Except T.B.(also a developer) argues almost exactly that: the Cell-friendly approach helps with 360 games. His view (forgive me if I'm misinterpreting), which seems quite sensible is: the more restrictive Cell forces you to treat parallelism in a more constrained, more controllable way. This can be applied to the 360 as well, helping you avoid race conditions or other pitfalls, particularly when quick additions/changes to the code are needed.
*Game Development Issues*
- Thread starter valioso
- Start date
- Status
Except T.B.(also a developer) argues almost exactly that: the Cell-friendly approach helps with 360 games. His view (forgive me if I'm misinterpreting), which seems quite sensible is: the more restrictive Cell forces you to treat parallelism in a more constrained, more controllable way. This can be applied to the 360 as well, helping you avoid race conditions or other pitfalls, particularly when quick additions/changes to the code are needed.
Of course this can be applied to the 360 as the lead platform also. I don't believe that a crappy developer will code poorly on the xbox 360 , yet when it comes to the ps3 that developer will step it up and not only get amazing performance out of the ps3 but also the xbox 360. They could have just steped it up for the 360. I think its just a bs reason. The 360 sells alot of copys of games and it has the larger install base esp when taking into account the western market. Why should the 360 gamers suffer any slights in development process by playing second fiddle to the ps3. Whats going to happen is that the xbox 360 ports will suffer or we will be stuck with a half assed version to make the minority happy....
Last edited by a moderator:
Why should PS3 games suffer by being second fiddle to the 360?
Crossplatform games by their very nature should be as near identical as possible on both platforms. If notionally leading on one particular platform can result in a better game for both, then it's good news all round, surely?
The argument is that a developer can use suboptimal methods on the XB360 and get respectable results; they would have no need to step it up. Than applying those same methods on PS3, they end up struggling. This imapcts their sales on the PS3. To address this they learn to develop for PS3 more effectively producing good results, and see that those same techniques can be applied to their XB360 efforts to improve what they were achieving there.
With this argument, if the developer doesn't 'step it up', their software will suffer on PS3 and sale accordingly poorly. Thus by necessity they will improve, in theory, which is where we see these reports of developers shifting focus to PS3. Bear in mind these PS3 oprimizations aren't hard-core voodoo magic. Any competant developer can learn the skills and techniques just as they've learnt whatever skills and techniques they've had for developing any games whatsoever. The major problem is the effort of learning new skills, when the old skills were perfectly fine for XB360 and PC, and the acquiring the new skills and adapting your methods are an added cost you'd rather do without. It's as if (figures just for illustration)
Old Way = 75% utilization of PC/XB360, 30% of PS3 utilization.
New Way = 90% utiliztaion of PC/XB360, 90% of PS3 utilization.
The effort required to learn the New Way jsut for PC and XB360 isn't worth it, but throwing PS3 into the mix, it becomes a requirement that benefits the other systems. It's an added workload for the developers, but a win for consumers, and some developers suggest even a win for developers once they've got on top of it.
Are you saying that without the PS3 there would be more recources for the 360 and therefor the 360 games would be even better?
A. The lead in install base is shrinking, so devs cannot simply ignore the PS3 base like it could back in 2007.
B. When has the 360 version suffered during port or lead PS3 game? Do you have data?
Mod : Keep it civil and on topic please.
Last edited by a moderator:
Oh they are, or better they were. That's why a lot of them are changing their tune. I'm not a manager/producer/etc..but many believe they can take any decision, even at technical level, because those smart hardworking passionate engineers will take care of it anyway. Unfortunately they were wrong, just wait and see if you don't agree with me.
I never said is free, I've just said it's a better practice IF you are developing a multiplatform game. As a developer you have to deliver on both platforms, not just one.
We can spend a week talking about pros and cons of each platform, this won't change the outcome which is: I have to develop a good game on two platforms, and but must be good on both.
All your argument is flawed because is based on a reality where PS3 doesn't exist, but it does exist, it doesn't matter how much you dislike it, IT DOES EXIST
If PS3 wouldn't exist then games created just for 360 wouldn't be so unoptimal, but working on two platforms dictates rules for both of them, it doesn't matter if you like one or the other.
I think he resource argument is pretty bogus.
Games go multiplatform to increase sales projections, which in turn increases budget and resources. Generally you wouldn't ad a SKU if the increase in resources wasn't offset by the increase in sales projections.
You can argue that building a single platform game can be more efficient, and that's likely true IMO. A developer is allso much more likely to spend the time optimising to a plaforms strengths if they don not have to worry about how a certain feature it will run on other platforms.
Having said that as hardware gets faster the gap closes a lot.
Games go multiplatform to increase sales projections, which in turn increases budget and resources. Generally you wouldn't ad a SKU if the increase in resources wasn't offset by the increase in sales projections.
You can argue that building a single platform game can be more efficient, and that's likely true IMO. A developer is allso much more likely to spend the time optimising to a plaforms strengths if they don not have to worry about how a certain feature it will run on other platforms.
Having said that as hardware gets faster the gap closes a lot.
There's also been discussion on this board that this new way of doing things may even prove fruitful on PC and next MS console in the future. The speculation hinges on the idea that the multi-multi-multi-symmetric cores aren't the way things will shape up. The thought was along the lines of '360 developers managed to avoid learning this stuff NOW, but they'll have to learn it eventually'. Lots of hidden assumptions there, though.
Actually it isn't the parallelism that's the issue on PS3 relative to 360, it's the memory architecture. The real question is what does the memory architecture of the future look like is it PS3 with it's seperate local memories with manually managed DMA transfers, or is it more conventional with some sort of automatically managed memory heirarchy?
I don't personally know, though I suspect long term it's more likely to be the latter. It'll be interesting to see which direction Intel goes in with their 8+ core systems.
Word of mouth is definitely strong especially in game shops because these are the people we buy games from and a-lot of them follow gaming news from mags and forums and can have an influence on the customer. I remember being in a gamestop when two parents came in with a teen to buy a ps3 but they ended up leaving with a 360 because the the sales associate.
It really does does put the argument to rest - if someone suggests that 360 Core is restricting devs from implementing HDD usage for replays, then one can look no further than at Halo 3, when hard drive is required for the replay feature (perhaps works with memory units as well, but in the game menu I was told that HDD is required for that) and the owners of the core console simply won't watch any replay. Can you watch replays of entire matches in Madden? The last time I played a sports game was NHL 99 (and I think I'll buy NHL 09 this year, because the game receives lots of praise) and if I remember correctly you could watch entire matches there after they were finished, as well as replay of the last action.
As for the issue on Madden 09 on PS3 - I suspect they simply didn't have enough time time to implement the feature correctly on PS3, it's not a case of "the reason for lack of HDD usage is probably the other console as always". They have only twelve months between each release and I doubt EA would allow a delay for one SKU, so the left it as it was.
Mintmaster
Veteran
You're missing the point. When you say that coding for Cell first helps Xenon, you are already making a comparison to a situation where the PS3 didn't make much influence on design considerations. I only explicitly mentioned the exclusion of PS3 to make the optimality point clearer.All your argument is flawed because is based on a reality where PS3 doesn't exist, but it does exist, it doesn't matter how much you dislike it, IT DOES EXIST
If PS3 wouldn't exist then games created just for 360 wouldn't be so unoptimal, but working on two platforms dictates rules for both of them, it doesn't matter if you like one or the other.
Maybe coding for the PS3 first instead of 360 results in a smaller overall game quality drop on 360 than gain on PS3, but the drop is still there.
Again, the no-PS3 case was just to illustrate optimality. Just because the practices that make code Cell-friendly also makes it run better on Xenon doesn't mean that implementing them will give you a better game. These practices and others are ignored when making a 360 game because they take away resources from other parts of a game, and the benefit isn't worth the cost.
For any given multiplatform budget, you will spend more on coding and less elsewhere if you want PS3 parity. This, and this alone, is my point.
You can force the 'cell' way of cpu coding on 360 coders as well. On PS3, you can make a very thin code layer that runs between spurs and other peoples code. Coders still have to adhere to the local store idea of things and hence arrange their data structures accordingly, but the lower level details of slicing up data to dma them in chunks, etc, are all hidden. This identical code layer can also be tweaked to run on 360, the difference being of course that spurs is out of the picture and the dma's can be yanked out. But to the other coders, everything looks exactly the same since they are dealing with a similar interface. They still adhere to a local store way of coding and arrange their data for that, even though they are coding on 360.
It doesn't take much to enforce this. The thin layer is cake to write, and the lead british-dude-with-a-ponytail coder can enforce it.
I got the impression that Content was the most costly part of any game this generation. Adding 15 million potential customers to your game, even if it required a complete seperate coding team, should reap economical benefits.
Unless we were told something different in regards to content?
Okay, what you're saying is:
SMP-optimal-model = Cell-optimal-model, and vice versa.
But Non-Cell optimal model != non-SMP-optimal model. So there is still a difference between these 2 models. What most are saying is people can get away with the non-optimal situations more on 360.
I look forward to the day when someone can show the "differences" due to the extra CPU power (visually or otherwise).
Except you're going from the assumption that the 'easy' way of working on the 360 is the optimal one, when other developers have mentioned that it ain't so. At the very least, it's not a given truth.
And I think you may be missing their point: following your thought exercise, suppose there's no PS3, not even Cell. What developers in this thread have argued, essentially, is that if some 360 developer came up with an architecture that follows the same principles as 'Cell-friendly architecture' in our universe, they'd have a more robust architecture for the 360, full-stop. What multiplatform development involving the PS3 does is it makes this more robust architecture necessary for good cross-platform performance, rather than optional for better single-platform performance.
This isn't as true as it used to be.
Engineering teams have gotten VERY large at some of the major publishers. In some cases as big as the content teams.
But it doesn't require a completly seperate coding team so the point is moot.
All this discussion is dependant on some concept that games are even vaguely optimal though, and for the most part it just isn't true. As engineer count increases, the more shipping a game becomes about software engineering and the less it becomes about technology.
Sure at the core tech level, things still have to be vaguely optimal, assuming your not just buying it in these days, but my experience with gameplay code (which is the bulk of the codebase) on large teams is that optimisation is way down the list of things people are thinking about, until it becomes a prformance issue.
If you mean Larrabee then Ars just had an article on it with a diagram that seems to suggest a conventional cache architecture but "local subset" is also interesting.
This is veering off into the next-gen (or next-next gen or whatever) hardware discussion, or there's probably a Larrabee thread somewhere).
- Status
Similar threads
