*Game Development Issues*
- Thread starter valioso
- Start date
- Status
Yes, 1080p it is.
*To other posters, please continue this line in Upscaling thread. Relevant posts are copied in there.
-AlS
eastmen said:
I remember reading that the 360 was the lead platform for it. As in that's where development started & the assets were created first.
MazingerDUDE
Regular
thegamesmaster
Newcomer
Yeah, i remember reading something similar in an interview...i will post the link if i find it again.
Leading on PS3 does NOT leave you with a as good or better X360 game (if the other choice is leading on the X360. Leading on the X360 is obviously better for the X360. )
What your trying to say is that leading on the PS3 and porting to the X360 gives you a better results, than leading on the X360 and porting to the PS3. The relative ports that is. The X360 version does NOT get better by leading on the PS3 vs leading on the X360, that much is obvious.
Now, since all devs are constrained by budgets, and PS3 development is more complicated and whatnot, it makes sense to consentrate your efforts on the platform that sells the most games aspecially when this platform also is the easiest to makes games on.
Leading on the X360 results in a better X360 game. Leading on the PS3 results in a better PS3 game and a relatively better X360 port, compared to a PS3 port of the X360 version. You do not end up with a game that plays better on both platforms (X360 lead, obviously is better than a PS3-> X360 port).
You may argue that combined, the PS3 lead and X360 port is "better", than the X360 lead and PS3 port, but as long as X360 development is easier, faster and the X360 consumers buys more games, why should they consentrate their efforts on PS3?
It means a worse game on the platform that sells the most games.
Last edited by a moderator:
I think this is the core of the matter. For most software it's:
1. Schedule
2. Budget
3. Features/quality
Since the 360 is deemed the more mature platform, - and importantly easier to work with for junior staff, the perceived risk is lower, hence it becomes lead platform.
The low number of people working on the PS3 version (judging by people's comments here) allows the studio to invest minimally in the PS3 version and cut their losses if the version fails to deliver.
Cheers
This has been discussed very often, and there are two important points here. The first one extensively discussed was that developers who lead on PC/360 typically tend to write a certain somewhat outdated programming style that does not center around data-streaming. The 360 is better at running this kind of software even if it's simply because it's easier to distribute different tasks over the three homogenous cores. You can't get away with this on the PS3 as easily as on the 360, as you'll run into performance issues and you'll be forced to go for a better (for instance streaming) model sooner. However, this model also gets more performance out of the 360, resulting in (hopefully) a better product. Depending on what your goals are, this may eventually lead to a better value for your programming bucks. Which leads to ...
You need to see this in the context of where the biggest part of the budget needs to be spent. If I've understood correctly, these days typically more than 75% of the budget is spent on art-creation, and only a very small part of the budget is spent on engine design/hardcore programming.
And it doesn't stop there either. Typically programming work on the second platform is allocated a lot less time to, because you don't have to reinvent the wheel - the actual structure of the game and logic and so on, most of that has already been figured out by the first team. The second team only has to translate this to another platform, a job to which traditionally a fraction of the developers will be assigned.
Currently, the ratio between PS3 and 360 worldwide is 15 / 20. Say that it takes you 25% of your budget to increase your market by 75%, then that clearly increases your chances on making a healthy profit.
Now what has been a problem is that the effort needed to bring a game to the PS3, particularly when your programmers come from a PC style programming background, is more than just a handful of port-guys. Particularly when you start working on this too late in your cycle towards release time, you're going to spend a lot of extra money and run out of time. This leads to a product that will potentially sell less on your second platform than it could have.
To make a long story short, good planning when you start, and recognising the effort required for the PS3 in time, and using that to design engines that make better use of consoles in general, and the PS3 in particular, makes you more money.
There are a few other factors in play here, but I think if you look around, you'll see that publishers are picking up on this and you'll see things are already changing.
Well the surprising part of the theory is that you do get a better XB360 game, because leading on the PS3 forces developers to rethink ways of doing things that gets better yields on the XB360 (and PC to boot) which otherwise are overlooked as the existing and less effective methods are used on those platforms. I dunno how true that is; that's something for the actual developers to argue!
Of course a huge amount of the quality of the non-lead title is dependent on porting strategy. You can port 'tightly' to a platform with high-quality reqrites, or port with hacked code copies. No amount of designing for one platform or other will yield good results with a platform if the porting strategy is rubbish, and leading on PS3 won't always lead to better XB360 versions even if the theory that designing for PS3 results in better XB360 designed code holds up.
Just force the devs to rethink ways of doings things to get better yield on X360 leads?
I understand that the solutions are often overlooked for simpler methods, but you cannot just say that PS3 lead means better X360 game compared to X360 lead. As long as the developer is open for new ideas to begin with, X360 lead >>>> X360 port from ps3 every single time.
There is nothing stopping these developers from using these benefitial solution that are being used on the PS3, other than talent\knowledge\time. Thus PS3-> X360 port is NOT more benefitial than X360 lead. Its rather straightforward logic and math.
I don't think you understand. When you architect an application from the ground up based on a certain type of software model and CPU architecture; considering the PS3 and 360 share the major similarity of having multiple cores/processors (be they SPUs, or cores), you'll naturally gain on both when you design that way. This is not to say that you can't make an application that just simply caters to one better than the other, but you get a net gain on both if your application attacks a problem in a manner that suits a multi-processor architecture.
Joker454 disagreed with that statment.
But it sounds like the overall quality of the game (so including both versions) would be overall better with the ps3 as the lead version.
But it's different than saying that it will lend to a better 360 version.
Anyway I can't think of multiplatform being optimal anyway.
Tiling is no longer an issue and not implement that much (solution like the one chosen in SC4 for example is weird to ridiculous imho). As Joker454 stated some RAM seems to end unused on the supposed "optimal" 360 renditions, etc.
That's not to say that efforts should not be more balanced in regard to the PS3.
But it sounds like the overall quality of the game (so including both versions) would be overall better with the ps3 as the lead version.
But it's different than saying that it will lend to a better 360 version.
Anyway I can't think of multiplatform being optimal anyway.
Tiling is no longer an issue and not implement that much (solution like the one chosen in SC4 for example is weird to ridiculous imho). As Joker454 stated some RAM seems to end unused on the supposed "optimal" 360 renditions, etc.
That's not to say that efforts should not be more balanced in regard to the PS3.
Yes, that's the crux of it as I understand it. If the XB360 were willing to embrace the New Ideas, then the XB360 version would be best. But if they're not, and the XB360 is comfortable coping with the Old Ways, choosing PS3 first does force them to do things differently, grow out of old habits and learn new ways fo thinking of things. The choices are :
1) Lead on XB360 with the best methods possible and get the best possible XB360 game, then port to PS3 which can handle the well-designed engine.
2) Lead on XB360 with a cheaper, more comfortable design that isn't optimal for the platform, and then port to PS3 which is a difficult struggle with lacklustre results.
3) Lead on PS3 with the best methods possible because the machine can't handle anything else well, then port to XB360 with those best methods in place.
Option 1 is the ideal given the market, but from the sounds from some developers, isn't happening, and as long as XB360 is the lead, it's bad for both systems.
What the best possible practices is for the PS3 isn't necessarily the best possible practices for the 360.
Yes, thinking about locality will be a performance win for both platforms.
But the differences in the memory model and in the optimal memory transaction size (cache line vs. optimum DMA size) as well as in differences between CPU/GPU performance specifics (raw GFLOPS, vertex/pixel/unified shader performance etc.) may lead to very different solutions.
Cheers
Yes, thinking about locality will be a performance win for both platforms.
But the differences in the memory model and in the optimal memory transaction size (cache line vs. optimum DMA size) as well as in differences between CPU/GPU performance specifics (raw GFLOPS, vertex/pixel/unified shader performance etc.) may lead to very different solutions.
Cheers
The idea is that for multiplatform development (see topic), nobody is likely to achieve 'the best possible practices' for any platform. Thus in a project where PS3 and Xbox360 are the machines behind considered, using the PS3 as the lead platform will lead to better results overall because it will necessitate the re-thinking of programming methods. This will benefit the Xbox360 in ways that leading with the Xbox360 are unlikely to bring due to its tolerance for non-optimal coding and human nature to go with the path of least resistance. In the end, this is theorized to bring the best results in a multiplatform environment where resources and focus is divided, unlike a first party or platform-exclusive studio.
The above is basically a summary of Shifty's point number 3.
I would like to add of course that yes, of course it is still possible to develop a game that makes the best possible use of a single platform. If you are sure that you'll be able to do something extraordinary on that one platform that is not possible on other platforms, and you think that will translate in a higher uptake on that single platform that compensates sufficiently for the lost sales potential on the second platform, or you can make a significantly smaller investment during development (particularly if you're not art-limited in your game, or if you're as smart as, say, Media Molecule and make content creation really easy ), or you get a nice bonus from the platform holder, or any combination of the above, then yes of course it can make a lot of sense to develop something for a specific platform.
), or you get a nice bonus from the platform holder, or any combination of the above, then yes of course it can make a lot of sense to develop something for a specific platform.This has always made me believe that the PS3 has lower performance
It's an indicator that the way to access the performance is more specific/less forgiving. As an example that showcases how this can be so, consider one CPU (A) that has a single vector unit that can process one vector per clock, clocked at 3 GHz, and another CPU (B) that has four vector units capable of processing four vectors per clock if packed right, clocked at 2 GHz. If you design and implement your data to work in batches of four vectors, you'll get 2*10^9 * 4 vector operations per second = 8 Billions ops/s. On CPU A you can only process one op per clock, for a maximum of 3*10^9 = 3 billion ops/s.
But if you design your code for CPU A and use that 3 billion ops/s and don't use a data format that matches the array of vector units on CPU B, CPU B would also run at a single vector operation per clock, and being clocked lower, would max out at 2 billion ops/s.
Cell requires programs to be properly designed to access it's potential, otherwise the potential just sits unused/untapped (inefficiently used).
But if you design your code for CPU A and use that 3 billion ops/s and don't use a data format that matches the array of vector units on CPU B, CPU B would also run at a single vector operation per clock, and being clocked lower, would max out at 2 billion ops/s.
Cell requires programs to be properly designed to access it's potential, otherwise the potential just sits unused/untapped (inefficiently used).
I think that it's rather less flexible... I mean you could use RSX to maximum with it's strenghts in mind like pixel shader performance and anisotropic filtering and Xenos will achieve similiar results because of its flexible unified shader architecture.... And both machines will be used in the most efficient way... But if you design a game on xO which make for example lot of vertex shader computation you can't port it to PS3 easily because it's not so good at this...
And there is opposite situation - if you design a game which uses lot of Cell resources (for physics, postprcessings, particle effects etc ) you can't port it easily to Xenon.
Just my "I dunno what I'm sayin" opinion...
- Status
Similar threads
