id Software's Official DOOM3 Benchmarks

Kinda funny how this thread died out so quickly and all the meaningless bickering came to an end :) and everything move to the other thread and so far after a few hundred posts there still hasnt been any flames :)

Now thats the kinda threads that are nice and relaxing to read :)
 
I just cant wait until ati is forced to turn off the bilinear cheat when trilinear settings are turned on. Expect to see major fps hit, ati cards will also have to use color minimaps like nvidia is using in upcoming benchmark tests. Last but not least, the question that puts everything to rest is why get X800 and hope?... :eek:
 
evilchris said:
Face it. NVIDIA pwned ATI. Period. If ATI fans want to downplay this, THEN DON'T EVER MENTION FARCRY AGAIN. Otherwise, ADMIT IT. ATI's OPENGL DRIVER AND HARDWARE SUPPORT SUCKS.

6800 GT = 5.6 Gigapixels/sec fillrate

X800 XT PE = 8.3 Gigapixels/sec fillrate

GT beats XT PE in Doom3. ATI has COMPLETELY dropped the ball with OpenGL. ADMIT IT.

Oh, I know, it's just one game right? No other games will use the D3 engine that matter, only little insignificant ones LIKE QUAKE 4.
Click to expand...

And Quake 4 will be the only Doom 3 based engine to be released next year and with the updated AI routines, increased poly counts and graphical touches that they always add to id's engines, even the GeForce 6800 Ultra won't be enough and will want to be on the next refresh or next new architecture for the cards. You can expect a 15fps drop if not more across the board on Quake 4. Also, keep in mind that development on Quake 4 has been in the works since the Doom 3 engine has been code complete and that's been over three years now, which means, sans titles already under development, but not released yet, we won't be seeing them until 3rd or 4th quarter of 2006 at the very earliest and by that time, these cards won't be to special anyway, You don't know what you are talking about. ADMIT IT. There will also be a lot more games utilizing the Source engine than Doom 3, because the development time on the Doom 3 platform takes a significantly longer amount of time and the tools and mod kits aren't as plentiful compared to Source. That goes without saying that Doom 3 is the graphically superior engine though.
 
HeavenX said:
I just cant wait until ati is forced to turn off the bilinear cheat when trilinear settings are turned on. Expect to see major fps hit, ati cards will also have to use color minimaps like nvidia is using in upcoming benchmark tests. Last but not least, the question that puts everything to rest is why get X800 and hope?... :eek:
Click to expand...

I'm not so sure it's necessary given that ATI's brilinear filtering on the X800xxxx platform looks better than Nvidia's 6800xxxx platform with trilinear filtering on most scenes.
 
i know there is about a hundred games that will be using the havok 2 engine. (including tribes: vengeance). so is the half life 2 engine been totally re-done (now called 'source') or is it a heavily modified havok 2 engine?
 
Havok 2 is a physics engine. The Source engine (half life 2 its always been called the Source engine) uses havok 1. havok 2 is faster in all respects and tuned for consoles as well.
 
rancor said:
Havok 2 is a physics engines. The Source engine (half life 2) uses havok 1. havok 2 is faster in all respects and tuned for consoles as well.
Click to expand...

hmm i though havok 2 was more than just a physics engine but i guess it makes sense (physics for DE:IW and many many more).
 
well I don't even see why people buy it, it costs like 75k for something that 2 or 3 programmers can do in 6 months :/. Physics isn't that hard to code compaired to the actual renderer. (Got to know physics anyways to make the renderer lol)
 
Well sometimes time is money and who knows, they could be getting a great deal on it.
 
True but the physics engine that Unreal 3 uses is only 15k for a full site license, and its better then Havok by far.
 
rancor said:
True but the physics engine that Unreal 3 uses is only 15k for a full site license, and its better then Havok by far.
Click to expand...

the unreal 3 engine has just been shown. while the havok 2 physics engine has been around/announced for a year or two now
 
Thier physics engine has been around for a long time ;). Not as long as Havok but its been out there.
 
I, personally, could care less about Havok1/2. The Unreal Engine just looks so awesome.
 
TheCrimsonCrusader said:
You don't know what you are talking about. ADMIT IT. There will also be a lot more games utilizing the Source engine than Doom 3, because the development time on the Doom 3 platform takes a significantly longer amount of time and the tools and mod kits aren't as plentiful compared to Source. That goes without saying that Doom 3 is the graphically superior engine though.
Click to expand...

The development took longer because they were developing it from scratch. Now that it's done, it'll be much easier to implement in other games. id is also releasing their developer kits and tools they used to make the game to other people who are using the engine, etc.

Half-Life 2 engine was "easier" to code because they used some parts from the Havok engine, etc.
 
There will also be a lot more games utilizing the Source engine than Doom 3, because the development time on the Doom 3 platform takes a significantly longer amount of time and the tools and mod kits aren't as plentiful compared to Source. That goes without saying that Doom 3 is the graphically superior engine though.
Click to expand...


Highly unlikely. The source engine and doom 3 engine have very similiar capabilites, and game devlopement if anything like q3 or hl, they are equaly easy to develop with. When it comes to a developer that is buying an engine, they don't care about mods period. There is a big difference in making a full game and a mod. Both companies give full tech support to licnesees and a copy of the game source code too, with appropriate documentation for all tools and how that game was made.
 
Hehe, yeah I definitely think that the Doom 3 physics engine is at least on par with the modified Havok one used in HL 2. I mean come on, Carmack wrote it. Also i wouldn't base any any assumptions on the vids we've seen so far, I mean everyone saw the trailer where the invisible demon was tossing guys around, but that was really early on, and I'm hoping wasn't the final version of the physics engine, the zombie seemd to go static a bit too soon. Whereas, on the other hand, in the icon video the physics when Pinkie got shot were pretty impressive in my opinion.
Now while we might not see all the different effects of the Doom 3 engine, ie liquid effects (everyone keeps refering to how in half life 2, matresses float...... well yippee) given the game world and environment, the capability is probably there.
 
Have you ever wrote a physics engine? Doubt it. To even get a half decent physic engine you have to put a hell of a lot of time and effort into it. First you have to get a decent maths engine written. I don't just mean implement the add/sub mul/divs as operators and use that since that would be slow. No - this is really grind it down using intermediate structures for storage of temporaries and then grind it out to SSE/3DNow/Altivec/Whatever. Yes JC probably could do a true physics engine and it would be quite good, but HAVOC have had years of experience, they have PhD's writing the stuff (an advert for a physics engine programmer in Newcastle UK, was advertised at £50,000 a year). Besides previous id games had enough physics so it 'seemed' right which is very different to if it is right. HAVOC is about it being near-as right as within the polling constraints.

In short. The Doom3 physics engine will be nowhere near the HAVOC engine. It will implement enough maths so the rag-doll effect looks good, not so it is right. If JC would care to disagree then I would thrash out an argument with him. I started a physics engine (for car sims) and it is a real bitch to stop everything falling apart as it approaches zero state fluctuations.

Remember - everything is an approximation to a real.
 
rancor said:
well I don't even see why people buy it, it costs like 75k for something that 2 or 3 programmers can do in 6 months :/. Physics isn't that hard to code compaired to the actual renderer. (Got to know physics anyways to make the renderer lol)
Click to expand...

Wrong. HAVOK is written by PhD's. These aren't your average games programmers. I am an average programmer I can't get a physics engine right. Also those 6 months while those 3 programmers are writing a physics engine is 6 months they haven't devoted to gameplay. Id are fortunate in the fact that they don't get feature creep, they don' t have milestones, and are under zero pressure from the publisher.

Also, you don't need to know Physics to make a renderer. I wrote a renderer in software years before I wrote any physics.
 
Mournblade said:
Have you ever wrote a physics engine? Doubt it. To even get a half decent physic engine you have to put a hell of a lot of time and effort into it. First you have to get a decent maths engine written. I don't just mean implement the add/sub mul/divs as operators and use that since that would be slow. No - this is really grind it down using intermediate structures for storage of temporaries and then grind it out to SSE/3DNow/Altivec/Whatever. Yes JC probably could do a true physics engine and it would be quite good, but HAVOC have had years of experience, they have PhD's writing the stuff (an advert for a physics engine programmer in Newcastle UK, was advertised at £50,000 a year). Besides previous id games had enough physics so it 'seemed' right which is very different to if it is right. HAVOC is about it being near-as right as within the polling constraints.

In short. The Doom3 physics engine will be nowhere near the HAVOC engine. It will implement enough maths so the rag-doll effect looks good, not so it is right. If JC would care to disagree then I would thrash out an argument with him. I started a physics engine (for car sims) and it is a real bitch to stop everything falling apart as it approaches zero state fluctuations.

Remember - everything is an approximation to a real.
Click to expand...

Physics is the basis of a 3d engine. Just that lately CPU's are powerful enough to do more physics so they are being added. There are free physics engines out there that are as good as Havok. A Ph D means nothing, all ya really need is high school phyiscs to do what havok is doing now (and be a really good programmer). I'm sure you have heard of ODE, ya might want to take a look at it cause its got rag doll and vehicle physics just like Havok. It doesn't have fluids or air, but really those aren't that hard since they are an entity of thier own.
 
One last thing before I hit the sack: ATI haven't done anything to OpenGL in over a year.

Wow. Neither has MS since the introduction of it with NT4.0.

That's why we as programmers have shed load of trouble because all the new ARB extensions aren't implemented. Fortunately you only have to write the code to grab them once (or use one of the many professional standard libraries to get them for you) and it is (almost) plain sailing.
 
Hell I know people with PhD's and they are the dumbest people you'll ever meet. :rolleyes:
It's just a title and nowadays it means little to people that understand that. Thread crap anyone?
 
rancor said:
Physics is the basis of a 3d engine. Just that lately CPU's are powerful enough to do more physics so they are being added. There are free physics engines out there that are as good as Havok. A Ph D means nothing, all ya really need is high school phyiscs to do what havok is doing now (and be a really good programmer). I'm sure you have heard of ODE, ya might want to take a look at it cause its got rag doll and vehicle physics just like Havok. It doesn't have fluids or air, but really those aren't that hard since they are an entity of thier own.
Click to expand...

Lets hack this argument to pieces: Physics is the basis of a 3D engine Not it isn't: The graphics subsystem is completely different to the physics. The physics operates, and you are left with a position and a rotation in the 3d world.

Also it depends on what the physics engine does. Are we talking a fully constrained system or not? How high will we be polling? Where do you draw the line: Do you measure the flex in a carbon fibre wishbone suspension or just add a little fudge to the packers or put the emphasis in the tyre?

There are free physics engines out there that are better than HAVOK. Doesn't mean they are particularly fast. High school physics doesn't give you the necessary information to do a decent physics engine. You have to go out and do research yourself. If you understand the way it works, you are better than high-school.

"It doesn't have fluids or air, but really those aren't that hard since they are an entity of thier own". Those aren't hard!?. May I recommend that you actually research non-newtonian fluid dynamics before speaking? We have a doctor at my work who wrote a 160 page tome on non-newtonian fluid dynamics in a Rolls-Royce jet engine for his PhD. I am quite good a maths, but his was way beyond my comprehension.
 
Mournblade said:
Lets hack this argument to pieces: Physics is the basis of a 3D engine Not it isn't: The graphics subsystem is completely different to the physics. The physics operates, and you are left with a position and a rotation in the 3d world.

Also it depends on what the physics engine does. Are we talking a fully constrained system or not? How high will we be polling? Where do you draw the line: Do you measure the flex in a carbon fibre wishbone suspension or just add a little fudge to the packers or put the emphasis in the tyre?

There are free physics engines out there that are better than HAVOK. Doesn't mean they are particularly fast. High school physics doesn't give you the necessary information to do a decent physics engine. You have to go out and do research yourself. If you understand the way it works, you are better than high-school.

"It doesn't have fluids or air, but really those aren't that hard since they are an entity of thier own". Those aren't hard!?. May I recommend that you actually research non-newtonian fluid dynamics before speaking? We have a doctor at my work who wrote a 160 page tome on non-newtonian fluid dynamics in a Rolls-Royce jet engine for his PhD. I am quite good a maths, but his was way beyond my comprehension.
Click to expand...

Do you really want to discuss physics engines with me :). You don't know who I am, but lets just say I have worked on 5 different 3d engines with fully complient physics engines and that are just as good as havok if not better, and worked on engines that have used havok 1. ODE is hmm a smidgen slower then Havok 1, its about 15% slower then Havok 2.

Really you think you can use non-newtonian fluid dynamics in a 3d engine? What crock is that, this is what I mean when a Ph D doesn't matter. You try that and you will kill performance. I don't care how you do it its not possible on todays cpus in real time games period. BTW I too have a Ph D, in comp sci. and very good with math and physics as I have minors in both.

If you want to butt heads with me on real time renderering be my guest but things like high level physics are not used in 3d engines, the main reason is they are too complex for cpus to cacluate in real time. 3d engines for games get their speed because of inaccuracies or fooling the eye to show something that is not truely there, like bump mapping, or using the frenel term on water and animating its bump map.

Answer this for me in Half life 2 or even Unreal tournament 2003 or 04 how are they cacluating thier fluids? What is an optimal equation for real time rendering or water via vertex and pixel shaders?
 
Forgot to add in the last statement high school physics does give you what you need to write havok. I have never used anything more the Kenemtics and linear algebra in a physics engine yet. Also I haven't even seen anything higher then these in any physics engine for games. In a simulation program yes you are right they do require more accuracy and require algos based on higher level math and physics. Not in games though.
 
I wrote a train simulator - and not for PC but a proper 'real world' fully operational train cabin with proper lights, buttons, bells and horns. Not as exciting as cars/fps but it pays the bills.

At highschool in the UK I never did Kinematics - maybe it has changed in the 10 years since I was at school? Linear Algebra was a 6th form level subject (17/18 years of age).

I didn't say you can use non-newtonian physics in a 3d engine. I was responding to your statement that they were easy.

'fully complient' Compliant with what? There aren't any tests to pass (as far as I know). The best test is to fool the viewer to believe what they are seeing is real.

"3d engines for games get their speed because of inaccuracies or fooling the eye to show something that is not truely there, like bump mapping, or using the frenel term on water and animating its bump map."
Click to expand...
This is exactly what I was saying to Suicide King - it isn't real. It looks and responds like it is, but in all likelyhood there are fudges that really just avoid the costly maths.

Back to the point that JC's physics isn't likely to be as good as Havok's. Havok have a 3D engine that can simulate several different scenario's to a lesser or greater degree. John's probably (the only one who knows is him) simulates only what it needs to and only that, and does it very well.

I probably went off on one because I was tired. No offence? :)
 
I hear ya :), havok does have more options then most physics engines but when it comes to purely games, they aren't useful, over all there really isn't anything special in havok over other compairable physics engines out there with gaming in mind. (by complient I mean SDK type physics enignes).
 
Dunno why people see the havoc physics engine as being "all that". To me especially in its ragdoll effects it just makes the characters look as if theyre made out of rubber with no spine the way they flop around. Haliflife 2...well im beginning to think physics is the only gimmick that game has since its graphics can;t touch those of doom 3, that and every video we have seen on hl2 all they can do is show off stuff with physics....I think we get the point by now the game uses physics, don't need to show little setup sequences in just about every vid that just happen to show off physics to some degree.
 
the physics in hl2 looks awesome compared to ANY other game I've seen... so why not show the reality the physics engine has? I applaud it.

~Adam
 
Half life 2 was the first game to show Havok off, its tested and proven, even the Karma (math engine) engine (unreal tournment 2003 and 04) is equal to Havok (unfortunately they were bought out by Renderware), just that Havok has been around for alot longer. Havok is even used in 3ds Max, so authoring game models ready for use with in Havok is easier to set up if a studio is using Max. As cpus get more powerful I'm sure more of the advanced physics will be used. Just a matter of time.
 
Just got Doom 3 :), physics are just as good as half life 2, didn't see any water yet though lol, but hell its mars!
 
if you watch the doom3 trailer you look for a part....right after the demon picks up the chainsaw zombie and throws him....hell shoot an imp or something on the left side of the screen....watch it slide down the wall.......that really impressed me.....(im easily entertained) and thats good physics in my opinion....and you people that have doom3 already I HATE YOU :(
 
Stuff moves, bodies bounce, things die with pretty movements.. Yay physics!
:)

What I wanna know is how people can say stuff like "HL2 graphics cant touch Doom3" when all they have seen are movies and screens from an early HL2 beta.
Honestly, in just about every Doom3 screenshot and trailer I couldnt help but think "It looks so plastic!"
Dont get me wrong, im still getting the game. Tonight at 7PM in fact. But to say HL2 cant touch Doom3 in graphics is getting a bit ahead of yourself. I thought the Source engine DX9 effects demo looked pretty good, and that was a year ago.
 
ATI has announced that the 4.9 Cats coming out in September will have huge Doom 3 boosts. Just a heads up.
 
Mincemeat said:
ATI has announced that the 4.9 Cats coming out in September will have huge Doom 3 boosts. Just a heads up.
Click to expand...

4.9? weren't those the beta drivers used in the orginal benchmarks?

Yep they were, there is no huge boost, there won't be a huge boost for a long time for ATi Ogl drivers.
 
You must log in or register to reply here.
Back
Top