Game graphics that surpassed Doom3?

[Phil]

tag:

With mediocre optimisation and incredibly poor main RAM, I can play, for example, Unreal Tournament 2003 at 60fps.

I wouldn't think Unreal Tournament 2003 is anywhere in the same league as Doom 3, neither is the NV2a with the Radeon 9500. True, Xbox is a closed architecture so a good Doom 3 conversion is likely, it will depend very much on the effort invested by the devs who port it though. Realisticly, I think a port of Doom 3 will run somewhere at 30fps...

 

[Tagrineth]

I wouldn't think Unreal Tournament 2003 is anywhere in the same league as Doom 3, neither is the NV2a with the Radeon 9500. True, Xbox is a closed architecture so a good Doom 3 conversion is likely, it will depend very much on the effort invested by the devs who port it though. Realisticly, I think a port of Doom 3 will run somewhere at 30fps...

Just giving an example.

 

[Paul]

This argument seems to be out of your league. Think back for a second, there used to be a time where 700MHz PIIIs used to be choice, the fastest Intel offering. They used to be in gaming rigs all over the place. I have one in use right now. I play games, it seems to be working fine.

We are talking about how Xbox's CPU will fare with DOOM III. There used to be a time when a p2 was the fastest Intel offering too, what's your point?

And the XCPU isn't exactly a P3.

My whole point was that for what was avaliable for the time, the XCPU wasn't a good choice if you wanted to use it primarily to run games on.

 

[Paul]

Well I should add, it wasn't *as* good of a choice.

 

[Tagrineth]

Well I should add, it wasn't *as* good of a choice.

As what? P4 was still too immature; other than that, HAVE FUN fitting a SPARC, Alpha, or PowerPC chip into an otherwise-x86-oriented system. THEN have fun teaching all the devs Alpha assembly for their games.


P6-derivative and K7-derivative were pretty much THE ONLY CHOICES Microsoft had; and to be perfectly honest, I believe they picked the right one.

 

[Paul]

Carmack and his crew are going to have to do alot of coding if they want it to run 30fps on Xbox without any details lost.

Maybe a few million thrown their way by MS will convince them to put alot of time into it.

 

[Tagrineth]

Carmack and his crew are going to have to do alot of coding if they want it to run 30fps on Xbox without any details lost.

Eh... why do you keep dropping back to this same argument?

I almost guarantee my P3 800 will be able to run DOOM3 at least at 30fps, if not higher, and again that isn't that heavily optimised. And don't even point me at the alpha leak.

The only problem with porting to Xbox will be RAM; I have 384MB total (256MB main, 128MB video), so that helps.

The amount of RAM in Xbox will be the deciding factor; levels will have to be segmented more (sort of like Halo's constant checkpoints... what do you think those are for? ) to fit into the RAM in "chunks".

But, for comparison, Serious Sam didn't lose anything at all in its Xbox conversion. The Xbox version looks more or less identical to the game running on my 9500 Pro (no, I don't use TRUFORM, the frame rate just dies with it on), minus resolution. SS's levels were tightened up though, which is actually a good thing cos some rooms were just too big...


But anyway, with careful ops, I'm sure DOOM3 could fit well enough in Xbox's main RAM with some kind of virtual memory scheme on the hard drive. Which eliminates the last 'absolute performance bottleneck' and leaves us with a system that should run DOOM3 just fine.

 

[Paul]

Eh... why do you keep dropping back to this same argument?

Because that was the original argument at hand, how Xbox would run DOOM III. I then brought out the fact that the XCPU could very well be a bottleneck as it's not the most spectacular CPU for gaming.

 

[Tagrineth]

Because that was the original argument at hand, how Xbox would run DOOM III. I then brought out the fact that the XCPU could very well be a bottleneck as it's not the most spectacular CPU for gaming.

Yes, and we just explained to you why it isn't a bottleneck!

 

[Paul]

For DOOM III, it is. For the PC version it will be anyway, for Xbox? They will code for it.

 

[ninelven]

edited

 

[Teasy]

As what? P4 was still too immature; other than that, HAVE FUN fitting a SPARC, Alpha, or PowerPC chip into an otherwise-x86-oriented system. THEN have fun teaching all the devs Alpha assembly for their games.

Why did you just completely miss out AMD's CPU's there?

 

[Sonic]

She did mention the K7 AMD in the same post, she was just comparing other CPU architectures that differ from x86.

I honestly don't see how Doom3 will have that much trouble running on the Xbox without a few memory optimizations. The XCPU in the Xbox isn't all that bad for gaming, sure it might be general purpose and not meant for games at all but that doesn't mean too much considering most of the graphics work will be done by the NV2A. What really does the CPU need to do besides physics and AI, and maybe some sort of transformation that the vertex shaders can't handle? There's the game code, which should run fine as I really don't think iD is known for bloated code in its games or engines. There also might be some sound processing needed to be done, not too much of a chore. The half P3 in the Xbox will be fine for Doom3. Sure, there's going to be optimizations done to get it all working fine on the Xbox, but it's not going to be so intensive a task like getting Max Payne to run on PS2 was.

 

[Teasy]

She did mention the K7 AMD in the same post, she was just comparing other CPU architectures that differ from x86.

I just thought it was a bit odd that she didn't mention why the K7 wasn't a better choice then the P3.

 

[Tagrineth]

She did mention the K7 AMD in the same post, she was just comparing other CPU architectures that differ from x86.

I just thought it was a bit odd that she didn't mention why the K7 wasn't a better choice then the P3.

K7 line runs hotter, draws more power.

And the P6 line runs faster per cycle, except in heavy FP-oriented tasks (where K7 only wins because it's using 3 FPU's versus P3's one - and P3 compares remarkably well despite having 1/3 the theoretical FP performance).

AND the P6 line supports SSE.

And finally... who would you rather trust? Intel with one bloody f'n huge cash reserve, gigantic market recognition, and the capacity to custom-modify the core (which they did), or AMD who's putting every last cent they have in the x86-64 gamble?

 

[Saem]

My whole point was that for what was avaliable for the time, the XCPU wasn't a good choice if you wanted to use it primarily to run games on.

Which has shown to be erroneous time and time again on this forum. Power consumption and heat dissipation are important factors with respect to performance. The Thunderbird FPU advantage quickly evaporates when SSE comes into the equations. If programmers use SSE there were realworld cases where you could see 800MHz PIII keeping within a few percent of Thunderbird 1.1GHz. Games, especially physics presents these scenarios. Not to mention the Coppermine has overall superior Integer performance thanks to it's good branch predictor and the Thunderbirds' rather weak one. This is especially important in AI.

Not to mention AMD didn't have fab capacity to spare at the time nor did it have the Palimino. Not to mention they were asking for more cash up front IIRC. People brought up the mobile durons, which has similar "max" power dissipation. IIRC if you looked carefully, it wasn't really max it was under a 60% workload. So power savings features were being factored in. Not to mention mobile parts are more expensive.

 

[bbot]

Here's what John Carmack has to say:


http://slashdot.org/comments.pl?sid=33453&cid=3619372

quote:
'Our "full impact" platform from the beginning has been targeted at GF3/Xbox level hardware. Slower hardware can disable features, and faster hardware gets higher frame rates and rendering quality.'

 

[archie4oz]

As what? P4 was still too immature; other than that, HAVE FUN fitting a SPARC, Alpha, or PowerPC chip into an otherwise-x86-oriented system. THEN have fun teaching all the devs Alpha assembly for their games.

What would be so difficult about that? Certainly PowerPC or MIPS would've had no problem being integrated as they already are in current consoles, and considering that an Intel and AMD cpus use different bus protocols/signalling and topologies going with some other arch would've posed the same issues... I sure as hell would take a 7450 core over a P6 core!

Besides learning Alpha assembly is pretty easy compared to x86...

 

[Tagrineth]

What would be so difficult about that? Certainly PowerPC or MIPS would've had no problem being integrated as they already are in current consoles, and considering that an Intel and AMD cpus use different bus protocols/signalling and topologies going with some other arch would've posed the same issues... I sure as hell would take a 7450 core over a P6 core!

Besides learning Alpha assembly is pretty easy compared to x86...

I meant because the idea was it was supposed to be very PC-like so PC devs might decide to make Xbox ports.

Or do you suggest MS make a new version of DX specifically for SPARC support?

 

[archie4oz]

No but they do have older (and or cut-down) versions for PowerPC, MIPS, ARM and possibly Alpha... Considering the drastic changes in DX8 compared to it's predecessors building SDK support for a different ISA would likely be on the small side of issues to be dealt with... Granted going x86 did make things easier overall...