FiringSquad's new article

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

ViRGE

Elite Member, Moderator Emeritus
Oct 9, 1999
31,516
167
106
Originally posted by: gururu
I don't buy this preservation of 'reference' image stuff. JC admitted that they used a texture lookup table because it was faster than doing the math. He neglected to say that the lookup was faster only on Nvidia hardware. He admits that doing the math may actually be faster on certain hardware, but doesn't specify ATI or Humus or anything; the snub. It's a matter of what does the trick faster, the lookup or the math. Most sites agree as well that ATI's architecture DOES math a lot faster than Nvidia's, validating the gain in performance that the Humus tweak can offer (I got 20% increase with NO apparent artifacts). Lastly, no site has shown quality differences in either method.

In any regard, how could JC not know that the math would be faster on ATI. Of course he knew. For one reason or another, it seems he didn't just give a damn.
I think we're at a point where we're just going to have to agree to disagree, but I leave with this:
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.
 

ronnn

Diamond Member
May 22, 2003
3,918
0
71
yes was a major error by ID not to support all cards properly. If this can not be fixed for future games, the engine may have a short shelf life - as developers want to sell.
 

Childs

Lifer
Jul 9, 2000
11,313
7
81
Originally posted by: ViRGE
Originally posted by: gururu
I don't buy this preservation of 'reference' image stuff. JC admitted that they used a texture lookup table because it was faster than doing the math. He neglected to say that the lookup was faster only on Nvidia hardware. He admits that doing the math may actually be faster on certain hardware, but doesn't specify ATI or Humus or anything; the snub. It's a matter of what does the trick faster, the lookup or the math. Most sites agree as well that ATI's architecture DOES math a lot faster than Nvidia's, validating the gain in performance that the Humus tweak can offer (I got 20% increase with NO apparent artifacts). Lastly, no site has shown quality differences in either method.

In any regard, how could JC not know that the math would be faster on ATI. Of course he knew. For one reason or another, it seems he didn't just give a damn.
I think we're at a point where we're just going to have to agree to disagree, but I leave with this:
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.

Not to mention that Humus's tweak only really works on X800 and above. Only my 9800 Pro there was like a 2fps gain, so its hardly worth challenging someones credibility. Are you Humus's girlfriend or something? You seem to be taking this awefully personally.
 

jim1976

Platinum Member
Aug 7, 2003
2,704
6
81
Originally posted by: jrphoenix
Originally posted by: jim1976
Originally posted by: jrphoenix
Pretty humbling to see the 9800 run just as fast give or take as my GT :( (especially since my rig is very similar to theirs, except I have an extra gig of ram;)) I am going to get an Ultra in October to replace my GT... but still one can only imagine how much faster the x800's will be? Of course maybe my 16 pipes will pull away from the last gen card at 1600 x 1200 (that is the native resolution on my LCD and what I game at).

Pretty interesting read!

Where did you see GT being near XT? :confused:
At cpu lim 8x6? :D

Cmon you MUST be jokin of course...


No GT is not near the XT... it only keeps pace with the 9800 (on the graphs they show)... this is what made me sad because I have a GT :(

I still don't get it. Can you rephrase?
 

oldfart

Lifer
Dec 2, 1999
10,207
0
0
Originally posted by: Rollo

Everyone's tastes differ, especially at what makes a good game to them. I can't remember the last time I finished a FPS, and I will Doom3. I'm in the Recycling Center Level 2 now, and still loving it.

I didn't like HL at the time, the graphics were too primitive. To each their own.
Gameplay 1st, pretty graphics 2nd (not even).

I've played plenty of games that look nice but were not enjoyable. FWIW, the graphics that HL had in 1998 were as good as anything else out at the time except for maybe Unreal.

I'll take an enjoyable game with average graphics over a pretty one that is boring any day. (BTW, not saying D3 fits into either of those).
 

jim1976

Platinum Member
Aug 7, 2003
2,704
6
81
Originally posted by: oldfart

I've played plenty of games that look nice but were not enjoyable. FWIW, the graphics that HL had in 1998 were as good as anything else out at the time except for maybe Unreal.

My thoughts exactly... Unreal was the GFX revolution of the time. SFX/sound and atmosphere were pretty cool too. But when HL came out.... You know... :D
G-A-M-E-P-L-A-Y & I-N-T-E-R-A-C-T-I-O-N . 2 uknown words for any fps till that time... HL was probably the reason that I started playing intensively FPS.

I'll take an enjoyable game with average graphics over a pretty one that is boring any day. (BTW, not saying D3 fits into either of those).

Anytime man anytime... ;)
 

Childs

Lifer
Jul 9, 2000
11,313
7
81
Originally posted by: jim1976
Originally posted by: oldfart

I've played plenty of games that look nice but were not enjoyable. FWIW, the graphics that HL had in 1998 were as good as anything else out at the time except for maybe Unreal.

My thoughts exactly... Unreal was the GFX revolution of the time. SFX/sound and atmosphere were pretty cool too. But when HL came out.... You know... :D
G-A-M-E-P-L-A-Y & I-N-T-E-R-A-C-T-I-O-N . 2 uknown words for any fps till that time... HL was probably the reason that I started playing intensively FPS.

I'll take an enjoyable game with average graphics over a pretty one that is boring any day. (BTW, not saying D3 fits into either of those).

Anytime man anytime... ;)


IMO HL was overrated. I didnt even feel like finishing it the first time I played. Aside from lots of scripted events, there wasnt anything all that new. The mods are what made HL great and gave it longevity. It was a quality game, but overrated. If anything, there was less interaction because everything was so scripted. One thing that was great was the level load times. I wish more companies adopted that approach.
 

oldfart

Lifer
Dec 2, 1999
10,207
0
0
There was a lot that was new such as interaction with your environment. The ever popular wooden crates could be moved, broken, stacked, floated in water. Glass broke. I hate shooting a bottle and it just sits there. You could interact with NPC's. The sound was awesome. The game had a mission feel to it. You had a reason to get to the next level other than just getting to the next level. 3 way battles between you, the Marines, and the aliens. You could drive vehicles (limited). You got weapons and health in a more realistic way instead of them just "floating in space". There was variation. You were underground in tunnels, in an office, outdoors. It broke it up.

I did not like the whole Xen area though. That asteroid jumping thing was ....meh.

Compared to the other games I was playing at the time, Q2 and Unreal, it was WAY ahead. At the time it was.

HL = best SP game
Unreal = best graphics
Q2 = best online game (The Edge, Tokays Towers, The Frag Pipe....fond memories).
 

apoppin

Lifer
Mar 9, 2000
34,890
1
0
alienbabeltech.com
Originally posted by: oldfart
Gameplay 1st, pretty graphics 2nd (not even).

I've played plenty of games that look nice but were not enjoyable. FWIW, the graphics that HL had in 1998 were as good as anything else out at the time except for maybe Unreal.

I'll take an enjoyable game with average graphics over a pretty one that is boring any day. (BTW, not saying D3 fits into either of those).
I am . . . gameplay in Doom III is Banal . . . at best. A one trick-pony - "BOO!", over and over and over and over . . . , no damn AI, boring PDA messages that DETRACT from the "atmosphere" (which is as thin as Martian air), once you find the artifact you can fall asleep (i don't think i lost a single life after that "find"; since you get renewed when you use it [what kind of crap is THAT?] and then it's finally over - you have beaten the End Boss in one try and you survey the damage saying "that sucks". i only feel ok in that it cost me $25 (ultimately) for this crappy game . . .

YES, the game's ENGINE is NICE!

:roll:

gotta hope HL2 has better game play and a semblence of a story . . .

 

Childs

Lifer
Jul 9, 2000
11,313
7
81
Originally posted by: oldfart
There was a lot that was new such as interaction with your environment. The ever popular wooden crates could be moved, broken, stacked, floated in water. Glass broke. I hate shooting a bottle and it just sits there. You could interact with NPC's. The sound was awesome. The game had a mission feel to it. You had a reason to get to the next level other than just getting to the next level. 3 way battles between you, the Marines, and the aliens. You could drive vehicles (limited). You got weapons and health in a more realistic way instead of them just "floating in space". There was variation. You were underground in tunnels, in an office, outdoors. It broke it up.

I did not like the whole Xen area though. That asteroid jumping thing was ....meh.

Compared to the other games I was playing at the time, Q2 and Unreal, it was WAY ahead. At the time it was.

HL = best SP game
Unreal = best graphics
Q2 = best online game (The Edge, Tokays Towers, The Frag Pipe....fond memories).

I wouldnt dispute anything you stated, in fact I guess I agree with just about everything in the quote. I did forget how cool the marines were. And Xen was blah. Guess its been awhile since I really thought about it.
 

Drayvn

Golden Member
Jun 23, 2004
1,008
0
0
Originally posted by: gururu
Originally posted by: ViRGE
Originally posted by: Pete
I thought IQ was superior due to using math, rather than filtered texture lookups. As for not "compromising" D3's IQ, that sounds like more of an artist than a programmer issue.

I don't remember JC specifically denigrating Humus' patch b/c of lower or compromised IQ. His only comment on it, AFAIK, is in this B3D "interview,"">http://beyond3d.com/interviews/carmack04/index.php?p=2</a> and it doesn't mention costing or compromising IQ.

Note where he talks about all the artwork being done on non-fragment hardware; because Humus's tweak doesn't match the square/bias, it produces an image that deviates from the "reference" image, hence the comments I made about IQ. In a generic case, the IQ may be superior, but for Doom 3, this perfectly matches what id wants, and as Carmack notes, it's also useful for when you want a finite cutoff angle, which I'm assuming they're taking advantage of here.

I don't buy this preservation of 'reference' image stuff. JC admitted that they used a texture lookup table because it was faster than doing the math. He neglected to say that the lookup was faster only on Nvidia hardware. He admits that doing the math may actually be faster on certain hardware, but doesn't specify ATI or Humus or anything; the snub. It's a matter of what does the trick faster, the lookup or the math. Most sites agree as well that ATI's architecture DOES math a lot faster than Nvidia's, validating the gain in performance that the Humus tweak can offer (I got 20% increase with NO apparent artifacts). Lastly, no site has shown quality differences in either method.

In any regard, how could JC not know that the math would be faster on ATI. Of course he knew. For one reason or another, it seems he didn't just give a damn.


Thats what i thought, what Humus did doesnt take long, iD could have easily have implemented his work into the retail game but didnt, it just required a few numbers changed and a few replaced words and that was it. Since ATi and nVidia were working with iD, they must have known that ATi was reknowned for having a much better maths architecture than nVidia, but they didnt seem to want to try to implement this into the games engine, aagain all it needed was an extra set of instructions to tell it to change a few things in the engine and that was it. All i suppose the gamer had to do was set AF in the control panel. Valve knew that the FX card wouldnt run well on the DX9 stuff, and we all know that the FX are not as good as the 9700+, so they put in a new code to make it run well without having to sacrifice too much Quality, they actually went out of their way to put in new and SIMPLE coding to make each and every card run at playable frame rates with the best quality it can muster.

Now tell me, iD dont bother acknowledging that having a math calculation instead of a look up table is better for ATi and just go for the look up table which seems to run worse of ATi but not nVidia

Or Valve, at first it was ATi are much better, but later nearer release time, we have including extra coding so that the FX owners can play at reasonable frame rates without a lot of loss in Quality, acknowledging the poor perfomance in DX9 as HL2 does use it extensively

So i ask, tell me whose comprimising the GPU companies more? iD or Valve?

Everyone says that Valve are more in bed with ATi than iD were with nVidia, and that Carmack is a god and wouldnt exclude half his audience.... But what about helping ATi by putting a few bits of extra and SIMPLE coding to make it run faster, by up to 20% by what some review sites say.

And as for the IQ loss, but the end of the thread where Humus pout his tweak, they worked out a flawless tweak which gave no apparent Quality loss, but increased the peformance a lot.
 

Pete

Diamond Member
Oct 10, 1999
4,953
0
0
Originally posted by: Rollo
Wow. Nobody got what I was trying to say here at all.

I was trying to say HL2 accomodates a variety of hardware better than Doom3. No more, no less.

...

I suppose it would have been clearer if I said "5900s are better off with HL2 than any last gen card is with Doom3".
Agreed, on both counts. :)

It's a pity JC/iD chose to limit new cards to the old card's abilities simply to maintain IQ across all boards. I much prefer the ability to choose b/w IQ and speed, and that really one of the huge advantages of PCs over consoles in general. Point to HL2.

This is where I have to disagree with Virge, as there's no point in refusing to allow greater IQ from a newer card simply to maintain IQ. But iD is still a small company, IIRC, so I suppose they can't afford to test everything. Still, surely they could've found a few iD fanboys willing to run through the game on a few PCs lined up next to each other with different cards/paths to note any differences in IQ? As for a finite cut-off, that sounds like a one-line add for a math function.

Anyway, back to anticipating a good game. :)
 

nitromullet

Diamond Member
Jan 7, 2004
9,031
36
91
No GT is not near the XT... it only keeps pace with the 9800 (on the graphs they show)... this is what made me sad because I have a GT

The GT is roughly 20% faster than the 9800XT on the DX9 HL2 VST, but "close" on the CS Source tests. This is most likley because both cards run the CS test very well, however, the CS test is capped at 90 FPS. Apparently, these cards both can perform well above the 90 FPS cap, which is why you see them both average out in the 80's.
 

lordtyranus

Banned
Aug 23, 2004
1,324
0
0
It's a pity JC/iD chose to limit new cards to the old card's abilities simply to maintain IQ across all boards. I much prefer the ability to choose b/w IQ and speed, and that really one of the huge advantages of PCs over consoles in general. Point to HL2.

This is where I have to disagree with Virge, as there's no point in refusing to allow greater IQ from a newer card simply to maintain IQ. But iD is still a small company, IIRC, so I suppose they can't afford to test everything. Still, surely they could've found a few iD fanboys willing to run through the game on a few PCs lined up next to each other with different cards/paths to note any differences in IQ? As for a finite cut-off, that sounds like a one-line add for a math function.
Well, that's what the humus mod is for. Oddly enough, I've heard some pro Nvidia people (not Rollo), laugh at how the mod lowers IQ and is unacceptable.

Personally I fail to see how ATI's trylinear (even when it is visible) is as bad as lacking directx 9 effects (which apparently now is not a big deal).
 

nitromullet

Diamond Member
Jan 7, 2004
9,031
36
91
Personally I fail to see how ATI's trylinear (even when it is visible) is as bad as lacking directx 9 effects (which apparently now is not a big deal).

I think the difference is not that no one will agree that DX9 is preferable, but moreso that the DX8 codepath for the FX series is out in the open. Neither Valve nor nVidia have hidden the fact from anyone. I think this goes to show that most consumers value honesty over inflated FPS. This makes sense in the respect that with this out in the open, consumers can judge for themselves if saving a buck on an FX card is worth it or not.
 

AnnoyedGrunt

Senior member
Jan 31, 2004
596
25
81
The Xbit labs article (I think it was Xbit labs) that talked about ATI's buggy OpenGL driver touched lightly on the Humus mod. It sounded like the mod really helped when AF was being used because of the way Doom 3 handles the control panel AF vs. ingame AF. I don't think the mod actually does that much other than cover up an existing special case scenario. I certainly don't think iD had any intention of slowing down the ATI cards compared to Nvidia cards.

As far as DX9 goes, I definitely don't see much of a difference between it and DX8.1 in those screenshots. Occasionally there is a very minor differnce, but I don't think it would be that noticeable when actually playing the game. Therefore I don't feel that having DX9 is that important for HL2. I also don't think the "trylinear" optimization affects IQ too much, but I definitely think ATI needs something in their driver to allow the user to choose whether it is on or off.

Finally, HL was awesome on Earth, and sucked afterwards. So much so, that I never finished the game. Really, my favorite part was the marines and walking in on fights already in progress and picking off whoever seemed to be winning so they could kill eachother off. Hopefully HL2 has the good from the previous title and leaves out the lame stuff.

-D'oh!
 

Drayvn

Golden Member
Jun 23, 2004
1,008
0
0
Originally posted by: AnnoyedGrunt
The Xbit labs article (I think it was Xbit labs) that talked about ATI's buggy OpenGL driver touched lightly on the Humus mod. It sounded like the mod really helped when AF was being used because of the way Doom 3 handles the control panel AF vs. ingame AF. I don't think the mod actually does that much other than cover up an existing special case scenario. I certainly don't think iD had any intention of slowing down the ATI cards compared to Nvidia cards.

As far as DX9 goes, I definitely don't see much of a difference between it and DX8.1 in those screenshots. Occasionally there is a very minor differnce, but I don't think it would be that noticeable when actually playing the game. Therefore I don't feel that having DX9 is that important for HL2. I also don't think the "trylinear" optimization affects IQ too much, but I definitely think ATI needs something in their driver to allow the user to choose whether it is on or off.

Finally, HL was awesome on Earth, and sucked afterwards. So much so, that I never finished the game. Really, my favorite part was the marines and walking in on fights already in progress and picking off whoever seemed to be winning so they could kill eachother off. Hopefully HL2 has the good from the previous title and leaves out the lame stuff.

-D'oh!

The changes that are made through the Humus tweak is to change how the look up table is done, on the ATi machines its changed so that its a math calculation instead of just a look up, and the resulting fact of the look up, is that AF is only used on the textures it picks out, but when the Humus tweaks are put the resulting fact is that, putting on AF globally will stop the decrease in performance, i think ive got a good explanation here at Elite Bastards....

"As we can see, Humus' tweak has a much greater effect when using ATIs own control panel anisotropic filtering than when in-game filtering is used. The explanation for this is most likely quite simple - In the normal part of the game code altered by Humus, a texture look-up table was used to generate values. Using in-game AF, this is obviously not filtered as it is both unnecessary and computationally expensive. However, when AF is forced by the driver, it applies its filtering to everything, including the aforementioned look-up table, thus contributing to the large performance hits we've seen using this method.

Humus' tweak removes the look-up table code to replace it with code to calculate maths values instead - This offers a small performance improvement when in-game AF is used thanks to the X800 Pros speed at handling such operations, but when AF is forced at driver-level, the performance increase is vastly greater as the look-up table is no longer available to be processed by the anisotropic filtering mechanism, thus saving the amount of time normally required to be spent on filtering it.

With regard to image quality, there were no noticeable differences to the naked eye when this tweak was used compared to the normal Doom 3 code. It is no doubt not identical to the original image, but is perfectly acceptable and usable for game-play. "

And heres the link... http://www.elitebastards.com/p...;head=1&amp;comments=1

Also, and as for the DX9 and DX8.1 thing, if u read the first page of Firing Squads review, ull see what each one has over the other, here... http://www.firingsquad.com/har...geforce_fx_half-life2/

 

Marsumane

Golden Member
Mar 9, 2004
1,171
0
0
First reaction: WOW.... I wonder how the cards only seem to be effected THAT much in only the source engine... (more comments later)