*Game Development Issues*

[liolio]

Why do you assume lack of marked improvement in the 360 version would mean lack of effort with the 360, rather than having achieved greater utilisation in 360 with earlier versions and having less room to advance the game? eg. If XB360 usage was 80% of peak in Madden '07, and 50% on PS3, those changes in PS3 that shift it up to 80% utilisation wouldn't have a major gain for XB360.

Shortly my point wasn't to try to start a discussion or not in this way.
EA is not known to make the most of the systems they work with.
They will gain experience on the 360 too, so they have no only to catch up with last madden rendition but also with improvement the 2008 edition should have.
Would sound to me like a huge effort (or poor depending on the system you consider)

How many people refuse to buy a game because they don't feel it's improved as much in relation to a rival platform? That's very strange behaviour. "Madden '08 is the best football game ever, and looks fantastic on XB360, but because the PS3 saw a greater improvement over the last version, I flatly refuse to buy the XB360 version." I don't see any sense or popularity in that viewpoint.

No I say that ps3 lovers feel cheated , if both versions are identical it would be a great news for gamers (nobody will feel cheated, and most of talk will be about how cool the game is not popping fans war) and thus for EA (as the new systems market share improve I feel like less and less gamers are likely to own more than one system, may be later when systems will be really cheap this will change again).

My point was it was a bad news for MS, ie EA don't feel like it makes sense to push the 360 engine harder (for me no matter how good madden will be on both system, both systems will still have room for improvments and devs are supposed to have more experience on the 360 and acces to better/more mature tools).

MS's whole point was easily accessible hardware that didn't require the developers to jump through hoops. If developers didn't jump through hoops and got good usage from the hardware, then they're going to get less improvements over time. You can't have it both ways! Huge improvements in PS3 are more a sign of a developer unfriendly system at launch that hampered efforts, than a focus on one platform over an other.

I agree on what MS goals were but I'm not sure they manage to push out a hardware that live up to theirs own expectations.
In fact I post a comment about the state (and some concern in regard to MS long term strategy) of development on the 360.(from outsider point of view) here:
http://forum.beyond3d.com/showpost.php?p=1146081&postcount=40

Shortly MS provide good tools the design is more commun but it looks like it's not that easy to make good use of the xenos (as it seems broken in some regard).
It looks like a statuquo for me "a good enough mantra", but it's good enough right now if I were MS i would be concerned (or not may be drinking on my yatch... ).

Maybe it could be an interesting topic about the 360 only to prevent miss interpretations (my point is not that 360 games have to look better or better forever).

 

[joker454]

I know its off topic, but just a curios question on how the dev business works. Are people, like yourself, hired on project basis? Ie you do super crunch for a title and then 1 week after it ships your out of a job and need to find a new one? How does R&R fit in there, do you have to "pay" for that yourself or was it just special in your case ie wanting to move on etc.

We're usually hired more based on need. For example, a studio may need a tools guy for one or more projects so they put the word out and hire. Currently, there is a fairly desperate need for PS3 coders, every place I interviewed at basically wanted to hire people to get their PS3 versions at parity with their 360 versions. Hence why I'm now a PS3 only coder When hired, typically you are kept as long as you perform well, and as long as the company financials are in good shape. They spend time and money basically training you, so it doesn't really make sense for them to just keep you for one project and let you go. That's only for coders btw, artists may be different.

The OS memory should be on par on both platforms now. The only RSX limitation we know today is the slower vertex input rate. So they will have to work around it (e.g., implement better culling, find more efficient ways to store/read the assets).

I hate to harp on these kinds of details, so I'll keep it very brief. OS memory is not at parity, 360 still has lots more legroom. Plus, there is much more to RSX limitations that just vertex.

I can go on all day, but I'll just give one example. Lets say you have a shader that uses different samplers based on different circumstances. So for some pixels it may choose texture sampler 0, other cases sampler 1, etc. On 360 dealing with this is simple. Vertex side, you calculate which sampler you need, and simply pass this index value to the pixel shader. Pixel shader side it would use this index value to chose which sampler it needs, and one tex2d call gets you your value.

You cannot index into sampler registers on RSX. So what do you do? You can still calculate the index vertex side and pass it to the pixel shader. Pixel side, there's only 3 solutions I can think of. One is to use is if/else statements to determine which sampler to use. This is bad, but sometimes you have no choice. Or, you can sample from all your samplers, and use lerp's to choose which of the values to use in a branchless way. This is usually better than the above, but still not great. Lastly, if you are lucky then you can group your texures into one uber texture, and just monkey around with the texture coordinates to pick which part of the texture you want. This is even better since it's now just down to one sample and no branches, but only works if you can associate the textures together, which isn't always possible.

It's stuff like this that comes up everyday, which is partly why so few coders want to work on PS3. Then again, its because of this that PS3 coders are in huge demand and my mortgage is being paid off fast, so I guess I should be thanking Sony

But it sounds like the focus is to bring the ps3 version on parity, not makes the 360 version better.

There is a reality going on today that people won't like, but here goes anyways. It's better to leave 360 performance unused, and have both 360/PS3 versions run at parity. It just makes good business sense. Why? There's many reasons, but in the end having a deficient PS3 version gives bad press, and reflects badly on the studio. On the other hand, putting out two versions that are identical reflects very nicely on the studio and yields lots of good press. Whether or not technical boundaries of a given platform are maxed out turns out to be irrelevant. Making a good looking game is important, but making it run the same on both machines is becoming even more important. If that means leaving performance/memory unused, then so be it. Is this happening today? Yes. Does it suck? Perhaps. Does it really matter? I don't think so. In the end, the consumer doesn't really know whats going on, they are just happy that their version is running nice. 360 owners still get a good looking game and are happy. PS3 owners get a version that runs the same as the 360 version, and they are happy. Everyones happy, which is good for business.

 

[patsu]

I don't speak of the OS size, but UMA vs NUMA, ie for that kind of games I feel like you have way more room for optimization with for the former.

It depends on the problem in question. I do not think we can say either way without knowing what specific issues they faced.

Not the point, as I state if we speak about madden, it's clear that EA was doing a poor job on the ps3, so they had a lot of room. But the game were fine on the 360, so we will have wait to see the improvement on the 360, if they are marginal +> it's clear that they have spend most of their efforts n the ps3.
It's clear that ps3 owners are angry to get inferior renditions, EA and others gain more at making games the same than at making the 360 rendition really improves.
Ie if the ps3 is now really good (60 fps, etc.) but the 360 has improved less but is still on top, a part of the market could clearly chose to pass on the game.

That could just mean (i) it's easier to extract performance for 360 initially, and (ii) PS3 s*cks hard at unoptimized code.

Rockstar do use tiling as the game is 720p + AAx2

Cool, but the point is there are limitations on 360 too (without calling PS3 into the picture). Otherwise, the Bungie team would not have to compromise. Clearly the developers have to choose their best approaches based on the given time and complexity also. This is not a "good enough" game for both platforms. They can't afford to. As I understand, the developer received substantial help from both camps.

As I say the problem is not how much the ps3 version improve but how much the 360 versiohn improve. Obviously there is more room on the ps3.
But marginal improvements on the wouldn't be a good news for MS.
In regard to all of this, by reading some comments on the games forum, it sounds like the issue is in fact widely exagerated and that it's clear that a lot of pro reviewers are really biased (cf dmc4, and clearer now...)

So We will know soon but it makes your comments and mines even more, anticipated

Are they really exaggerated ? All of them mentioned "slightly". I think people just commented on what they saw. You have some that look better on platform A and some on platform B. The rest are in the eyes of beholder.

I am personally more keen on their next game after they learned how to take advantage of Cell properly. I believe there are still more work to be done.


EDIT:

I hate to harp on these kinds of details, so I'll keep it very brief. OS memory is not at parity, 360 still has lots more legroom. Plus, there is much more to RSX limitations that just vertex.

I can go on all day, but I'll just give one example. Lets say you have a shader that uses different samplers based on different circumstances. So for some pixels it may choose texture sampler 0, other cases sampler 1, etc. On 360 dealing with this is simple. Vertex side, you calculate which sampler you need, and simply pass this index value to the pixel shader. Pixel shader side it would use this index value to chose which sampler it needs, and one tex2d call gets you your value.

You cannot index into sampler registers on RSX. So what do you do? You can still calculate the index vertex side and pass it to the pixel shader. Pixel side, there's only 3 solutions I can think of. One is to use is if/else statements to determine which sampler to use. This is bad, but sometimes you have no choice. Or, you can sample from all your samplers, and use lerp's to choose which of the values to use in a branchless way. This is usually better than the above, but still not great. Lastly, if you are lucky then you can group your texures into one uber texture, and just monkey around with the texture coordinates to pick which part of the texture you want. This is even better since it's now just down to one sample and no branches, but only works if you can associate the textures together, which isn't always possible.

It's stuff like this that comes up everyday, which is partly why so few coders want to work on PS3. Then again, its because of this that PS3 coders are in huge demand and my mortgage is being paid off fast, so I guess I should be thanking Sony

Interesting. But even so, if they use more textures and more detailed textures, wouldn't the pop-in be even more obvious on the 360 version ? All I am saying is both platforms have different strengths and weaknesses. It's more than picking out one attribute and say it's difficult/impossible to do. The issues highlighted may imply Sony guys have to work harder to achieve similar effects, but overall... there are still advantages on their platform for specific areas.

 

[assen]

There is a reality going on today that people won't like, but here goes anyways. It's better to leave 360 performance unused, and have both 360/PS3 versions run at parity. It just makes good business sense. Why? There's many reasons, but in the end having a deficient PS3 version gives bad press, and reflects badly on the studio. On the other hand, putting out two versions that are identical reflects very nicely on the studio and yields lots of good press. Whether or not technical boundaries of a given platform are maxed out turns out to be irrelevant. Making a good looking game is important, but making it run the same on both machines is becoming even more important. If that means leaving performance/memory unused, then so be it. Is this happening today? Yes. Does it suck? Perhaps. Does it really matter? I don't think so. In the end, the consumer doesn't really know whats going on, they are just happy that their version is running nice. 360 owners still get a good looking game and are happy. PS3 owners get a version that runs the same as the 360 version, and they are happy. Everyones happy, which is good for business.

This is the position Microsoft found themselves in last generation, when they were significantly behind in install base, and with a markedly more powerful hardware. How could they get themselves in the same trap again??

 

[Shifty Geezer]

I can go on all day, but I'll just give one example...

It's stuff like this that comes up everyday, which is partly why so few coders want to work on PS3.

How does this get dealt with in the PC space? I haven't heard lots of complaints about G70 or other achitectures being pigs to code. Do devs just leave it to the DX API and not know how DX handles that specific case? Can the same PC system's not be used on PS3, or does the PC application just run terribly inefficiently and gives people good reason to buy the next latest, greatest GPU? One of the big pluses to RSX was it mirrored the hardware devs have been using for years, so it's shocking to hear that that hardware is now hard to write for!

 

[nAo]

DX/OGL drivers hide lots if juicy/naughty architectural details

 

[Nesh]

Well based on some of the forum's game dev comments I am wondering if the PS3 is a significantly weaker platform than the 360. Looks like the 360 can do anything the PS3 can plus more? Although I wonder at the same time how each console fare if developers put all the efforts to create an exclusive game that squeezes out the most of each console's hardware.

There are some exclusive titles on the PS3 I d like a developer to comment on if they can be achieved 100% accurately on the 360. Such exclusive games like Uncharted, Heavenly Sword, Killzone 2 and Rachet and Clank.

How does the Cell differentiate the PS3 as a platform and does it differentiate it enough??

I want a good conversation mind, and not a vs war

 

[Shifty Geezer]

DX/OGL drivers hide lots if juicy/naughty architectural details

But does that result in inefficiencies? What is the dev's experience of wanting to implement different samplers in different circumstances in DX, and how is that reflected in the hardware use? If joker454 were trying to do exactly the same thing on a G71 as he is on RSX, how much trouble would that be to write in DX9 and how efficiently would it run on the hardware?

 

[liolio]

Patsu I won't make a long answer, if you read the post I linked (one of mine not trying to pretend it is great or full of insight) you would see that I agree too some extends and that in fact I have more "concerns" (slight in fact I still don't know if I will go with a ps3 or a 360 ) for Ms in the long run than hopes.

I disagree is on the hardware, I think that while the ps3 is more powerful, it's not that that will hurt MS first. It's the prowess of some teams working on Sony's exclusives and so the software (speak about a paradox for a company that advertised so much the tools it is providing).

Joker really nice to have your insights

 

[Nesh]

Wait. Do we know for sure if the PS3 is more powerful? I d like to know if there is any specific information supporting this, because for around 2+ years I ve been reading mixed opinions on the subject or intentionally ambiguous or vague statements

 

[liolio]

This is the position Microsoft found themselves in last generation, when they were significantly behind in install base, and with a markedly more powerful hardware. How could they get themselves in the same trap again??

In fact parity is a good thing for them in regard to ps3 potential

But MS weakness is clearly that they are not willing to secure some talents and provide them with ressources and enough freedom.
MS is really in need for some dev teams to put the 360 strengh to good use and you need something built up from ground for the 360.
And worse even if MS manage to gather talents in devs team that would be on part with sony ninjas, they would run behind schedule ie insomnia/naughty boys etc, are already working on their engine V2 if not 3.

 

[patsu]

liolio, I agree the software + available hardware resources will make the final difference.

My real questions is: Are we maxing out the hardware already ? I would think at this point, people now understand and know how to avoid the limitations on both platforms. They may have also figured out their own "nextgen" workflow. However, the developers still have not exploited the hardware as much as they would like.

e.g., GT5P have damage, more polished environment and online infrastructure to complete; R2 will add the missing streaming component, plus significant improvements in visual, physics and online combat; NaughtyDog needs to go back to enhance their water and god knows what. MotorStorm 2 needs to work out their load time, split screen, plus even more crazy physics and levels. I am sure similar things can be said on the 360 side.

Cross platform developers may be half or one step behind the exclusive ones.

 

[joker454]

Well based on some of the forum's game dev comments I am wondering if the PS3 is a significantly weaker platform than the 360. Looks like the 360 can do anything the PS3 can plus more?

Hold on, let's backtrack a bit. I definitely would not say the PS3 is significantly weaker. In fact there are some obvious parts of it which are far superior (cpu and storage). The problem is that, in the 'general' sense, it is weaker. By 'general' I mean for typical multi platform studios that don't want to maintain multiple engines and codebases, and who deal with a variety of game types in one studio.

For example, I think PS3 is far better suited to deferred rendering than 360 is. This advantage though is limited in the real world. First, forward/deferred rendering reminds of of the risc/cisc arguments back in the day. Which is really better? Sure, you can demonstrate situations where deferred rendering is better, but the reverse is true as well. If you are a studio that makes a variety of games, then do you really want to gamble a chunk of time/money on deferred? What happens if its better for one type of game, but worse for another? Also, memory is already very tight, can you spare more for deferred? The big problem though with going deferred, is what if you want to make a 360 version? Financially, supporting just one platform doesn't make sense anymore. I'd wager that Insomniac must be wondering how many more millions they would be pulling in if they had 360 versions of their games. As previously mentioned though, at least in my opinion, 360 is not a good platform to go deferred. So what do you do? Does the studio build and maintain two engines? Not a chance. For most multi platform studios you stick with a more traditional rendering engine. This does not play as well to the PS3's strengths, so that's why I say in the 'general' sense, it leaves it at a disadvantage. Plus it has yet to be proven that deferred yields better results. We'll see I guess with the inevitable Gears 2 and Killzone comparisons. What deferred has shown is that it takes tons of time and money to implement, which only leads to scare more studios from it.

There are some exclusive titles on the PS3 I d like a developer to comment on if they can be achieved 100% accurately on the 360. Such exclusive games like Uncharted, Heavenly Sword, Killzone 2 and Rachet and Clank.

I think Uncharted visually could be done on 360, but it would need a total rewrite. If they just ported what they have and tried running it on 360, it would run like total ass. I forget what their tech docs said, but I think they run 8 passes or so through their engine. Yikes, that would be like ~3-4ms just spent on resolves on the 360, let alone bringing the cpu to slideshow levels! So, ignoring dvd space limitations for the moment, could they get Uncharted visuals on 360? I think so. Could they port Uncharted to 360? No chance, it would have to be re-written. And there in lies the rub. For Naughty Dog it makes no difference, they are owned by Sony, so they are PS3 only, period. For the rest of us, financially it must work on both platforms so we can't go the Naughty Dog route.

 

[Shifty Geezer]

I'd wager that Insomniac must be wondering how many more millions they would be pulling in if they had 360 versions of their games.

Hmmm, I don't know that they would be. Insomniac are doing just fine, ranked in the top ten companies in the US to work for. Going multiplatform would be a whole load more work and potentially impact their performance as a top-tier developer as they have to compromise their products to fit two different machines. You can always make more money, but there are sacrifices involved to.

So I think being exclusive still has merits, especially as long as cross-platform titles remain difficult to implement. If you have the choice of releasing a mediocre game across two platforms, or a top-tier AAA version on just one, the latter may well have better returns. Depending on the company and who's funding you, it might actually be worth considering.

 

[Arwin]

This is the position Microsoft found themselves in last generation, when they were significantly behind in install base, and with a markedly more powerful hardware. How could they get themselves in the same trap again??

Or you could argue that they wanted to get themselves into this same trap, but this time on the receiving end. If they had the better hardware last time, this time they've just tried to get something that is close enough, but can be sold at a lower cost. So basically, they have aimed for the PS2's position in terms of hardware, and tried to benefit even more from their software development support strengths and PC crossover stuff. The biggest answer that Sony has and that Microsoft is not always successful at answering, is their excellent software support. Though not perfect, the Sony development strategy is both adventurous and technically proficient. Microsoft's still occasionally lacks in both of those. But so far, they're doing very well I'd say.

 

[makattack]

Interesting. But even so, if they use more textures and more detailed textures, wouldn't the pop-in be even more obvious on the 360 version ? All I am saying is both platforms have different strengths and weaknesses. It's more than picking out one attribute and say it's difficult/impossible to do. The issues highlighted may imply Sony guys have to work harder to achieve similar effects, but overall... there are still advantages on their platform for specific areas.

I think the "pop-in" can be attributed to the fact that on the 360 version, assets are streamed off DVD vs. the PS3 version off HDD, which features a mandatory 3+ GB installation.

 

[liolio]

I just want to point out that MS exclusive has failed to deliver so far (I put aside GeowIII as UE III is availlable to anyone).

WHat would be a good use of the 360?
I feel like native resolution is irrevelant to costumer (halo III, Cod for best sellers MGS/MGO might be the next one, and countless of others).

What resolution would allow for AAx4 while needing only two tiles? (EDIT didn't some calculation would be pretty low ~520p)
Go for it, it's likely to be good enough if the package is good enough. -(EDIT 520*960 is really close to sd but I'm not sure that the average costumer would care if the package is good, lot of thing going on/ smooth/DOF/some motion blur/etc. some game are 576p so why not assume this route?)

use the shading power left for more complex shaders/ particles/whatever you can run on the gpu to offload the cpu).

Try to use the tesselator.

find some use for the left bandwidth on the edram

For the cpu it's trickier as the xenon seems bitchier than what a lot of people think:
far from optimal altivex implementation.
L2 subject to cahe traching
no that much cache.
So it needs more than its fare share of optimisations

The sad part is that nobody want to go though the expanse of creating something custom on the 360 / rare who seems iddle...
I don't even speak about trying to reach the dedication some sony team go through to put out technical marvels...

 

[joker454]

Hmmm, I don't know that they would be. Insomniac are doing just fine, ranked in the top ten companies in the US to work for. Going multiplatform would be a whole load more work and potentially impact their performance as a top-tier developer as they have to compromise their products to fit two different machines. You can always make more money, but there are sacrifices involved to.

It depends on how they built their engine. If it's 'relatively' close to a typical forward renderer, then porting it to 360 would be easy. They wouldn't even have to do it themselves, they could pay a third party to do it all for them so it wouldn't impact their native working environment at all. They are doing fine for sure, but when you see the multi million sales of other games like COD4, etc, I can't help but wonder how many more units their Resistance and Ratchet franchises would sell if available on 360.

 

[Shifty Geezer]

Sure they'd sell more, but would they be happier? Kinda like I could work longer hours and work more, and get more money, but I don't want to If you can find a niche for yourself and you're happy there, why start chasing after something else? A company doesn't have to pursue growth in international megacorporation size, even though that's a common goal!

 

[patsu]

I think the "pop-in" can be attributed to the fact that on the 360 version, assets are streamed off DVD vs. the PS3 version off HDD, which features a mandatory 3+ GB installation.

The install is only 3+ Mb (3340Kb), not 3Gb.
http://forum.beyond3d.com/showpost.php?p=1153113&postcount=634