Doom Perfomance

Let's go back to the interview Carmack gave to VoodooExtreme where he mainly talked about the new doom engine.

In the first part, he said the following:

"I am hoping that the absolute top-of-the-line system available when the game ships will be capable of running it with all features enabled and anti-aliasing on at 60hz, but even the fastest cards of today are going to have to run at fairly low resolutions to get decent frame rates". "Many will choose to drop a feature or two to get some speed back, but they still won't be able to get near 60hz."

In the second part he said the following:

The card that the new Doom will work best with is not available as of yet. But there are some good ones in progress. The existing cards on the market are as follows:

nVidia GeForce 2: We are using these as our primary development platform. I play some tricks with the register combiners to get a bit better quality than would be possible with a generic dual texture accelerator.
(Editors note: id's primary development platform today is GeForce 3 - Watch the MacWorld Doom movie and read Carmack's last .plan update for more about it).

ATI Radeon: All features work properly, but I needed to disable some things in the driver. I will be working with ATI to make sure everything works as well as possible. The third texture unit will allow the general lighting path to operate a bit more efficiently than on a GeForce. Lacking the extra math of the register combiners, the specular highlights don't look as good as on a GeForce.

3DFX Voodoo4/5, S3 Savage4/2000, Matrox G400/450, ATI Rage128, Nvidia TNT2: Much of the visual lushness will be missing due to the lack of bump mapping, but the game won't have any gaping holes. Most of these except the V5 probably won't have enough fill-rate to be very enjoyable.

3DFX Voodoo3, S3 Savage3D/MX, Matrox G200, etc: Without a stencil buffer, much of the core capabilities of the renderer are just lost. The game will probably run, but it won't be anything like we intend it to be viewed. Almost certainly not enough fill rate.

Zaldron, if by Carmack's words the fastest cards of today will run the game like shit (that's what he meant), then how come the ancient TNT2 won't have any gaping holes?
I know that it will be missing lots and lots of new effects (Real Time Bumpmapping, Real Time Shadows, etc... due to lack of Vertex and Pixel Shaders) but by Carmack's words, it is enough to be enjoyable even if it is fill-rate limited! (Meaning it will run fast enough to be enjoyable but not very).
So what I don't get is how come the first GeForce won't be enough to play the game if a TNT2 is enough?


My specs are:

P3 1GHZ

256mb SDRAM

Creative Annihilator Pro (Geforce DDR chipset)

Sound Blaster Live! Value

What frame rates do you think I will get and I will I be able to play in 1024x768 resolution with most options on? (guess based on what you know).

Remember, the MacWorld demo was run on a 733 G4 and I have a 1GHZ P3!
Carmack has a 1GHZ P3 as well which makes me happy :)

P.S

"We did a ton of testing the last two weeks while we were putting the demo together".
"The 733 G4 was not as fast as my 1 ghz PIII in any of the trouble areas.
Apple is doing a lot of good work, but the CPU's just aren't as fast as the x86 ones."

John Carmack

This is the comment Carmack made on Slashdot regarding the frame rate in the Doom demonstration on MacWorld.


Now, remember the exclusive news I didn't want to give you all cause I wasn't sure they were true?
Well, I still am not sure but I will give them anyway.

Taken from my source (translated from english to russian):

"Currently, the minimum machine at id consists of PIII 350 (they must have meant Pentium II 350) with 128mb of ram and the first GeForce."
"In the next 15 months (editors note: a clue to the release date?), when AI and additional details will be added and the code will be optimised, the minimum system requirments might rise to a Pentium III 550 with the second GeForce."

The news also mentioned that Gray Matter Studios (currently working on Return To Castle Wolfenstein, the amazing sequel to Wolf3D using the Q3TA engine) helped id in the creation of the models for the demonstration and also helped them in the making of the demonstration itself. (that must be the main reason cause of which id allowed them to use some part of the code of the new doom engine).

I honestly believe that there is a slight chance that everything I said here is true (the news mentioned the plan update of George Broussard too and he did update it).

Bah. By fall the GeForce 3 will be old news. The Radeon 2 and GeForce 3 Pros will rule the market. By spring, the GeForce 3 will be ancient and completely obsolete.

Spring and fall are the video card market hot seasons. If we can assume that Doom 3 won't ship for exactly 1 more year... then we can have GeForce 4's (or whatever the nVidia successor chip is).

I'm assuming the GeForce 3 MX will have 64 megs of memory. That's probably the card I'll get.

Having a non-GF3 card won't halt the Init phase common in every id engine. The world, models, animation system, renderer and other subsystems will load, but without being enhanced by the new features.

No T&L? The game will run like shit.
No Pixel/Vertex shaders? No bumpmapping, dynamic lighting, highlights in the world. Expect classical gourad shading.

Every element necessary to play the game will be there, but then again, that's like playing Q3A engined games with no textures, without weapons being drawn and no sound...

About your specs :

WHEN DOOM3 COMES OUT, GO FIND A TOP OF THE LINE VIDEO CARD.

Remember that OpenGL pretty much follows Carmack's direction. Buy something that runs DooM 3 even better than the GF3 and you'll have a card ready to play at great speeds most of the DooM-engined games will see shortly.

deadnail said:

By spring, the GeForce 3 will be ancient and completely obsolete.

Yes, video hardware has been moving faster than Moore's law lately, but isn't this a wee bit of an exaggeration ?

That depends, spring here were I live? (almost end of the year) In that case, he's right.

my new specs (hopefuly):

Dual P3 1 GZh's
1024 mb pc133 SDRAM
GeFORCE 3 ultra deluxe
two 45 gb ibm deskstar HD's going raid.

Dual P3? I would go for an Itanium.

As soon as Microsoft pulls out the 64-bit compatible C++ compiler in their Developer Studio, games should start to smoke...

Lord FlatHead said:

Yes, video hardware has been moving faster than Moore's law lately, but isn't this a wee bit of an exaggeration ?

No, it's not!

The GeForce 2's came out after the GeForce. In what way was the GeForce superior to the GeForce 2? None, it's completely obsolete... hence, the savage price drop to get rid of the dinosaurs to make space for the new wave.

99% of computer users are using completely obsolete equipment. Why? Something new comes out the next month. Most people that buy computers are buying an obsolete machine. The P4 1.7 ghz system costs too much? Well, let's get a P3 1ghz instead. You just bought something obsolete.

Hell, my SNES is obsolete as hell but I still have fun with it. =)

Dima said:

, Matrox G400/450, : Much of the visual lushness will be missing due to the lack of bump mapping,

Surely he is wrong here? My Matrox has bump mapping, and wasn't Matrox the 1st to introduce it?

Very well said. A lot of people are sweating cannon balls because they think DoomIII is going to require a GeForceIII card. You've shown that Carmack had the engine running on earlier cards, and that even cards like Voodoo5 will run the game to some degree.

There is one thing, though-- all you really need to do per-pixel bump mapping is the dot product instruction. I think most GeForce cards have this, and any other card that is designed around the Direct7-type api's. Carmack even came up with a way to do it without any special hardware, and might have made it into QuakeIII, but he decided it was just too much of a drain on the art department, and it wasn't a nessesary effect for the game they were making. I think he did it by using a separate normal map for each direction, or something, and just calculating the net light reflecance value.