*Game Development Issues*

Status
Not open for further replies.
I think you guys misunderstood assen. He illustrated his quotation marks "will get used to PS3" /end quotation marks with "things will be easier next year. PS3 development will always need more effort, better programmers, etc."

His first "comment" wasn't a very good choice of wording (but not everyone here speaks English as a native langauge).

No it isn't flawed because PS3 development doesn't get any harder over time!!!

Assuming there is no evolution and complication in the code, then that is correct. But games in 2009 will be more complex and demanding the system "do more with less".

Some issues will be trivial in that "good practices" as well as tools&code already ironed out can be leveraged; in other areas it will be the same issues of getting the same performance from a SIMD architecture with small local memory with more manual management and a smaller total system memory configuration.

Sure 360 development may advance at the same rate but that doesn't mean PS3 ports will get any worse..

It all depends. We are seeing some of both right now -- some really bad ports as well as some really good "ports" (co-development).

It all depends on the studios. Are they picking designs that work well on both systems? Do their techniques translate well to the SPE architecture? Are they trying to port to the PS3 or is it true co-development? Are they giving the PS3 adequate resources so they can adequately design systems that make efficient and, minimally, equal end results (performance and quality) as the 360 effort?

Marketshare and development costs are going to be major factors--as well as the continued quality improvements in the tools Sony and their partners make available. While the cool physics system you developed in your 2006 may carry over in large parts, your player animation systems was spartan in your FPS and now that you have a 3rd person game with intricate animation demands you will need to not only create a more robust system to deal with the complexity, but also find smart ways to map this adequately to the PS3 system.

As individuals become more accustomed to the conventions and quirks of the platform they'll start to innovate & work out their own solutions to exploit it..

One of the hurdles is that the number of people who can adequately tame the SPEs is thinner than the number of whom can do so with multiple cores. Both are small pools, but no one appears to disagree with the fact that right or wrong the SPEs are posing a larger hurdle to to existing programmer talent pool. They may be lazy, unexcited, or just plain bad at their job. But it is these very people companies are relying on to create these games. While tool quality will improve, "good practices" will be shared, and general understanding of the system--what works well, what doesn't--will be gained through hands on experience, these very same forces will be met with further deadlines, increases complexity--and more importantly, high expectations and standards. If IW released CoD2 in 2007 it would be considered a poor game. 360 and PS3 devs cannot shoot for what was good in 2005, but must be on the cutting edge in 2008. I am not a prophet, but I think there is a middle ground--some things will become much easier and common "game breakers" can be avoided, but there will also remain a lot of reinvention of the wheel and struggling to get the right algorhythm in place to efficiently produce the results you require. And the cynic says: SMM made a comment about some dev houses hating MS. There are also individuals who hate Cell and would like to see it die. If the marketshare doesn't grow, sometimes you can create your own self fulfilling prophecy.

So trying to say that "PS3 ports will never get any better" is a flawed statement & not the alternative..

I think it is very likely that many PS3 ports in 2008 will still be suffering. More will ascend closer to parity, and in certain situations (like nAo's case) even exceed, but there are enough market factors moving right now that there will continue to be poor ports on the market.
 
The *will* get used to the platform, but it will never be just as easy as developing for a SMP machine (which, in turn, will never be as easy as developing for a single-threaded machine). So the reasoning "next year we'll get better ports because developers will get used" is flawed - unless, as you say, you fire your Xbox 360 developers and your 360 version doesn't get any more complicated.

Good point!
 
The *porting* itself will be easier, but the initial development for the asymmetric CPU of the PS3 will be still significantly harder than if the lead version was on a SMP machine (PC/360).
The interesting question is will the total amount of work be more or less depending on which platform you start out with and will the release times of the two versions be more predictable?

The *will* get used to the platform, but it will never be just as easy as developing for a SMP machine (which, in turn, will never be as easy as developing for a single-threaded machine). So the reasoning "next year we'll get better ports because developers will get used" is flawed - unless, as you say, you fire your Xbox 360 developers and your 360 version doesn't get any more complicated.
Actually I have a very hard time understanding your logic. SMP will always be harder than single-threaded development, but does that imply that developers will never "get used to it" and improve the quality of their work?

In the same way if you once learned how to organise your data to work well for streaming CPUs, don´t you think it will be easier the next time especially if you realise that those data structures perform better on the 360 as well, and perhaps make such data structures standard on your lead platform whichever it would be?

BTW, I believe the incentive of publishers do deliver decent PS3 games on time will increase as the installed base of the PS3 is getting larger. I can see that as long as the installed base of the 360 was 4-5 times larger than the PS3, it made perfect sense to rush the 360 version to market to get quick return on your investment and leaving the PS3 owners waiting, but the installed base of PS3s are now reaching numbers that the publishers start to care more about, it may very well reach 8 million units before the end of the year.
 
i think he is saying if your developing for the 360 and porting to the PS3, that PS3 ports relative to their 360 counterparts could still quite easily be in the same position as they are now because devs will also be getting more out of the 360.
So the PS3 devs are going to have to get better performance gains over a given time period then the 360 guys to make up the ground.
 
What assen is saying is that compared to the Xbox360, the PS3 paradigm requires more effort than just throwing unoptimized code at the hardware and letting it bear the brunt.

As for what itsmydamnation wrote, writing good code is a win on both consoles and both should continue to improve; however on PS3 the performance benefits are more pronounced since it's less tolerant of subopmtimal code than the 360. Developers have already stated that most optimizations for the PS3 will increase performance on the 360, but cases where you see an improvement on the PS3 side such that it surpasses performance on the 360 are not unheard of.
 
Isn't the current situation actually based on the different treatment of the graphics pipeline rather than the difference in CPU (power)? Most of the time differences can be found in different textures, AA levels, etc. that are totally independant of the CPUs.

I'm pretty sure developing an algorithm that fits CELL´s needs and organizing data structures for SPE usage is not much harder than to developing for a SMP (inorder!) system that sufferes severe cache issues when not treated carefully enough and has to fight latency much more than the other architecture (because of the memory system).

It's not like you can throw unoptimized code at each of the systems and get good results out of it automatically. And some seem to forget that developers had a little more time getting used to 360 (as it's out a year longer on the market ;) ).
 
Last edited by a moderator:
I'm pretty sure developing an algorithm that fits CELLs needs and organizing data structures for SPE usage is not much harder than to developing for a SMP (inorder!) system that sufferes severe cache issues when not treated carefully enough and has to fight latency much more than the other architecture (because of the memory system).
Sure but if you first have gone through the hassle of getting an algorithm working on the 360 that splatters related data all over the heap, you will go through some pain to rewrite and debug that piece of code to run efficiently on the SPUs later on.

But you are probably right that so far there are still a lot of unoptimised code running on the 360 and that makes PS3 ports harder, as it has less general purpose cores which tend to handle that kind of code better.

Regarding the GPUs they seem in broad terms to have similar capabilities from what can be read on this forum, but the RSX seem to require the developers to take advantage of both memory pools to run efficiently which probably has been neglected in many ports so far.

I share SedentaryJourneys sentiment that writing good code is probably a win on both platforms, the game devs can of course have different opinions on what good code is, some may prioritise efficient data structures while some may put more emphasis on high abstraction layers. Hopefully they will be combined in a nice way as the platforms mature.
 
Assuming there is no evolution and complication in the code, then that is correct. But games in 2009 will be more complex and demanding the system "do more with less".
No.. They'll demand the system to "do more with the same"...

Also generalising that all developers will be looking to push the envelope with respect to technical performance is a bit of a stretch.. I know many developers who'd be happy to stick with the same second generation engine, concetrating the majority of futher resources into fleshing out tools & optimising the engine & production pipeline to reduce overall production time/costs.. This will be the most financially "smart" direction to take for thrid parties looking to develop multiplatform games since they'll never have the resources to compete against exclusive first party teams in trying to push for the greatest visual "wow" factor on any single platform.. I'm sure most would opt to make further visual improvements following advances in the team's artistic experience and direction over trying to push the hardware to new limits.. Sure you'll have exceptions like when a designer says "hey! lets have a scene where we have the player fight 1000 opponents on screen".. If the engine doesn't support this then they'd probably push for it provided it was feasible within budget however these instances would be dealt with as a case by case & not having the teams invest any more into pushing for perf gains unless it was truely "necessary"...

Once you have a codebase & tool suite as big as many of these next gen teams efforts thus far, it becomes extremely unlikely (from a financial perspective) that a company would want tro scrap it all and start again for the sake of an additional 20% more pixel + poly pushing power..

It all depends. We are seeing some of both right now -- some really bad ports as well as some really good "ports" (co-development).

It all depends on the studios. Are they picking designs that work well on both systems? Do their techniques translate well to the SPE architecture? Are they trying to port to the PS3 or is it true co-development? Are they giving the PS3 adequate resources so they can adequately design systems that make efficient and, minimally, equal end results (performance and quality) as the 360 effort?
Exactly.. & this is exactly why you can't generalise and say "PS3 ports will always be worse" because there are exceptions to the rule even now meaning that it just isn't true across the board..

Marketshare and development costs are going to be major factors--as well as the continued quality improvements in the tools Sony and their partners make available. While the cool physics system you developed in your 2006 may carry over in large parts, your player animation systems was spartan in your FPS and now that you have a 3rd person game with intricate animation demands you will need to not only create a more robust system to deal with the complexity, but also find smart ways to map this adequately to the PS3 system.
This should be too difficult since animation is in particular an area that the SPUs should be greatly suited to and as a result I wouldn't see why your animation system couldn't run as well if not better on both platforms.. The secret is not to write it for the 360 first and port it over afterwards, but to design the high level specification of the system acording to a list of defined requirements (inputs & outputs) & then allowing your engineers on both sides to use their experience and expertise with the hardware to build the system from the ground up (or on top of the current system in place) whilst leveraging the hardware peculiarities from the ground up.. Again in this way you're never going to have to worry about trying to port a software system to a hardware platform it wasn't designed to run well and you can have the benefit of using the same animation system on the surface which utilises two completely separate implementations underneath..

One of the hurdles is that the number of people who can adequately tame the SPEs is thinner than the number of whom can do so with multiple cores. Both are small pools, but no one appears to disagree with the fact that right or wrong the SPEs are posing a larger hurdle to to existing programmer talent pool.
The thing is though that all these pools are growing with time & it's alot easier to have seniors educate juniors on how to code well on a platform than it is to have externals coming in and trying to educate your seniors AND juniors whilst your in production of a title..

If IW released CoD2 in 2007 it would be considered a poor game. 360 and PS3 devs cannot shoot for what was good in 2005, but must be on the cutting edge in 2008. I am not a prophet, but I think there is a middle ground--some things will become much easier and common "game breakers" can be avoided, but there will also remain a lot of reinvention of the wheel and struggling to get the right algorhythm in place to efficiently produce the results you require.
The solution?

Change your requirements..

It's alot easier to meet them if your expectations of what you can and cannot do with your engine is already known (having the engine already established after shipping games on it..) In the end consumers will care alot more about advances in gameplay, story telling and production values than they will about graphics going forward & i'm pretty sure that if Mass Effect 3 (the third in the trilogy) presents the exact same level of graphics fidelity in 2010, albeit with much much greater movie-like production values, small, subtle additions but strong artistic deviations, gamer's wouldn't be calling the game poor by any means..

And the cynic says: SMM made a comment about some dev houses hating MS. There are also individuals who hate Cell and would like to see it die. If the marketshare doesn't grow, sometimes you can create your own self fulfilling prophecy.
With all due respect, the quality of bad PS3 ports is hardly going to dictate the market share of the platform so I fail to see how this is relevant?

I think it is very likely that many PS3 ports in 2008 will still be suffering. More will ascend closer to parity, and in certain situations (like nAo's case) even exceed, but there are enough market factors moving right now that there will continue to be poor ports on the market.
Sure I agree with that.. All platforms have had bad ports of games from other platforms since the days of the SNES.. The point however is that it's not something that we'll see more of as a general case especially with regards to technical performance in the realm of framerate & IQ.. The biggest differentiating factors will be in area where one platform displays greater leverage of it's strengths over the other (e.g. PS3 version having less grass in a scene or less particles or less online features or more physically interactive objects in the world..)
 
Is it way out to suggest that at some point it´s possible that the PS3 "ports" will start to pull ahead of the 360 versions?
 
It's possible, but I doubt it will happen (unless every PS3 exclusive that's made ends up looking considerably better than multiplatform games)

Thats a fairly blanket statement I guess, of course if they do find an advantage, some developers would make the PS3 version better, the "big ones" with their yearly updates won't bother though.
 
Is it way out to suggest that at some point it´s possible that the PS3 "ports" will start to pull ahead of the 360 versions?

It´s probably more likely you will see some PS3 versions pull ahead of 360 ports. IIRC Midway and Sega has announced they will use the PS3 as their lead platform. Maybe more will follow later on if it works out well for them.
 
No.. They'll demand the system to "do more with the same"...

With the same hardware, yes. I didn't convey this well, but the "do more with less" was in terms of game complexity and what they are doing now with X% of resources they will be asked to do the same (or more!) with less resources to make way for new features. As games become more complex there will be the expecation of a) more/better at the current designs/techniques and b) addition of new ones.

So lets say you are doing a racer and have a fairly robust physics engine for car movement and you have this running on 4 SPEs. This is version 1. In version 2 you improve the physics engine to correct some issues and make it a little more robust--and toss in 50% more cars. But you also now want to do some realtime deformation. You still have your 4 SPEs for this budget, but now you are going to "dedicate" one SPE to the realtime deformation. So now you have 3 SPEs to "dedicate" to your car physics engine, but with a 50% higher workload due to new features and improvements. (There won't typically be "dedicated" SPEs but more likely a job/task model).

That is how I arrived at "do more, with less". In this example "do more/better physics, but with less resource budget--because some of that budget is being thrown toward the increasing complexity of the new version".

I am out of time, so I have to end here. But that is all I ment by "do more with less". Not that there was less console there, only that as games mature they will need to find ways (more optimizations, better approached, etc) to make way for their feature bloat.
 
With the same hardware, yes. I didn't convey this well, but the "do more with less" was in terms of game complexity and what they are doing now with X% of resources they will be asked to do the same (or more!) with less resources to make way for new features. As games become more complex there will be the expecation of a) more/better at the current designs/techniques and b) addition of new ones.

So lets say you are doing a racer and have a fairly robust physics engine for car movement and you have this running on 4 SPEs. This is version 1. In version 2 you improve the physics engine to correct some issues and make it a little more robust--and toss in 50% more cars. But you also now want to do some realtime deformation. You still have your 4 SPEs for this budget, but now you are going to "dedicate" one SPE to the realtime deformation. So now you have 3 SPEs to "dedicate" to your car physics engine, but with a 50% higher workload due to new features and improvements. (There won't typically be "dedicated" SPEs but more likely a job/task model).

That is how I arrived at "do more, with less". In this example "do more/better physics, but with less resource budget--because some of that budget is being thrown toward the increasing complexity of the new version".

I am out of time, so I have to end here. But that is all I ment by "do more with less". Not that there was less console there, only that as games mature they will need to find ways (more optimizations, better approached, etc) to make way for their feature bloat.

Their are (practical) ways to get around that particular problem you suggested as an example however I don't want to get into that..

Otherwise those are pretty valid points..
 
"We never wanted to release PES 2008 in this state on PS3," said Seabass. "In fact, we thought about delaying it but different factors meant we had to [release it]."

The PlayStation 3 version of PES 2008 is plagued with problems, including horrific slow-down and game-breaking slowdown during online matches - problems Seabass puts down to being overstretched by multi-platform development.

...

Despite problems and a general feeling of disappointment with the PS3 game, Seabass is confident that future iterations will be vastly improved.

"Look at the way the PS2 version panned out from the first to the last game. We're well on course to follow suit on PS3," he said. "PES 2009 is in development, but after hearing the criticism, we'll be taking it back to the drawing board. It might not be next year, but it will be soon."

http://www.videogamer.com/news/23-11-2007-6958.html
 
I agree with this.. However only recently they've begun to show promise in understanding there weaknesses & taking steps to rectify them..

The purchase of TheCreativeAssembly & the "Comdemned" guys show that they CAN in fact make good business decisions & hopefully they still have a bright future in the industry..

Now back to multi-platform development.. :D

Didn't Warner buy Monolith?
 
PS3 development makes for better 360 game

http://www.psu.com/PS3-development-makes-for-better-360-game-News--a0002052-p0.php

Nick Channon, Senior Producer for Burnout Paradise explains how setting the PlayStation 3 as the lead platform over the Xbox 360, actually makes the Xbox 360 version a better game.

"From our point of view, we’ve had no issues with dealing with either machine; that’s what I mean by that. It seems that some games haven’t always transitioned between the two well. From our perspective leading on Playstation3 has meant Xbox 360 has given us a great product."
 
As they themselves noted, they were big fans of the PS2 and very excited about the PS3. I am interested to see how the 360 version fairs and how resources were shared, what challenged faced, concessions made, etc. Anyhow, funny source "you do the math" i.e. PS3 is better.
 
Well, the guy who wrote the article itself is drawing some really weird conclusions from the statement as well as from the relationship between PS3 and 360. I mean, a lot of people would also say that putting the PS3 as the "lead" platform for a number of things pays off on the 360, but certainly not for the reasons this guy seems to think. Moreover, as some people have said earlier in the thread the question for many people is how much it's worth. If they already have pretty good results on the 360, is it a big enough difference to warrant the extra work? For some of us, it is definitely worth it, and for others, not so much. For most, I think, they see it as a necessary evil just to get moving on the PS3 whether the additional payoff on the 360 was all that great or not... and that's a source of a lot of screaming.
 
Status
Not open for further replies.
Back
Top