id Software's Official DOOM3 Benchmarks

Dyslexic said:
I hope not for thier sake.
Click to expand...

Makes you wonder what it was like running these benchmarks with guys from Nvidia and ATi in the same room. I wonder how they behaved.
 
3 things about the benchmarks:

1. last week the CEO of iD said that PCI-E helped nothing. The benchmarks do show a pretty nice improvement with PCI-E.

2. he also said there was virtually no performance difference between the 6800 and x800.

3. "The benchmarking was conducted on-site, and the hardware vendors did not have access to the demo before hand, so we are confident that there is no egregious cheating going on, but it should be noted that some of the ATI cards did show a performance drop when colored mip levels were enabled, implying some fudging of the texture filtering. This has been a chronic issue for years, and almost all vendors have been guilty of it at one time or another. I hate the idea of drivers analyzing texture data and changing parameters, but it doesn't visibly impact the quality of the game unless you know exactly what to look for on a specific texture. On the other hand, the Nvidia drivers have been tuned for Doom's primary light/surface interaction fragment program, and innocuous code changes can "fall off the fast path" and cause significant performance impacts, especially on NV30 class cards."
:rolleyes: at both ATI and NVIDIA. ATI should at least allow users to turn off trilinear hacks during gameplay. NVIDIA needs all the help they can get using ARB2 with a NV3x, but there's driver recognition isn't going to stop because performance is to critical. I'm kind of glad I don't have a 5900 anymore. ;)
 
^eMpTy^ said:
Makes you wonder what it was like running these benchmarks with guys from Nvidia and ATi in the same room. I wonder how they behaved.
Click to expand...

Man I would have loved to have been a fly on the wall during that one. :D
 
One thing we all need to take into consideration is this "quick" test was done using timedemos. We all know time demo's DO NOT represent actual game play. I can not wait to see real benchmarks using FRAPS with AI inaction with enemies running and firing weapons at you. This might be the missing peice to how ATI just got their ass handed too them.

On a side note: How do you think the ATI rep felt as his top of the line card was getting clowned by the middle level competition's card? OMG the Nvidia flamer must have been laughing his ass off the entire time.
 
pxc said:
3 things about the benchmarks:

1. last week the CEO of iD said that PCI-E helped nothing. The benchmarks do show a pretty nice improvement with PCI-E.

2. he also said there was virtually no performance difference between the 6800 and x800.

3. "The benchmarking was conducted on-site, and the hardware vendors did not have access to the demo before hand, so we are confident that there is no egregious cheating going on, but it should be noted that some of the ATI cards did show a performance drop when colored mip levels were enabled, implying some fudging of the texture filtering. This has been a chronic issue for years, and almost all vendors have been guilty of it at one time or another. I hate the idea of drivers analyzing texture data and changing parameters, but it doesn't visibly impact the quality of the game unless you know exactly what to look for on a specific texture. On the other hand, the Nvidia drivers have been tuned for Doom's primary light/surface interaction fragment program, and innocuous code changes can "fall off the fast path" and cause significant performance impacts, especially on NV30 class cards."
:rolleyes: at both ATI and NVIDIA. ATI should at least allow users to turn off trilinear hacks during gameplay. NVIDIA needs all the help they can get using ARB2 with a NV3x, but there's driver recognition isn't going to stop because performance is to critical. I'm kind of glad I don't have a 5900 anymore. ;)
Click to expand...


I forgot all about those statements he made. Maybe he made them before Nvidia optimized them for D3 as he did mention that he got the absolute latest drivers from both companies at the time if testing. Nvidia made improvements in framerate and who knows what ATI did if anything.
 
d34dly said:
Man I would have loved to have been a fly on the wall during that one. :D
Click to expand...
Except for their PR people, I don't think IRL they act like some of their moronic fans do on web boards. :p
 
pcx, the PCI Express system was a 3.6GHz 4GB DDR2 533MHz PC. The AGP systems were using 3.2GHz and only 2GB of RAM. The difference is CPU bound.
 
davidj said:
Many thnx to a results oriented [H] team. As usual great work guys!!!!!!!!!!!
Click to expand...
Thanks to all you guys that appreciate the work. :) We are at id's offices now working on the hardware guide. It will contain all that good stuff that you have come to expect from [H] evaluations.

(And BTW, the game friggin ROCKS.)
 
am I the only one that wished they would benchmark older video cards... 9800, 9700, geforcefxs and even geforce4s?

also, as someone else said... in systems withOUT 2gb and 4gbs of ram... maybe 512mb and 1gb?

not everyone spends $2000 on their pcs... the vast majority have a 2ghz PC with a 200 dll video card and 512mb-1024mb of ram

not a 3ghz pc, a 400 dll video card and 2gb of ram :/

while those benchmarks ARE appreciated (to see how most of us will be able to play doom3 in 2-3 years when we can afford computers like the ones used to benchmark), i'd like to see older systems being benched. thx.
 
^eMpTy^ said:
Was I just imagining things or did they say they were using the 8.05 driver from ATi?

What Catalyst version does this correspond to? 4.8? And has there been any comment on whether or not this driver version contains ATi's rumored new OpenGL driver?
Click to expand...

Looks like they are based on a version two releases away since 4.7's are based on 8.03.

/copied from ATI's website on the release notes for CAT 4.7

"This particular posting updates the ATI Display Driver to version 8.03."
 
OriginalReaper said:
pcx, the PCI Express system was a 3.6GHz 4GB DDR2 533MHz PC. The AGP systems were using 3.2GHz and only 2GB of RAM. The difference is CPU bound.
Click to expand...
That's disappointing if the game is still CPU bound at 3.2GHz. :(
 
Hooligan said:
am I the only one that wished they would benchmark older video cards... 9800, 9700, geforcefxs and even geforce4s?

also, as someone else said... in systems withOUT 2gb and 4gbs of ram... maybe 512mb and 1gb?

not everyone spends $2000 on their pcs... the vast majority have a 2ghz PC with a 200 dll video card and 512mb-1024mb of ram

not a 3ghz pc, a 400 dll video card and 2gb of ram :/

while those benchmarks ARE appreciated (to see how most of us will be able to play doom3 in 2-3 years when we can afford computers like the ones used to benchmark), i'd like to see older systems being benched. thx.
Click to expand...

We are not worthy. There is no review site out there that does benchmarks on realistic systems. As was mentioned it was Id that did the testing. I am waiting to see what real life benchmarks from a wider range of systems will show.
 
MrHappyGoLucky said:
One thing we all need to take into consideration is this "quick" test was done using timedemos. We all know time demo's DO NOT represent actual game play. I can not wait to see real benchmarks using FRAPS with AI inaction with enemies running and firing weapons at you. This might be the missing peice to how ATI just got their ass handed too them.

On a side note: How do you think the ATI rep felt as his top of the line card was getting clowned by the middle level competition's card? OMG the Nvidia flamer must have been laughing his ass off the entire time.
Click to expand...

I agree that FRAPS are the only way to go but in some cases there was almost a 20fps gap between cards and that is huge when you are only getting 70 to 80fps. Even in real game stuations with AI and gun fire where things can change alot, that is an awful big swing in the other direction ATI needs to make.
 
JKA Represent said:
im glad its cpu bound---it indicates how powerful the gpus of this generation really are
Click to expand...

DOOM3 uses less CPU than UT2004 from what I've heard.
 
The question is how much of this 15% discrepancy in performance is due to hardware failings, and how much is due to the fact that ATI's OpenGL drivers are, shall we say, lacking?
 
MrHappyGoLucky said:
One thing we all need to take into consideration is this "quick" test was done using timedemos. We all know time demo's DO NOT represent actual game play. I can not wait to see real benchmarks using FRAPS with AI inaction with enemies running and firing weapons at you. This might be the missing peice to how ATI just got their ass handed too them.

On a side note: How do you think the ATI rep felt as his top of the line card was getting clowned by the middle level competition's card? OMG the Nvidia flamer must have been laughing his ass off the entire time.
Click to expand...

Actually thats FALSE and i've confirmed it with other video card reviewers on other sites. What other games besides Far Cry does the AI not interact with the player in the demo? They do in the UT2004 demo's i tested last night.

Testing in-game with FRAPS causes the results to be influenced by the CPU more and the AI will react differently every time giving you mixed results that are unpredictable.

Go run some UT2004 time demo's if you dont believe me. They represent in-game gameplay the same as in-game testing with FRAPS and the results are STATIC instead of off the wall.

If you will look at some of HardOCP's results they are incredibly mixed.

Take a look at the results i was showing here:

http://www.hardforum.com/showthread.php?t=780578&page=2

Testing in-game with FRAPS is NOT more accurate then recorded Time Demo's. Far Cry is just a poor benchmark because it doesn't playback the AI interaction and gunfire in its demo's.

But seriously, if an X800 can't outperform a 6800 WITHOUT AI interaction then how can it possibly do so much better WITH AI interaction?? Especially if you realize the AI is controlled by the CPU.

There have been more then a handful of people say gameplay was smoother on the 6800GT when they switched from an X800Pro.

FRAPS can be detected and optimized for the same way as any other benchmark.

And if you will remember all the Far Cry SM 3.0 benchmarks, the performance results are different for each level and the results were also different on each site for the same level. That means the exact spot tested and the AI introduced influences the fps results and you can't be for sure which card is faster in the game.

I'm sorry, but even with in-game testing with FRAPS, ATI doesn't have a chance in Doom 3. Not when its fully OpenGL and supports UltraShadow II Technology.

Did you not see the City of Heroes benchmarks that Hard did using FRAPS in their latest BFG 6800GT OC review? City of Heroes is OpenGL.
 
This would be a good time for alot of you X800 owners to try to return your cards and get 6800 GT or Ultra's.
 
i just dont see the logic in buying a video card JUST for ONE game. After a while D3 will get old just like every other game. Yes, I pre-ordered too...but, it just seems like people place such a heavy emphasis on one game and how that determines their buying decision. I mean, for true hardcore gamers, what are you supposed to do? buy both cards so you can play HL2 and D3 ? and just switch cards when you want to play the other game?
It seems like people are more concerned about numbers, benchmarks, frames per second etc, than actually wondering if the game is going to be good at all or how much fun they're going to have playing it. (of course it'll be fun, right?)

I think that I will just play it an ENJOY it that way it is on the system I have without worrying about the guy next to me getting an additional 10 frames per second above me.

I can't wait for Aug 3rd....woo hoo!
 
Ummm didn't Carmack say himself that the AI and Physics don't come into play as much in Time Demos?
 
im still waiting to choose on what card to buy....i want to see what ATIs rewritten opengl drivers do.............
 
... ATI doesn't have a chance in Doom 3.
Click to expand...

Uh, I think they pretty much stated in the article that ANY CARD will be more than playable, and give a satisfactory experience. Saying that it "doesn't have a chance" is very misleading. These were all benchmarks on extreme configurations, and while the nVidia cards are certainly performing better, it was stated quite clearly that the ATI cards were all still VERY playable.

Yes, I'm disappointed that my XT didn't perform as well as I would have liked to see. Am I going to run out and replace it? Of course not, it works well enough for what it is. I might have harder feelings if I'd paid full price for it (or even half-price ... don't ask), but the card plays everything I have well, and everything I expect to have for a couple of years well, so I'm fine with it.
 
BlackDragon323 said:
i just dont see the logic in buying a video card JUST for ONE game.
Click to expand...
Think about how many games use the Quake3 engine. Now, think about the future games that will use the Doom3 engine. Ok. lol... :D
 
SuperRob said:
Uh, I think they pretty much stated in the article that ANY CARD will be more than playable, and give a satisfactory experience. Saying that it "doesn't have a chance" is very misleading. These were all benchmarks on extreme configurations, and while the nVidia cards are certainly performing better, it was stated quite clearly that the ATI cards were all still VERY playable.
Click to expand...

This isn't misleading at all. This isn't about providing a "good" experience, this is about providing the "best" experience...and ATi has no chance of providing the "best" experience in Doom 3.
 
xSyzygy666x said:
im still waiting to choose on what card to buy....i want to see what ATIs rewritten opengl drivers do.............
Click to expand...


haha, have fun waiting for that one. In response to the above poster, of course people buy a card for a game, they get excited and want to play a game very much at the highest quality they can. This has always sold games. The 'game' to buy cards for before this announcement was Far Cry (or for those more hopeful, Half LIfe).

It isn't like this card WON'T PLAY OTHER GAMES, lol. IN the case of Nvidia, it plays other games very well, usually within a statistical dead heat with the XT. So when one card really trounces another card and feeds it baby milk like a momma....people take note.

This should do alot to help the GT's already great sales.
 
oozish said:
haha, have fun waiting for that one. In response to the above poster, of course people buy a card for a game, they get excited and want to play a game very much at the highest quality they can. This has always sold games. The 'game' to buy cards for before this announcement was Far Cry (or for those more hopeful, Half LIfe).

It isn't like this card WON'T PLAY OTHER GAMES, lol. IN the case of Nvidia, it plays other games very well, usually within a statistical dead heat with the XT. So when one card really trounces another card and feeds it baby milk like a momma....people take note.

This should do alot to help the GT's already great sales.
Click to expand...

Not disputing your comments, but can you point me to your source for the release date of ATi's new OpenGL drivers, and the stats on how great the GT's sales are?
 
well i think u guys forgot that the 6800 series got 32x0 mode for the doom3 engine instead of 16x1 on 5900 series and the x800 series. i think the benchmark 6800 kick the x800 ass............ but i only play my game at 1028x768 cause of my crappy monitor. i own a x800 xt and i really hope they will optimise the opengl for it...... in reality i dont think it will happen......... lol if i am rich of course i will buy both...... just to show off
 
^eMpTy^ said:
Not disputing your comments, but can you point me to your source for the release date of ATi's new OpenGL drivers, and the stats on how great the GT's sales are?
Click to expand...

You don't have to have solid numbers... if you spend enough time on various forums you can tell. The great reviews and people searching high and low for a GT, it's pretty obvious they are doing well.
 
... this is about providing the "best" experience...and ATi has no chance of providing the "best" experience in Doom 3.
Click to expand...

Come on ... "best" is always a moving target. Everytime a new CPU comes out, "best" is going to change. Hell, haven't we all learned from the past that a simple driver update can change what is "best"? In fact, the card itself may be a great performer, but if the system is tuned poorly, that "best" might end up performing poorly.

And considering the fact that Doom 3 is reported to look the same on things even as low as a GeForce 4 MX, what exactly is "best" anyway? A couple extra FPS? Bah. I'm willing to call a spade a spade ... the 6800 series is outperforming the X800 series across the board. But please, don't start throwing subjective terms out there, because it's different for everyone. Stick with things that can be proven if you want to make a solid argument. Things like "performs better" and "costs less". Because if you want to say that something is the best, I'll just say that I like the way my X800 complements the colors of my case, and that makes it the best in ANY game. :p
 
PucSpifo said:
I take it you haven't played Splinter Cell: Pandora Tomorrow? Only 4 players in MP and it is one of the best MP experiences I could have imagined. I think DOOM 3 is really trying to play up the horror aspect of the game, and that is much easier to do with 4 people sneaking around as opposed to 16 people blowing each other up.
Click to expand...

Amen to that, can you imagine the terror, the surprise, the HOLY SHIT WHERE DID THAT COME FROM!!!???? :eek:


Heheh, I am planning a Doom3 LAN as soon as I can get my hands on the game....
 
^eMpTy^ said:
Not disputing your comments, but can you point me to your source for the release date of ATi's new OpenGL drivers, and the stats on how great the GT's sales are?
Click to expand...

no, i don't have sources...I'm commenting based on all the basic knowledge and hearsay I've read about in countless forums. ATI has been in dire need of an opengl overhaul since I can remember, so those of you waiting for them to really kick it up a notch and present nvidia with a challenge in the opengl forum...I just wouldn't get your hopes up.

In regard to GT sales...I've seen 3 shipments come in @ frys...currenty frys has none. GT's are sold out all over the place online. So I'm basing my comment about the great sales in GT Land on personal experience and I could be wrong.

But I doubt it.
 
d34dly said:
You don't have to have solid numbers... if you spend enough time on various forums you can tell. The great reviews and people searching high and low for a GT, it's pretty obvious they are doing well.
Click to expand...

I agree completely, I was just wondering if there were some numbers somewhere.
 
d34dly said:
Ummm didn't Carmack say himself that the AI and Physics don't come into play as much in Time Demos?
Click to expand...

I would hate to think how the X800's would perform in Doom 3 with MORE AI and Physics lol.

The AI is controlled by the CPU so the performance drop you get from added AI interaction is going to be primarily up to the CPU.

Time demo's are created by ACTUALLY playing the game and just recording the action. When you record it and play it back then its the exact same scene over and over again. The AI doesn't react differently and the results are more static. If you make two runs in-game with two different cards you dont have a chance in hell of getting the same results twice.
 
^eMpTy^ said:
This isn't misleading at all. This isn't about providing a "good" experience, this is about providing the "best" experience...and ATi has no chance of providing the "best" experience in Doom 3.
Click to expand...

That is a rediculous statement to say that ATI will never be able to beat the 6800 in D3. If I read the article correctly wasn't the point of it to let people know that a wide range of systems can play the game just fine? I never once saw Kyle write that the whole point is to tell people which card to buy. He did say how the cards ranked but only because he knew that the readers wanted to see that.
 
You must log in or register to reply here.
Back
Top