Doom 3 on Dual 2Ghz/9800, run well enough to preorder Options

[flargh]

Doom 3 is GPU bound on the Mac on the same video card, but at much lower framerates compared to the PC. I don't think you can blame this on Aspyr or "porting overhead". The source is much more likely to be Apple's drivers. Even if it's ATI and nVidia that ultimately create the drivers, Apple is responsible for QA and performance testing. They must demand quality product from their vendors because, after all, it's their OS.

Since my article went up, I've heard back from people who point out that if you run a similarly clocked Athlon chip with the r_skiprender test, it'll return a much higher number than the PowerPC. I have to admit that at the outset, I wasn't aware of this -- I don't have an Athlon to test on.

But that definitely suggests than there's more than just OpenGL drivers at work here. I don't dispute that they probably play a factor, and may play even a significant factor, but driver optimization clearly isn't the only issue.

[Algol]

the truth is just that these games are designed for PCs first and then PORTED to apple. Then on top of that apple is not considering a gaming platform and thus steve jobs and his goonies have not gotten around to including the newest best OpenGL drivers. I think tiger will have newer drivers if you want to know the truth.

[Endymion]

Well, let me see.....

[a lot of non-specific stuff snipped]

If you some other explanation for the performance issues with Doom 3, I'd love to hear your theories.

Thanks for trying to answer the question for him, but I'm specifically looking for what he says Apple's GL is deficient in--if you'll look at his previous posts he's made the accusation before and had nothing to back it up. Telling us that a senior OpenGL ARB member doesn't know what they are doing with the API is a pretty tall order. I will admit that features have crept in over the years, but the games that use or need them have been far between. At this point I can only think of the shadowing issue with UT2004 that separates it from the PC version, and since he is speaking of Doom 3 here I would presume that he is implying that there is a disparity with the PC version here also--not just performance but features, as with the UT2004 shadows. And that is really considering a historical perspective, what right now is this continental divide with Macs' OpenGL that you think makes it a third-world citizen?

OpenGL performance is about as good as it gets on the Mac and the tools that Apple provides to debug and crank up the speed are second to none.

As for other matters it should be noted the review used an ATI X800, and the 6800 has been known to spank it with this game on other platforms.

And speaking of other platforms, here's one that performs way less than the Windows version also.

[dj phat 2000]

still does better then the Mac port though. And this was id doing the work right? From start to finish.

[flargh]

I think tiger will have newer drivers if you want to know the truth.

Uh, no.

[flargh]

still does better then the Mac port though.  And this was id doing the work right?  From start to finish.

If we're still talking about Doom 3, no -- Aspyr was heavily involved in the development.

[mrimac00]

Alright, if Apple or the OpenGL implementation or even the GPU drivers aren't the problem, who or what's to blame? Porting overhead? Our CPU?

[Endymion]

still does better then the Mac port though.

Please note the video card used--those are all nvidia cards with 6800 coming in the fastest. The macworld article used the X800. With overhead factors considered the Linux version is losing nearly half its speed to Windows. All of those tests were done on the same x86 hardware. At this time the Mac 6800 driver is a generic FX one, meaning, the same driver that enables the 5200, so if the driver is incapable of enabling the 6800's massive stencil/z buffer ability it may not fare as well as that on the Mac yet.

[wickedsteve]

Review

Doom 3 Reviewed on 3 different Macs

[Algol]

Review

Doom 3 Reviewed on 3 different Macs

OMG Those numbers are horrible. Apple's are fast, especially the G5 as can be seen from tons of benchmarks. They ride along with the best with many applications. But something is going horribly wrong with these game ports. Or the drivers on apple are horrible or who knows what. I'm going to barf.

[PeopleLikeFrank]

OMG Those numbers are horrible. Apple's are fast, especially the G5 as can be seen from tons of benchmarks. They ride along with the best with many applications. But something is going horribly wrong with these game ports. Or the drivers on apple are horrible or who knows what. I'm going to barf.

Those numbers were achieved with a 9600, a Geforce 5200, and a laptop. Not exactly ideal for a game like D3. If they had put an X800 or GF6800 in the G5s, it would have been better. I know people are disappointed with the numbers for this game on the Mac, but expecting it to run fast on anything but a fast G5 tower with a very modern video card is asking for that disappointment.

or maybe you're just joking.... I can't tell anymore with all these people ready to cry because of framerates... ;P

[monkeyphonix]

More cold hard benchmarks courtesy of Macworld. This time on Dual G5 and X800. Mac performance is CPU bound and apparently the higher cycles of the Pentium mean better numbers. Or something like that. I'm just looking forward to playing it at 640x480 at 20fps on my G4 1.5! Read for yourself:

http://www.macworld.com/news/2005/03/02/doom3/index.php

[Algol]

Those numbers were achieved with a 9600, a Geforce 5200, and a laptop. Not exactly ideal for a game like D3. If they had put an X800 or GF6800 in the G5s, it would have been better. I know people are disappointed with the numbers for this game on the Mac, but expecting it to run fast on anything but a fast G5 tower with a very modern video card is asking for that disappointment.

or maybe you're just joking.... I can't tell anymore with all these people ready to cry because of framerates... ;P

I wasn't joking. I realize they were testing it with old cards, but compare those results with the 2.5Ghz G5 X800 results... eek... not too good lol I mean the vastly faster system doesn't seem to really give you the boost you would expect.

[Mr.Rogue]

Those numbers were achieved with a 9600, a Geforce 5200, and a laptop. Not exactly ideal for a game like D3. If they had put an X800 or GF6800 in the G5s, it would have been better. I know people are disappointed with the numbers for this game on the Mac, but expecting it to run fast on anything but a fast G5 tower with a very modern video card is asking for that disappointment.

or maybe you're just joking.... I can't tell anymore with all these people ready to cry because of framerates... ;P

The Macs and graphics cards used in that review were all that were available to us, and to be honest most of the gaming demographic probably won't have, or cannot afford, a top end G5 with an X800 or GF6800. So the results show performance on the most common configuration of Apple's new machines.

[Dr. Geebs]

It's been said again and again: Doom3 does extra detail scaling above and beyond what the user sets as low, medium or high detail; so with a faster/better featured graphics card the framerate won't necessarily be much higher but you See More Stuff.

[Arenzera]

The peecee world is taking notice of these benchmarks and they are laughing their asses off at Apple and Macs. Although it is painful to have to listen to the ridicule from the peecee gamers, the real damage to Apple gaming is it sends a clear message to all the graphics guys with x86 boxes out there thinking about picking up a Mac for coding on to not bother.

And we are laughing our asses off at the day to day crap they deal with - viruses, a infuriating and hindering to use OS, junk for hardware. Need I go on? So, then you ask who got the crappier deal - us or them?

You have a point thou, Apple could optimise the OpenGL drivers (like they could produce a superbeast compiler) but simply choose not to.

Kiel :-)

[fugimax]

I know someone from Aspyr reads these forums, so..

What's the excuse for these benchmarks?

There have been hints at "clock speed" or "apple's drivers," but I'm interested to know specifically where the bottlenecks are.

(Note: A 2.5gHz P4 does much better than these benchmarks...so it's not raw clock speed that's at fault.)

And a more constructive question perhaps:

What are Aspyr's plans for Doom3 maintenance? Are you still working with Apple on performance tweaks in the drivers or code? Do you plan on releasing performance updates and not just required updates to match id updating the PC version? What's in the pipeline aside from the 5.1 sound that should have been included to begin with?

[Space_Pirate_Kil...]

http://www.barefeats.com/image06/mvp-doo.gif
These are the PC Benchamarks from Barefeats.com.

[dj phat 2000]

Of which we get about 1/2 the Xeon's FPS. Now, while they have a faster clocked CPU of about 1GHz more then the G5, and a faster video card in the X850. We are off buy about 1/2 there speed. And that is on the slow end. The AMD's SMOKE everything in this game. And in gaming in general. Personally, we should be running this game faster then the Xeon chips do. If AMD's lower clock speed is any indication of how strong there system architecture is. What does it say for the G5? I mean, we are just as fast if not faster in other area's of performance. That is of course out side of gaming. So, what's the problem? So far, at best we can say it's Apples Open GL software that maybe holding the game back. The fact that the game is not SMP aware. Just thinking that if the game was SMP aware, could we squeak out at least another 10FPS or so? Just that alone would be a MAJOR difference in playability on G5' and an of course G4's.

If all of this is not the case then, I don't know what else to say. Maybe Apple should give up Open GL and work on getting Direct X over to the Mac. Or something. Cause this sucks.

[Endymion]

Um, did you read the Linux comparison? It lacks SSE2 and suffers a big penalty because of this on the same hardware for any PC. Do you think a G5 has SSE2? Don't hold your breath about DirectX on the Mac, only Microsoft can do that, and they don't wish to, nor will they ever. Then there's the matter that even having DirectX won't magically give the hardware SSE2.