Carmack's Hands On Impressions of Xbox 360 & PS3

[Neb]

Far more memory in each pool? I doubt 320MB~ VRAM is the current mainstream, let alone Mac.

I hardly doubt it is the current mainstream since many budget cards and G71 cards have 512MB to begin with (but it could be country specific). But that really doesn't matter since it is not what dictates the limits. Thats why a lot of PC games whants 512MB VRAM for highest settings, and that why there exists ingame options so that the lowest denominator don't act as a bottleneck!

Look, at the current stuff he's showing, it freaking amazing. A multi-platform running at 60fps...it looks very good...He makes some disparaging remarks about the PS3, and now he's a noob, stupid, lazy? Has anyone else managed to do what he has done?

I am amazed so many bashes him, I mean he has showcased a very good looking game (shadows, textures, geometry etc) that runs at 60fps.

considering all the talk about 60fps being the real deal, now there is a FPS game that will run at 60fps AND look good at the same time, on all platforms!

 

[ninzel]

Irrelevent, I didn't say UE3 was better (I'm no fan of UE3 titles so far either from an art direction standpoint), the subject was about the skin shaders. The main character from the movie has plasticy old Doom-engine look and his hat looks like it coulda been ripped right out of D3 art assets. What am I comparing to? How about Crysis, Heavenly Sword, even Bad Company. iD's character models always look like wax figures who have ingested too much colloidial silver and came down with Argyria.

.

I've yet to see any dev do skin well yet. Ya sure a lot of devs do the wax figure look,but then a lot of devs do the opposite with skin that has no real qualities. The absence of waxiness,doesn't automatically mean the presence of realness.

 

[AlexV]

I've yet to see any dev do skin well yet. Ya sure a lot of devs do the wax figure look,but then a lot of devs do the opposite with skin that has no real qualities. The absence of waxiness,doesn't automatically mean the presence of realness.

Considering how difficult skin-rendering is, due to its peculiarities WRT interaction with light, and complex minor detail, that's hardly surprising. The thing is that the waxy look jumps right out and blows any kind of illusion whatsoever, whilst something less "specularish"(think what HL2 tried to do, that was fairly good looking, though simple, and the illusion was better IMHO than either D3 or FarCry Mme. Tussaud's approach) is considerably better. There are also some titles that will approximate(well, fake) SSS, and that should provide some improvement.

 

[Subtlesnake]

Now, that's not true and please stop trying to put words into my mouth and paraphrasing things I never said. I did not say Carmack had no concept of the power of the PS3.

You mean the statement that Rys qualified with "I'm looking at you now, PSman"?

He clearly states that the 360 and PS3 are closer than any other 2 consoles so his opinion is pretty evident. What I did say is that he doesn't really care for the PS3 architecture - which is probably why id hired a PS3 engine specialist to optimize their id Tech5 engine on the PS3 (a task Carmack did not want to undertake).

Whether he likes a particular piece of architecture is a separate issue from whether he can program effectively for that architecture. If Carmack only hired the Cell specialist because he didn't want to do all optimisation himself then all that proves is his temperament!

You're going to have provide evidence that Carmack isn't capable of adopting the appropriate programming model and effectively developing for the Cell.

There is no question Carmack has adapted to multicore processor development on the PC, but he is on record as preferring single core high MHz processors.As for Cell development, bringing multicore development paradigms that work on other multicore processors onto the Cell without adapting them to suit the radical architecture of the Cell will usually result in poor performance.

Yes, and Carmack himself has stated many times that you can't treat the Cell like a regular multicore processor. You said that Carmack "[won't] even contemplate switching to a data centric view of processing where the flow and packaging of data is just as important as the code processing it", and yet in his Quakecon 2005 keynote he does just that:

"Now the tough decision that you have to make is how you deal with the CPU resources. Where you might say that if you want to do the best on all the platforms, you would unfortunately probably try to program towards the Sony CELL model, which is isolated worker threads that work on small little nuggets of data, rather than kind of peer threads, because you can take threads like that and run it on the 360. You won’t be able to get as many of them, but you can still run, you know you got three processors with two threads, or three cores with two threads in each one.

So you could go ahead and make a game which has a half dozen little worker threads that go on the CELL processor there, and run as just threads on the 360 and a lot of PC specs will at least have hyper threading enabled, the processor’s already twice as fast, if you just let the threads run it would probably work out ok on the PC, although the OS scheduler might be a little dodgy for that -- that might actually be something that Microsoft improves in Longhorn.

And that’s kind of an unfortunate thing that that would be the best development strategy to go there, because it’s a lot easier to do a better job if you sort of follow the peer thread model that you would have on the 360 but then you’re going to have pain and suffering porting to the CELL.

"I’m not completely sure yet which direction we’re going to go, but the plan of record is that it’s going to be more the Microsoft model right now where we’ve got the game and the renderer running as two primary threads and then we’ve got targets of opportunity for render surface optimization and physics work going on the spare processor, or the spare threads, which will amenable to moving to the CELL"

Another quote I found from his keynote:

"In terms of where the things I think hardware should be evolving towards, honestly, things are going really well right now. The quibbles that I make about the exact divisions of CPUs and things like that on the consoles, they’re really essentially quibbles. The hardware’s great, I mean everybody’s making great hardware, the video card vendors are making great accelerators, the consoles are well put together, everything’s looking good."

http://www.freakygaming.com/compani...pt_of_john_carmacks_quakecon_2005_speech.html

In his 2006 keynote he also says that the SPEs should be treated as "acceleration engines" rather than processors, where tasks are parcelled out for the accelerators from the main thread.

 

[Arwin]

Regarding that last comment, I'm starting to understand that some developers are now starting to move towards SPU centric development. Rather than put everything on the PPU and send out some stuff to the SPUs, you put everything onto the SPUs, the SPUs themselves divide work among each other, and the PPU is left over for some specialist stuff that the PPU.

 

[betan]

I don't know why we have two threads for pretty much same stuff, but for those who care, I remember reading WWDC Mac demo was using a single core, and a single OpenGL thread.

 

[Cheezdoodles]

I don't know why we have two threads for pretty much same stuff, but for those who care, I remember reading WWDC Mac demo was using a single core, and a single OpenGL thread.

because the other thread was in the technical forum, but people rather wanted to debate Carmacks merits than the tech that is used in this engine

 

[ninzel]

Considering how difficult skin-rendering is, due to its peculiarities WRT interaction with light, and complex minor detail, that's hardly surprising. The thing is that the waxy look jumps right out and blows any kind of illusion whatsoever, whilst something less "specularish"(think what HL2 tried to do, that was fairly good looking, though simple, and the illusion was better IMHO than either D3 or FarCry Mme. Tussaud's approach) is considerably better. There are also some titles that will approximate(well, fake) SSS, and that should provide some improvement.

I suppose you're correct,the waxy look does jump out more. Whereas the less shiny look just blends in with the rest of the game.

 

[pjbliverpool]

Far more memory in each pool? I doubt 320MB~ VRAM is the current mainstream, let alone Mac.

Well Macs a Mac and not a PC like Nebula's statement was targeted at. The majority of gaming PC's may still be 256MB of video RAM but thats still 256MB guarenteed which is at least on par with both consoles while 512MB and above GPU's are extremely common and easy to come by these days.

Certainly is you wanted to build a half decent gaming PC today, you should aim for at least 320MB IMO.

Obviously system memory size doesn't even warrent comment.

 

[pjbliverpool]

Im suprisingly impressed with what id have achieved here. The graphics looks excellent, at least on UE3 level which considering its running at 60fps on all 4 platforms is damn good. I honestly expected that level on cross platform compatability and performance to seriously hamper the graphical quality but it looks like I was wrong.

In still hoping the game will scale in some way on the PC above just resolution but even looking as it does now, it should be up there amoungst the best.

 

[Xenon]

You mean the statement that Rys qualified with "I'm looking at you now, PSman"?

I believe there were other paraphrases which were misinterpreted.

Whether he likes a particular piece of architecture is a separate issue from whether he can program effectively for that architecture. If Carmack only hired the Cell specialist because he didn't want to do all optimisation himself then all that proves is his temperament!

You're going to have provide evidence that Carmack isn't capable of adopting the appropriate programming model and effectively developing for the Cell.

Once again, I believe my main point was that Carmack would rather not deal with the Cell architecture because it's his least favorite architecture. To say that he's incapable of "programming effectively for that architecture" is a gross misstatement. If John wanted to devote his time to the Cell processor then I have no doubt that he would be able to contribute best practices that would be adopted by all.

Yes, and Carmack himself has stated many times that you can't treat the Cell like a regular multicore processor. You said that Carmack "[won't] even contemplate switching to a data centric view of processing where the flow and packaging of data is just as important as the code processing it", and yet in his Quakecon 2005 keynote he does just that:

And I still stand by that remark. I mean, has he switched to a data centric model? Do you think the id Tech5 PC and 360 engine have been radically modified to support a data centric processing perspective that uses schedulers to parse out work? I highly doubt it. His keynote addresses what is required to efficiently run on the Cell, not that he's adapted his entire code base to it. And why would he? When you have a development paradigm that is perfectly applicable to the PC, Mac, 360 why would you switch to using a paradigm that is only preferable on the PS3?

In his 2006 keynote he also says that the SPEs should be treated as "acceleration engines" rather than processors, where tasks are parcelled out for the accelerators from the main thread.

That's good to know, however, the SPE's are really PowerPC (601 type) processors that are more than "acceleration engines". Yes, each SPE is a self contained computer that doesn't need to be controlled by the PPE.

 

[abbaon]

Assuming their website is up-to-date, they've got 34 employees. I don't mean to demean Mr. Hollenshead's job, but I guess one could say there are 33 people working on it directly.

Willits says 26 on the team. Cloud and Dean are on Quake Wars, and you can probably take out Nix, Stratton and Jackson, but I don't know who else to discount.

 

[Farid]

This thread is going nowhere. It seems most people didn't even listen to that keynote in full and just jumped on a random out-of-context quote to display some of their home brewed drivel.

Carmack make some perfectly valid points in this keynote. But more importantly, he said nothing new, outrageous or unexpected. He has always been vocal about his dislike of multi-core architectures, so why would anyone expect him to be positive toward an asymmetric, using local memory pools and different ISAs, multi-core architecture such as Cell? He points that some people may like that, but he, for one, doesn't.

And that's his prerogative to do so, he's not fond of complex multi-core processors and he says just that. He never said he would stop working on these multi-core platforms, he isn't trying to blame the hardware for a bug ridden release, nor is he denying that multi-cores CPUs (asymmetric as well, see some of AMD/Intel's projects) represent the path the industry is now following, he just make known that he doesn't like the multi-cores, Cell in particular, since it's the epitome of multi-cores (7 available cores, two ISA including a new one, SPEs addressing local RAM, etc...).

Based on that previous point, he goes on to say that working with the PS3, and thus Cell, requires much more work and nuturing to get things running well. What is so wrong or negative about that? It's just a factual statement. Furthermore, it is nothing we didn't know or hear before. He even goes on to say that the PS3 version of the engine "will be just as good or potentially better in some ways, but it will have required much more effort than the other platforms." [Paraphrased by Shacknews] A claim that can hardly be considered negative, no matter how you want to slice it.

The PS3 supports a more complex architecture, it has immature development tools compared to the top-notch stuff available on the Xbox platform, and thus it's harder to work with... More news at 11'? Seriously, what about these claims hasn't been discussed many time son this very board? Carmack will not say that the PS3 is easy to work with just because some people want their favorite piece of CE to be the "most powerful, easy to code for and best selling" machine out there.

On the same page, he says that the game memory footprint is a problem since the PS3 has less available memory, because of the bloated OS SCEI came up with. Here again, what is supposed to be new or wrong about that? His plan is to have the same assets on every platforms, he has two 512MB consoles to work with, he thus expects to have the same amount of memory to work with. And when he doesn't get the same available on both platforms, he gets annoyed about it. Nothing groundbreaking, here as well.

Now with that said, let's try to get some valuable discussion going on in this thread. The whole confrontational passive agressive nonsense, between clueless, drivel spouting console supporter faction A versus console supporter faction B trying to justify their irrational hate for a CE product misusing these statements made by Carmack and the nonsensical reaction from the other "faction" as an opportunity to channel their own unrelated negative feelings toward an unanimated kit of electronic machinery, will only get this thread locked and many posting rights suspended.

 

[ihamoitc2005]

Carmack

Carmack is saying what many other devs have been saying for years now, I could be wrong but I remember him a few years ago criticizing multicore CPUs without aknowledging the fact that superfast single core CPUs are not pratical anymore.

This is true my friend, I think 2 years ago he said he prefers one big very powerful processor than multicore because it has less development cost and complication. It is no surprise for me that he does not like PS3 because it has 7 times more cores than he likes, and also cores are different types, and also sometimes one type is main core sometimes other type is main core. Maybe it gives too many choices for him so he does not know which approach to have for this design because each choice has different complications and benefits. I think for him CELL is not a good design but for other developer with different approach it can be good. I think this is why he does not like PS2, it has too many different processors and methods. This is my guess.

 

[nAo]

I don't know any developer that does not share Carmack's point of view, we would all prefer to work on an incredibly fast fast single core machine. Said that this incredbly fast single core macihnes don't exist and won't for the foreasable future, and it seems to me that a lot of devs got this right in their minds a long long time ago..
Replace 360 and PS3 CPUs with an intel or amd OoOE core and then we would have now a last year PC as next gen console, saying that the transition was done too early is not, imho, an extremely smart thing to say, at least from a technological standpoint.
Yes..Wii is outselling everything out there but do we really care about it, again, from a technological standpoint? I don't, most of you don't as well (given the fact that this is a technology oriented forum) and Carmack does not as well.

 

[ERP]

And a trillion others on machines between ZX81 and AtariST. Of course people would only remember those if they played games in that era.

Was MidiMaze something you worked on as well?

Nope I just played it a lot in the dorms at University.

 

[Subtlesnake]

Once again, I believe my main point was that Carmack would rather not deal with the Cell architecture because it's his least favorite architecture. To say that he's incapable of "programming effectively for that architecture" is a gross misstatement. If John wanted to devote his time to the Cell processor then I have no doubt that he would be able to contribute best practices that would be adopted by all.

Well, Carmack himself has stated many a time that he prefers developing for the 360, and is not a fan of the choices Sony's made. So I don't think you'll find anyone disagreeing with you there. And with your acknowledgment that Carmack could in fact program effectively for the Cell, then all the argument really comes down to is whether Carmack is prepared to develop for the Cell. So you're questioning his commitment, and I don't see what grounds you have to do that.

And I still stand by that remark. I mean, has he switched to a data centric model?

The issue is whether he contemplated doing it (not whether he actually did it), and in his keynote he did just that. I mean all we have are Carmack's actual words, so what grounds are there for superimposing your beliefs over what Carmack actually said?

Do you think the id Tech5 PC and 360 engine have been radically modified to support a data centric processing perspective that uses schedulers to parse out work?

No, because Carmack has already said that a 'peer threading' model is best for the 360 and PC.

His keynote addresses what is required to efficiently run on the Cell, not that he's adapted his entire code base to it. And why would he? When you have a development paradigm that is perfectly applicable to the PC, Mac, 360 why would you switch to using a paradigm that is only preferable on the PS3?

To get the PS3 up-to-par with the other two platforms, running at 60 FPS? Carmack has told us that a lot more effort has been put into the PS3 port, but yes we don't know how they're managing the SPEs and what role Carmack played in this and what the PS3 specialist is being used for, and that means that any claim about Carmack's role is pure speculation. He's told people that he's prepared to look at a 'worker thread' style system for the PS3, so it seems rather arbitrary to me to decide that actually he isn't, or actually he doesn't want to put the required work in, or actually he'd never commit to the required programming model.

 

[Cheezdoodles]

I don't know any developer that does not share Carmack's point of view, we would all prefer to work on an incredibly fast fast single core machine. Said that this incredbly fast single core macihnes don't exist and won't for the foreasable future, and it seems to me that a lot of devs got this right in their minds a long long time ago..
Replace 360 and PS3 CPUs with an intel or amd OoOE core and then we would have now a last year PC as next gen console, saying that the transition was done too early is not, imho, an extremely smart thing to say, at least from a technological standpoint.

But lets moderate that statement, we know that we wont have any incredibly fast single core machine. But what about being O-o-Oe or not ? Isn't this also a significant part of this? That console cpu's are in order based, and the AMD\Intel cpu's are not?

A bit off topic, but do you think we would been better of with a quad-conroe than a Cell?

 

[Arun]

Replace 360 and PS3 CPUs with an intel or amd OoOE core and then we would have now a last year PC as next gen console, saying that the transition was done too early is not, imho, an extremely smart thing to say, at least from a technological standpoint.

I don't think I agree with that completely... I look at it as more of an engineering time & budget problem. A Core 2 Solo wouldn't be very big if it existed, but IBM never created a CPU core with an IPC comparable to that, and if you look at the direction they're taking with the Power6, possibly never will.

I don't think anyone would deny that one Conroe core and one less SPE per die would have been a more compelling solution than the current CELL. Obviously, the problem with such an alternative is that IBM doesn't design Conroes, and Intel doesn't design SPEs. Ah well!

 

[patsu]

To get the PS3 up-to-par with the other two platforms, running at 60 FPS? Carmack has told us that a lot more effort has been put into the PS3 port, but yes we don't know how they're managing the SPEs and what role Carmack played in this and what the PS3 specialist is being used for, and that means that any claim about Carmack's role is pure speculation. He's told people that he's prepared to look at a 'worker thread' style system for the PS3, so it seems rather arbitrary to me to decide that actually he isn't, or actually he doesn't want to put the required work in, or actually he'd never commit to the required programming model.

It takes more than just "'worker thread' style" to optimize for PS3. The memory layout, who (SPU/PPU) does what and how many, the way the data are encoded and grouped to hide latency manually, etc. will be affected non-trivially.

If JC is "only" prepared to look at worker thread, then there is still much room for improvement. Then again, code and opinion do change. We may need to give JC and Olick some more time to sort things out. There is only so much 1 person can do after-the-fact. A PS3 optimized version will take time and extra resources regardless of whether JC is pro or against Cell design principles.