV20/R200 paths.
Click to expand...

Quite concise, much like Carmack's code. :)

Anyway, my point above again is that his intent wasn't to optimize, it was to maintain parity between the paths. Does that make optimizations like this one irrelevant or somehow "illegal" according to game "ethics". Not at all. I'm sure that if Carmack himself had an issue with this optimization, he would have said more on the subject.

That said, now that the "hack" has itself been further optimized to a single line, I'm getting even more performance out of it, and still no artifacting. I'm sure that if they wanted to, they could write a whole new R400 specific rendering path. Would it be worth the effort? Not likely. Would it be "cheating" somehow? No, not any more than the existing specific hardware rendering paths are "cheating". Just because Carmack had a reason to not do it himself doesn't make it less valid to do it after the fact.

The new Humus/ATI optimization is at ends with this method - an approximation instead of mathematical equivalency - but it is faster for ATI owners.
Click to expand...

It _was_ at odds. Carmack's response clued Humus into how to make the hack more accurate using a different technique. Now the hack is mathematically accurate, and gives an even larger performance boost.

Where is all of ATI's money now if they were so big then?
Click to expand...

They bought ArtX. They invested in the future.
 
Ok this started out as a discussion about the improvements on the ati perfomance side. It would be nice to get back on that discussion or just lock the thread and start it again in the ATi forum. I mean no disrespect but it gets old seeing the same names with the same old links and same arguements about how great the 6800's are. They are great cards. But some people actually bought and like *gasp* thier x800's! So really, if you don't have the hardware and don't really know what is going on, why post negative garbage. Who cares we are here to see if this is of any help performance wise. Why is it a big deal?
 
GeniusInABottle said:
A little OT, but humus is like the John Kerry of ATi.
Click to expand...

So basically you're saying Humus' optimizations are all talk & propaganda with no real substance or usefulness to them? :D
 
gsboriqua said:
So really, if you don't have the hardware and don't really know what is going on, why post negative garbage. Who cares we are here to see if this is of any help performance wise. Why is it a big deal?
Click to expand...

Because this performance hack was posted by someone who knows what they're doing. This person knows the hardware. And this person just so happens to work for ATI, so the conspiracy theorists assume that this isn't on the up and up. And they have nothing better to do than bitch about what they don't have full understanding of.
 
After ATI smokes Nvidia in Half Life 2 all you nvidiots, at least the smarter ones anyways, will go digging for these types of "hacks" as well. And then all of a sudden it will be perfectly fine thing to do.
 
tranCendenZ said:
So basically you're saying Humus' optimizations are all talk & propaganda with no real substance or usefulness to them? :D
Click to expand...

Yea i think that pretty much says it all for Humus.

He's come up with some really nice ideas to give the X800's an advantage and as the threads on Beyond3D progress its finally realized that most of these ideas will never be used or implemented lol.
 
killerD said:
After ATI smokes Nvidia in Half Life 2 all you nvidiots, at least the smarter ones anyways, will go digging for these types of "hacks" as well. And then all of a sudden it will be perfectly fine thing to do.
Click to expand...
Do you have something to tell us?
 
SuperRob said:
http://www.beyond3d.com/forum/viewtopic.php?t=14874&postdays=0&postorder=asc&start=220



Quite concise, much like Carmack's code. :)

Anyway, my point above again is that his intent wasn't to optimize, it was to maintain parity between the paths. Does that make optimizations like this one irrelevant or somehow "illegal" according to game "ethics". Not at all. I'm sure that if Carmack himself had an issue with this optimization, he would have said more on the subject.

That said, now that the "hack" has itself been further optimized to a single line, I'm getting even more performance out of it, and still no artifacting. I'm sure that if they wanted to, they could write a whole new R400 specific rendering path. Would it be worth the effort? Not likely. Would it be "cheating" somehow? No, not any more than the existing specific hardware rendering paths are "cheating". Just because Carmack had a reason to not do it himself doesn't make it less valid to do it after the fact.



It _was_ at odds. Carmack's response clued Humus into how to make the hack more accurate using a different technique. Now the hack is mathematically accurate, and gives an even larger performance boost.



They bought ArtX. They invested in the future.
Click to expand...

I'm not disagreeing with you. But I will say that when some random guy goes tooling around in videogame shader code, that's the very definition of a hack.

If you like hacks, more power to you, I don't think it's a bad thing...and I'm glad JC chimed in to lend a little more credibility to the situation...
 
KayossZero said:
I don't see any "Personal" attack so what are you trying to prove? Many Nvidiots, expecially Empty came in and thread crapped while legit people who we're trying to have a civil discussion got left behind in the dust, and thats pretty much that.
Click to expand...

When I don't know what I'm talking about I keep my mouth shut...but that's just me
 
KayossZero said:
I don't see any "Personal" attack so what are you trying to prove? Many Nvidiots, expecially Empty came in and thread crapped while legit people who we're trying to have a civil discussion got left behind in the dust, and thats pretty much that.
Click to expand...

I wanna know why everyone keeps calling people that like nVidia nVidiots in every other post.

If you dont know already, its considered flaming and you can banned for it.

You dont see me calling everyone that doesn't like nVidia an ATIdiot even if they ARE an idiot.

I think if some of you can't dispute the technical aspects of these cards without throwing around !!!!!! slang then you shouldn't post at all. I've done my best to refrain from doing it myself.

I've more then proven my point of view in my posts so if someone wants to call me an nVidiot they can because their just ignoring the facts presented.

And if you dont want nVidia users posting in certain threads then for gods sake post them in the ATI section instead of the main forum.
 
So let's get back on topic.

So far what's the average boost X800 users are seeing? And have all the artifacts been eliminated yet?
 
^eMpTy^ said:
When I don't know what the fuck I'm talking about I keep MY mouth shut...

I'm not disagreeing with you. But I will say that when some random guy goes tooling around in videogame shader code, that's the very definition of a hack.

If you like hacks, more power to you, I don't think it's a bad thing...and I'm glad JC chimed in to lend a little more credibility to the situation...
Click to expand...

No, that's not the definition of a hack.

Hacker's Dictionary said:
hack: 1. n. Originally, a quick job that produces what is needed, but not well. 2. n. An incredibly good, and perhaps very time-consuming, piece of work that produces exactly what is needed.
Click to expand...

THAT is the definition of a hack, albeit with the obvious omission as it pertains directly to software "... in as few lines of code as possible". This kind of performance increase by altering a single line of code is the definition of a hack.

Also, Humus is not "some random guy tooling around in videogame shader code". He works for ATI. He'd arguably know a great deal more about how to make the code work for his company's hardware than "some random guy". It took a great deal of knowledge to know how to make these optimizations, you couldn't stumble across this stuff accidentally.

So if you want to cast aspersions on what's happened with this, stick to the name calling, because you haven't got a shot on the facts.
 
killerD said:
After ATI smokes Nvidia in Half Life 2 all you nvidiots, at least the smarter ones anyways, will go digging for these types of "hacks" as well. And then all of a sudden it will be perfectly fine thing to do.
Click to expand...

I'd bet good money this guy's running on a 9200. :p
 
SuperRob said:
No, that's not the definition of a hack.



THAT is the definition of a hack, albeit with the obvious omission as it pertains directly to software "... in as few lines of code as possible". This kind of performance increase by altering a single line of code is the definition of a hack.

Also, Humus is not "some random guy tooling around in videogame shader code". He works for ATI. He'd arguably know a great deal more about how to make the code work for his company's hardware than "some random guy". It took a great deal of knowledge to know how to make these optimizations, you couldn't stumble across this stuff accidentally.

So if you want to cast aspersions on what's happened with this, stick to the name calling, because you haven't got a shot on the facts.
Click to expand...

He's a software engineer for ATi who is not on thier driver team and not working on Doom3...you may find this impressive...but I do not.

He did this in his spare time...the HACK is completely unofficial...it is supported neither by id nor by ATi...and it requires you to open and modify compiled files in your Doom3 installation...that my friend is a hack if I have ever seen one...you can church it up all you want to, but a hack is a hack...call it what it is.
 
I don't really care how this works, as it appears that it may very well. What i would like to know is if i will see anything like this for my 5900nu. As the revised hack does not seem to degrade image quality i would love to get a 5-10 frame jump off my hardware, i might be able to jump to 12x10.
 
^eMpTy^ said:
So let's get back on topic.

So far what's the average boost X800 users are seeing? And have all the artifacts been eliminated yet?
Click to expand...

Humus said:
This of course explains why it doesn't look exactly the same. This is also a good thing. Now we don't even need to use a POW function, but we'll be fine with a MAD_SAT, which make it even faster. Got 21% boost in the timedemo with this. Now I still don't know exactly what exponent he's approximating, or if he's using different lookup tables for different materials, but I've tried this and it seems to look the same, but our fellow artifact hunters may want to verify that.
Click to expand...

There ya go...
 
Since everyone seems to be only concerned about ATI's OEM business now I will attempt to bring the thread back on topic.

I have an AthlonXP2500+ Barton, 512 DDR400 Corsair XMSPro mem, and an ATI 9800 Pro running the 4.8 CATs. Before the REVISED hack I was getting 28.9FPS. After this hack I am now getting 34.8fps in timedemo1. I run at 1024x768 High Quality. And I also did the image_cachemegs thing and changed it to 96.

I see absolutely to degradation in image quality at all.

Not sure how the whole engine does its job and all but I didn't open one single file in any of my D3 directories. All I had to do was put a VFP file under /base/glprogs. And I believe compiled files can't just be opened and changed at will.
 
Status
Not open for further replies.
Back
Top