*Game Development Issues*

[liolio]

Yeah, but it's not much of a secret that you're more than a little on one side of the fence.

The only thing this ensures is that the PS3 version doesn't look inferior to the 360 version. If that's your goal (and for Sony this would be a huge win) then this is good news, if you're just looking for games to be as good as they can be then you'd want the PS3 version to be brought up to the level of the 360 one, not the reverse.

I mean, Naughty Dog says they're only using 30-40% of the PS3 power, where's the problem?

I agree with you and Eastman.

And to some extend it makes me think of some Joker454's comments about the part of the budget that goes into coding.
PS3 force editor to invest more into coding, ok. 360 also benefit from some of this efforts, ok.

Doesn't say that editor could not spend more money on coding on the 360 IF they want to.

And as a possible buyer of the 360 I'm concerned by this shift especially if editors start to remove/not consider AA to make versions even, etc.

 

[Shifty Geezer]

A spin off would be is the cell overpowered in regard to the overall system?

Not like have spare of cpu power is bad but in regard to system limitations like:
GPU power
RAm available
Bandwidth
etc.

Like Ms wanted a cpu good enought to saturate the graphic sub system, it could be like cell even mostly 'untaped" is enought to sature ps3 other parts?

Only if you assume the workload should be balanced. If you have a preconception that graphics stuff is done on GPU and everything else on CPU, and you see the CPU handling graphics stuff, then it appears the CPU is overpowered and is having to make up for a weak GPU. But if you just view the system as processing potential to be used however you want, then whether something is being processed on the CPU or GPU doesn't matter, as long as it gets done, and the overall system has great results.

The real issue would be one of optimized efficiency. If the GPU can't handle some workloads you'd expect of it, and the CPU is being used in a way that is less efficient than a GPU doing the same job, then the transistor budget is likely imbalanced and some of those CPU transistors should have been apportioned to the GPU instead. If the CPU can handle the workloads as well as the GPU, it's not an issue.

It's no different to GPGPU work. You say the XB360 is better balanced with a CPU that feeds the GPU. Now if a dev decides to use Xenos to calculate some physics, would you say the CPU was now too weak for the GPU, even though it's seem well balanced up to now, or would you say that the developer is just spending their processing budget in a different way?

 

[liolio]

I bring the Ms old statement as an illustration, my point wasn't to compare with the 360 and more indeed a real question
And I know that cell helps it's not my point, it's really a ps3/ps3 question.

Like ok I run some particles on cpu, an impressive collisions system, some culling,postprocessing etc.

Ok you hit 40% utilisation then you could (supposition so my question) be limited by others factors:
for physic for example you run out of memory to keep tracks of the different objects for example
You face limitation due to communication between the gpu and cpu
Even if you cull a lot of polygone you can (if you want reach higher cpu utilisation) you can still send more than the gpu can swallow (after culling)
etc.

That what I mean by overpowered/oversized.

How good or bad is the xenon is ireleant, from what we know it's not a good cpu and it's not up to the task as Ms hinted that games are mostly cpu limited

 

[scificube]

I bring the Ms old statement as an illustration, my point wasn't to compare with the 360 and more indeed a real question
And I know that cell helps it's not my point, it's really a ps3/ps3 question.

Like ok I run some particles on cpu, an impressive collisions system, some culling,postprocessing etc.

Ok you hit 40% utilisation then you could (supposition so my question) be limited by others factors:
for physic for example you run out of memory to keep tracks of the different objects for example
You face limitation due to communication between the gpu and cpu
Even if you cull a lot of polygone you can (if you want reach higher cpu utilisation) you can still send more than the gpu can swallow (after culling)
etc.

That what I mean by overpowered/oversized.

How good or bad is the xenon is ireleant, from what we know it's not a good cpu and it's not up to task as Ms hinted that games are mostly cpu limited

You would be memory limited, or bandwidth limited, or triangle setup to vertex processing limited.

As the CPU is not the limiting factor in those scenarios it is not the limit.

It does not mean the CPU is overpowered. If the can CPU push more towards the GPU than it could ever handle its logically because the CPU should be able push as much as the GPU can handle and still process the myriad of other game tasks requested. If the CPU could not do this then those other tasks would become the limiting factor to the GPU getting fed.

The CPU would be over powered if there was no practical way the other components in the system could allow it to be put to good use they themselves has no need of its services. I don't think anyone can say that about Cell in the PS3.

As for comments about Uncharted using only 30%-40% of Cell processing capability ( despite it not being a very telling number to begin with ) Uncharted was a first effort on the PS3. The code was developed from scratch as ND moved from their GOAL programming language to C++ and of course introduced new toolchains, and programming techniques. Noteably ND had other challenges to face as ND mentioned they need to really really come to grips with shaders....which they did quite well. ND already divulged that they would have liked to implement fluid dynamics but the development schedule didn't permit it. This comment alone should suffice to say they didn't feel they had tapped Cell dry or that there was no practical target for the remaining processing power of Cell they left untapped with their first PS3 title.

In short I think the numbers ND gave out don't tell us much other than they were learning and adjusting like everyone else to a new console during much if not all of Uncharted's development. Still, most would agree, ND did came up with some pretty impressive results for their efforts.

 

[betan]

A spin off would be is the cell overpowered in regard to the overall system?

Not like have spare of cpu power is bad but in regard to system limitations like:
GPU power
RAm available
Bandwidth
etc.

Like Ms wanted a cpu good enought to saturate the graphic sub system, it could be like cell even mostly 'untaped" is enought to sature ps3 other parts?

While that may be so in some specific cases, I'd say the main reason behind underutilization is CBE's own structure (EIB latency, DMA, Local Store size etc). You don't need to look at the rest of the system really. There are (arguably artificial) examples which use Cell to the fullest literally. Those examples are enough to reason what runs good, what doesn't (as good).

That said, I can see how a bigger bandwidth to VRAM (both ways) would open door to trivial streaming applications that would run very well on Cell.

 

[scificube]

While that may be so in some specific cases, I'd say the main reason behind underutilization is CBE's own structure (EIB latency, DMA, Local Store size etc). You don't need to look at the rest of the system really. There are (arguably artificial) examples which use Cell to the fullest literally. Those examples are enough to reason what runs good, what doesn't (as good).

That said, I can see how a bigger bandwidth to VRAM (both ways) would open door to trivial streaming applications that would run very well on Cell.

I would say the state tools, the initial learning curve, and the fact affinity can't be leveraged in PS3 are/were just as substantial if not more so.

 

[T.B.]

Nothing involving an SPU is a piece of cake.
That said, the reason for lack of HDD usage is probably the other console as always.
(Of course it's also equally likely Sony may not be allowing continues streaming to HDD, but we know a couple of Sony games using background streaming, so maybe not).

You can basically do whatever you want with the HDD (well, there are size limitations, of course). As to writing to HDD from the SPUs, that's not going to work. SPUs can transfer to and from XDR, they don't have access to I/O (yay for public information!).

What's weird -for me- is that they seem to be recording animation frames as if animation is not deterministic (not even psedo-random).

Yeah, I was wondering about that. It kinda relates to the HDD issue, I guess. From the way he tells it, they simply dump frame-by-frame animation states into RAM and play that back, instead of doing animation/start/end triplets. Once the buffer is full, they drop frames to free up space. Now maybe they figured that they generate more data/frame then they can write to HDD. Maybe they need to full HDD bandwidth for something else. Who knows?

(Unless anyone has the game and can report little-red-light activity during gameplay. )

 

[nAo]

Yeah, but it's not much of a secret that you're more than a little on one side of the fence.

It's not much of a secret that a lot of ppl, you included, are not able to discuss this matter on a technical level and have to resort to these kind of arguments.
Where were you when I wrote about this methodology, perhaps dozen of times?
The search function is your friend, I started talking about it in 2005, even before putting my hands on any next gen console SDK/devkit, just with the information mostly available at that time (patents and industry rumours).

The only thing this ensures is that the PS3 version doesn't look inferior to the 360 version. If that's your goal (and for Sony this would be a huge win) then this is good news, if you're just looking for games to be as good as they can be then you'd want the PS3 version to be brought up to the level of the 360 one, not the reverse.

Wrong. This ensures, or at least promotes, a more sane development process for the sake of the whole project (and the reasons have been discusses and debated so many times that I won't bother to repeat them for you again). What ppl like you don't understand is that it's not about pimping your favourite project or console, it's about shipping on time a good multiplatform game.
What part of multiplatform you don't get? If you ship the game and the 360 version is ok while the PS3 version is rubbish (or vice versa) you, as a game developer, have failed. Obviously this doesn't mean that both versions have to be exactly identical.

I mean, Naughty Dog says they're only using 30-40% of the PS3 power, where's the problem?

Last time I checked ND doesn't work on multiplatform games, avoid retarded comments pls.

 

[T.B.]

What ppl like you don't understand is that it's not about pimping your favourite project or console, it's about shipping on time a good multiplatform game.
What part of multiplatform you don't get? If you ship the game and the 360 version is ok while the PS3 version is rubbish (or vice versa) you, as a game developer, have failed. Obviously this doesn't mean that both versions have to be exactly identical.

This is really a key point. You *do not want* one version to be inferior to the other. People will be less likely to buy your game for one platform if it looks better on the other platform. Nobody wants to play a gimped version. And we do want the money. Simple as that.

Another thing people often falsely believe is that there is a cell-optimal-model and an SMP-optimal-model. This is not really true, at least not for games. Almost everything you do runs faster in the "cell-model", even if you run it on Xenon. It also results in much cleaner code, which I guess people who haven't worked on large games cannot really understand. A lot of code is written with unmissable deadlines two days away, so if people simply multi-thread around to fork off some component and run it in parallel, you will see some pretty fun race-conditions.

So this is not really a 360-vs-PS3 thing. Both architectures need this model *badly* to perform well. 360 is a bit more friendly towards the old ways, which is an advatage early in the cycle. It might even be an advantage for the entire cycle, at least for performance-light titles. It's no free lunch, however.

 

[Shifty Geezer]

This is really a key point. You *do not want* one version to be inferior to the other. People will be less likely to buy your game for one platform if it looks better on the other platform.

Do we have any figures supporting this view, or some research? I know muliplatform issues make a lot of noise in the internet, but are sales really impacted by the knowledge one platform is inferior to another? Are there PS3 titles that people who would enjoy the game in the PS3 condition as-it-is aren't buying because they've learnt that the XB360 game is superior? I wouldn't have thought so. I would have thought that the condition of the game sells or sinks it, regardless of competing platforms. If a PS3 title is blurry and low-framerate, that'll be why it doesn't sell. I can understand internal and external forces encouraging equal platform performance, but I can't see it associated to sales regards differences.

 

[betan]

You can basically do whatever you want with the HDD.

Including the install and persistent data space, or just system cache?
I'm a little surprised at the freedom though, with all those annoying "this game saves automatically" messages.

As to writing to HDD from the SPUs, that's not going to work. SPUs can transfer to and from XDR, they don't have access to I/O (yay for public information!).

I believe at least on Linux, PS3 hypervisor reroutes SPE IO calls to OS@PPE via soft interrupts. I think standalone performance is pretty similar to PPE IO calls.

Yeah, I was wondering about that. It kinda relates to the HDD issue, I guess. From the way he tells it, they simply dump frame-by-frame animation states into RAM and play that back, instead of doing animation/start/end triplets. Once the buffer is full, they drop frames to free up space. Now maybe they figured that they generate more data/frame then they can write to HDD. Maybe they need to full HDD bandwidth for something else. Who knows?

Is it not possible that their replay system is just designed to work on HDDless systems and they simply didn't bother to implement an alternative?
ps: To be clear, that's not a lazy dev argument despite the look.

 

[aselto]

Is it not possible that their replay system is just designed to work on HDDless systems and they simply didn't bother to implement an alternative?
ps: To be clear, that's not a lazy dev argument despite the look.

I don't know how the replay system in Madden works, but these HDDless systems have games with replays that are 1.5 hours long. So please, put this argument to rest already.

 

[Mintmaster]

Wrong. This ensures, or at least promotes, a more sane development process for the sake of the whole project (and the reasons have been discusses and debated so many times that I won't bother to repeat them for you again). What ppl like you don't understand is that it's not about pimping your favourite project or console, it's about shipping on time a good multiplatform game.

I'm still not convinced that project directors are misallocating resources or making bad technical decisions, which is a fundamental but unstated component of your point of view.

Say there was no PS3. Would games be better if programmers spent less time on AI, gameplay, debugging, etc? That's what it would take for them to make the engine written "more sanely" so that it has nice data access patterns and is friendly to multithreading and in-order execution. You can't make something Cell-friendly for free, and you can't get Xenon benefits for free.

If the PS3 is changing your programming techniques, then you will end up with a worse game on 360 than you would otherwise. The only time this doesn't apply is if you make the assumption that given the same resources, the absence of PS3 causes games to be created less optimally. That's a pretty lame assumption...

 

[betan]

I don't know how the replay system in Madden works, but these HDDless systems have games with replays that are 1.5 hours long. So please, put this argument to rest already.

Unless someone claimed HDDless systems cannot have games with long replay duration, I don't see how your post is related to any argument at all.

 

[DrJay24]

Do we have any figures supporting this view, or some research? I know muliplatform issues make a lot of noise in the internet, but are sales really impacted by the knowledge one platform is inferior to another?

I don't know of any hard numbers, but there are examples where this has happened.

Check the first month sales of RB6:V2, you will see a relatively low PS3 number because of word-of-mouth. The online was broken (patched now?) and it may have affected sales. Compare it to competent multi-platform games like COD4, BO and DMC4 of roughly the same time frame.

March NPDs:
RSV2 360: 752k
RSV2 PS3: 155k

Feb NPDs:
DMC4 360: 295k
DMC4 PS3: 234k

Jan NPDs:
BO 360: 144k
BO PS3: 83k

COD4 360: 330k
COD4 PS3: 140k

I suspect the PS3 version of RSV2 is missing like 200k in sales due to being a poor version.

 

[Shifty Geezer]

I don't know of any hard numbers, but there are examples where this has happened.

The problem with the examples is we don't know the reason why less PS3 units were sold. We can guess it's because people heard the XB360 version was better. It could also be because the platform demographic is different. Or another game was competing for the market's dollars. Or other general reasons. In your case, if the online was broken, that'd be reason not to buy it, not that the XB360 version looked better.

 

[obonicus]

Do we have any figures supporting this view, or some research? I know muliplatform issues make a lot of noise in the internet, but are sales really impacted by the knowledge one platform is inferior to another? Are there PS3 titles that people who would enjoy the game in the PS3 condition as-it-is aren't buying because they've learnt that the XB360 game is superior? I wouldn't have thought so. I would have thought that the condition of the game sells or sinks it, regardless of competing platforms. If a PS3 title is blurry and low-framerate, that'll be why it doesn't sell. I can understand internal and external forces encouraging equal platform performance, but I can't see it associated to sales regards differences.

But isn't that exactly it? Most games are 'okay', and if the PS3 version is worse, then it's bad. In contrast, look at SC4: the PS3 version is the 'okay' one, while the 360 one has a higher render resolution. Perhaps as a result (rampant speculation warning!), we have reports that SC4 sold 2mil copies already.

 

[Shifty Geezer]

I don't know how the replay system in Madden works, but these HDDless systems have games with replays that are 1.5 hours long. So please, put this argument to rest already.

That doesn't put the argument to rest at all. If the issue is memory, XB360 has an extra 40 MBs to record data to for replays, why can't the devs write that data to HDD instead? That's the issue...

 

[wco81]

if they wrote replay to disk while you're playing the game, wouldn't that affect performance of the game?

Disk IO being orders of magnitude slower than the game which is loaded in RAM at the time?

 

[DrJay24]

The problem with the examples is we don't know the reason why less PS3 units were sold. We can guess it's because people heard the XB360 version was better. It could also be because the platform demographic is different. Or another game was competing for the market's dollars. Or other general reasons. In your case, if the online was broken, that'd be reason not to buy it, not that the XB360 version looked better.

I don't think we can blame the demographic (COD4, Ao2, etc. sold fine) or other recent releases (they were all multi-platform in this time frame). I didn't know the discussion was restricted to graphics, I was just giving an example of an "inferior version", which was inferior for online reasons.

We will never know why something did or didn't sell, we have to deduce it with the data we have.