New 360 Pic

[WishMaster3K]

Re: New 360 Pic

Be warned, this may be a long post, I know it took me an hour to type out with all my editing and such. Enjoy!

I like how manatra read the argument like all of us did and responded with:

Doesn't matter how powerful the PS3 is with it's multi-core architecture. The fact is, developers don't have the tools to utilize it properly. The same can be said for the Xbox360. . . .This is the main reason why the PS3 and Xbox360 aren't going to be very big jumps over the PC in terms of graphics.

Kinda like a cop out. But in any case, keep in mind that PCs would not be where they are at now if it wasnt for the foresight of engineers to create a standard every 5 years. Looking at PCs NOW, they are talking about incorporating a third card into programming, and that's a physics card. Did PC games even need to *think* about something like a physics card even 8 months ago? Even more so, the PS3 is readily able to adapt. Forgive me for not being able to provide a link, I will do so asap.

Last generation we threw bits out the window. This generation we'll throw CPU speed out the window. Pretty much everyone and their mom can make a CPU at 4ghz, but it's what the CPU will be able to do in terms of direction that will define true power. As it seems, simply taking CPU speed is two-dimensional, whilst the PS3 and XB360 are moving into three-dimensional CPU computing directions.

FURTHERMORE, it's also obvious that simply using "Hyper Threading" is a moot process, simply a faux pau. Likewise is using two GPUs, simply because one is coded to create the left side of the screen and one is programed to show the right. PCs do some nice things and get their supporters to feel that a lot of hardware is being put to work, but quite frankly, the new systems will run circles around current and future PCs in terms of scope and bredth of gameplay.

On the PS3, I suspect games will take on entirely new dimensions, with each individual enemy being controlled by independant and ADAPTIVE AI. Imagine a football or fighting game where your opponent adapts to your play style and counters it, forcing you to lose your linear format of playing.

Or imagine MGS4, if you will: Each enemy following their own predicable path, but not simply going back to their predescriped paths. Isnt it odd that simply hiding in a locker would deter guards? With the ability to program deeper AI and stronger language, games are new beasts. "No where to hide" makes sense now, because enemies wont simply "Give up" on looking for you now.

The argument that "they'll be harder to program for" is such a weak one, because in LESS than one year we'll already be seeing brilliance on these consoles. A lot of money is to be made, simply because programmers have to upgrade their own styles in order to make money. It's no use to BITCH about technology advancing, because that's the nature. I recently got Cisco Certified in June, and I can guarantee you that by January when new protocols come out, i'll have to re-take classes.

Same concept with consoles. In 3 years, the best thing you'll have on your resume isnt "I can do Open GL, Direct X, etc etc," it'll be "I programmed (Game Here) on Xbox360 and/or PS3. I understand how to properly thread a game through multiple cores and etc etc."

A developer is only as good as the material they produce on what is given to them, and seeing as how consoles decided to stop being LitePC without OS and compatability issues, instead opting to push developers to make greatness, games limited to the PCs will only suffer.

Even if you throw in $5000 worth of equipment into a Liquid-Nitrogen cooled Alienware system from Nasa, owners of PS3's and 360's will have something more than "Games that look really, really, really cool."

[Rai-Jin]

Re: New 360 Pic

lol my graphics on PC look waaaay better then that >.<

[fuz]

Re: New 360 Pic

Now this brings us to another interesting problems, the Nividia GPU programmable pixel shader... This is both good and bad... it's good that you can do some amazing things, but at the cost of more customized coding. Well there should be tools to speed that, but it's more work in the end obviously. I doubt this is a requirement, but it's another thing to think about in the ps3 development cycle.

And finally a tid-bid from the PS3 news trinkling out. It seems that PS3 is even MORE powerful then the demos we saw(assuming they're rendered rt) The 1st development boxes sent out(which the demos were made in) had no GPU, and had a lower clockspeed Cell CPU( at about 75%). This means that whatever strangeous conditions those demos were made in, should be easier on the final version.

Er, isn't it ATi with the progamable pixel shader?

Regardless of the cpu, the majority of the work will still be on the GPU. The cpu's just a multi-core DSP in the most basic sense.

You really don't need that many individual processing units to have good game AI. It's incredibly stupid to build a whole new processor to support more AIs, and you don't need too. You just need less precessing penalty when dealing with it, 2 cores is already enough to run current AIs. You really need much more memory to support such a complex program than more cores, since AI can't learn if it can't remember.

I dunno if anyone has played the F.E.A.R. demo, but the AI in that game is quite possibly best orchistrated show of team effort against the player. It's unusually agressive, works in all (random) locations, and literally has every npc gang up on you from all sides, as if they are using the radio to communicate. IT shows coordinated balance of offense and defense, pinning you down with one, while another sneaks up on you. I don't even think that's an evolving AI, just a program with a static list of pre-defined instructions, but it's damn hard to play against despite being 'weaker' in terms of game stats.

Making cinematic images isn't hard, but making a game like that is. It is incredibly time consuming, and hence expensive to go after such production values. Think of it like a rdm post refresh, haste, and staffs; now that you have more tools to do better, your job is just that much harder, not easier. Games like that have more artists than programers to fill all that content needed. Eye-candy that can be quickly programed in, say procedural textures, dynamic smoke,and lighting don't cost much to put in, but will only take you so far if the design is sloppy.

[WishMaster3K]

Re: New 360 Pic

Nice usage of RDMs, but being one, on the flip side, it's only initially hard to do everything needed once we get our mid-game abilities. True, with convert, refresh, haste and the addage of staves into our casting cycle, things are more complicated, but hasnt it been proven that we can easily adapt to the heavier workload?

[kuu]

Re: New 360 Pic

Er, isn't it ATi with the progamable pixel shader?

Regardless of the cpu, the majority of the work will still be on the GPU. The cpu's just a multi-core DSP in the most basic sense.

You really don't need that many individual processing units to have good game AI. It's incredibly stupid to build a whole new processor to support more AIs, and you don't need too. You just need less precessing penalty when dealing with it, 2 cores is already enough to run current AIs. You really need much more memory to support such a complex program than more cores, since AI can't learn if it can't remember.

I dunno if anyone has played the F.E.A.R. demo, but the AI in that game is quite possibly best orchistrated show of team effort against the player. It's unusually agressive, works in all (random) locations, and literally has every npc gang up on you from all sides, as if they are using the radio to communicate. IT shows coordinated balance of offense and defense, pinning you down with one, while another sneaks up on you. I don't even think that's an evolving AI, just a program with a static list of pre-defined instructions, but it's damn hard to play against despite being 'weaker' in terms of game stats.

Making cinematic images isn't hard, but making a game like that is. It is incredibly time consuming, and hence expensive to go after such production values. Think of it like a rdm post refresh, haste, and staffs; now that you have more tools to do better, your job is just that much harder, not easier. Games like that have more artists than programers to fill all that content needed. Eye-candy that can be quickly programed in, say procedural textures, dynamic smoke,and lighting don't cost much to put in, but will only take you so far if the design is sloppy.

ATI and 360 is doing that shared memory thing I believe.

And it's always a bad thing to say, "they don't need it" because it's backward thinking. Even top pro's thought they didn't need 2 meg of ram either back in the days

Writing multi threads is a interesting thing, it's largely unstepped in the gaming world. Only a hand full of people who played with Sega Saturn and perhaps some mac people, but doing "something" isn't hard. It's when you're trying to do it everywhere it will be hard. Some threads can be made to be run independing no problem, i've explained it somewhat in another post. Another thing to to consider after talking to people who know this stuff is to write "short lived" threads, or parallel in small bits. Basically you write you code in such a way, that it won't keep going too far so you don't trip yourself in the long run. Well only a theory at this point without talking to actual developers.

Your rdm analogy is decent but flawed. Yes you are juggling stuff, but programing isn't all manual. No programer in his right might would not reuse code if it applies. And there is middleware to buy like the UT engine to insert into your own. Also you reuse your own tools obviously for each game. Some will also hire out independent consultants, etc. It's not each game you start from scratch.( which I guess answers Wishmaster3k's question= yes it does get easier the more you do it)

Cinematic isn't as expensive as you think. There must be a give and take. They can be outsourced, reused, modified, and even done in advance and depending on what you do save on programing. They are also simpler to program for. Examples I can think of are Nipponichi games. Those games are a mass of reused hand drawns, but due to how the game is presented, it's relatively cheaper to just do it that way. Then you have inhouse powerhouses like Squaresoft which can punch of stuff easily with their teams.

[Macht]

Re: New 360 Pic

Nice usage of RDMs, but being one, on the flip side, it's only initially hard to do everything needed once we get our mid-game abilities. True, with convert, refresh, haste and the addage of staves into our casting cycle, things are more complicated, but hasnt it been proven that we can easily adapt to the heavier workload?

Seems a little of topic, but ok....

Good points were made about the processors and programing for them. However some things should be pointed out that cell processor does that I feel was misunderstood here.

First of programming the PS3 and performance, the programmers have taken some decidedly single threaded game code and ran it on pre-preproduction hardware with a processor optimised for multi-threading and stream processing (most likely with an immature compiler). Result was the performance wasn’t what they expected is not so much news as blindingly obvious.

Getting the full potential from a Cell will be more difficult than programming a single threaded PC application. Multiple execution threads will be necessary as will careful choice of algorithms and data flow control.

This is not exactly new problems and solutions have long existed to them. Multiprocessor or even uni-processor servers have been doing this sort for thing for years. It’s an old hat to BeOS programmers and many other programmers for that fact. You will likely see technologies appearing from other areas which take the pain out of thread management.

Also be aware that the SDK for the PS3 is much different then it was for the PS2. There are various systems that are in development for controlling execution on the Cell so developers should have plenty of options, this compares well with PS2 development which was primarily done in assembly and was highly restrictive.

Much of the PS3 is not only catering to performance but it's also catering to software development. Most of the programmers you see complaining are just babies wining afraid that they'll have a harder time creating their stuff when in many ways it'll actually be easier, in other ways it will be harder though but again many solutions to the harder stuff has already been out and done on many other systems for many years.

Aside from that seems it's still being missed that cell processor is also capable of just being a scalar processor but enhanced. As for that guy's interview if you look he's slamming on All next-gen developments. The interviewer decided he'd just focus a big part against PS3, another report that's more bias view on it.

Double Post Edited:
On top of that the interview seems like a shameless method of promoting a crappy technology design such as Valve Software's Steam Technology, which is pretty crappy in comparison to Digi Stream (Think that's what it's called).

I also forgot to mention that this Newell guy just gave away that he has little knowledge in Vector Programming (Which probably more of the main reason he's complaining) when he stated.

There are incredibly few programmers who can safely write code in the PlayStation 3 environment. And I totally see why Sony wants people to write code that runs on seven SPEs and a central processing unit, because that code is never going to run well anywhere else

The actually vectorizing part of the programming (Which takes advantage of the SPEs) happens in the compiler, so if it isn't compiled that way it's going to work fine were ever else. Just that it would work normal like anything else and not take full advantage of the speed potential that the SPEs when using a vectorized program and compiled as such could do for it.

[kuu]

Re: New 360 Pic

This is not exactly new problems and solutions have long existed to them. Multiprocessor or even uni-processor servers have been doing this sort for thing for years. It’s an old hat to BeOS programmers and many other programmers for that fact. You will likely see technologies appearing from other areas which take the pain out of thread management.

Not quite true. On the surface, it may look like it, but if you shift through various blogs and interveiws, writing for farms and massive database like calculations, is NOT like writing for a multi core game.

Lots of technicalities, but it bascially boils down to that games require more "creativity" as they work in a way that not many people who just program on farms or super computers do(straight calculations as oppose to whatever juggling a certain genre of game might do). BeOS might be decently close, but most of it is done in the low level kernel systems, not much upper layer programs. This does not mean that not having those expertise's is worthless, knowing the theories behind something always makes things easier to learn and do.

But anyway, besides the saturn people, many programers will or have been getting their feet wet for the future( multi core is the near future for us it seems)

Also be aware that the SDK for the PS3 is much different then it was for the PS2. There are various systems that are in development for controlling execution on the Cell so developers should have plenty of options, this compares well with PS2 development which was primarily done in assembly and was highly restrictive.

Much of the PS3 is not only catering to performance but it's also catering to software development. Most of the programmers you see complaining are just babies wining afraid that they'll have a harder time creating their stuff when in many ways it'll actually be easier, in other ways it will be harder though but again many solutions to the harder stuff has already been out and done on many other systems for many years.

We don't know much about the SDK yet, but with this generation it's almost certain it will not JUST be for games, but for programs as well, as Sony wants this to be a home multimedia center(re). It also helps that IBM is in it, with their expertises can give us some nice code snippets and tools.(See the satalite demo ^^)

Double Post Edited:
On top of that the interview seems like a shameless method of promoting a crappy technology design such as Valve Software's Steam Technology, which is pretty crappy in comparison to Digi Stream (Think that's what it's called).

I also forgot to mention that this Newell guy just gave away that he has little knowledge in Vector Programming (Which probably more of the main reason he's complaining) when he stated.

The actually vectorizing part of the programming (Which takes advantage of the SPEs) happens in the compiler, so if it isn't compiled that way it's going to work fine were ever else. Just that it would work normal like anything else and not take full advantage of the speed potential that the SPEs when using a vectorized program and compiled as such could do for it.

OMG...that guy doesn't know vector processes?? Wtf is he a developer or a code monkey? He's in what field again?

But whinning definitely. Sure there is cause to whine, it's not getting easier, but then money makes pain go away in these cases. And it's a growing market out there, with lots of competition.

[fuz]

Re: New 360 Pic

For multi-core vs single, it's not about giving them more than what they want. They gave multi-core at a penalty to single thread performance. There is a fixed amount of die real estate, and instead of making 1-2 cores, they went with a lot of small ones. This isn’t like RAM, or memory in general, it’s about the more effective computing setup given a similar number of transistors.

I don't know if this is the future of computing, but do I think they did it just for the sake of doing it.

While this offers flexibility as you can now define a core to do a certain task, e.g. one for 3d sound, another for physics, another to help calculate lighting, etc, when load balancing, it's much trickier, since not all these processes take the same amount of power. So if it is not utilized well, it can be extremely inefficient. The tech demos were overblown, since they were static which makes it easy to balance. The overhead and complexity of dynamic processing in games is going to be a whole lot more difficult to allocate proper resources for, on the fly.

We already know that dual core processors aren't twice as fast as a single (in most tests), the 7 cores on the PS3 aren't going to be OMFGBBQ faster than the 360's due to the same overhead penalty in allocating resources to feed all those processing units the correct amount of data to keep things as efficient as possible. If the low bandwidth rumor between the SPEs is true, then load balancing to keep unnecessary communication to a minimum will make it that much harder to extract high performance.

Then again as I said, it might not even matter as the CPU in PC games plays a very minor role compared to the GPU. We don’t know which will be the bottleneck.

While my rdm analogy isn't perfect (jee, why don't you find one?) it does highlight that more work must be done within the same budget, be it MP, $, or time . I don’t mean it just for programming, but content as well. Rdm also build upon recycled abilities, but then you add more on top of that as things evolve. It's quite easy to tell a good rdm from a bad one when the workload scales up, more spells, more equips, PT members gain new abilities, mobs gain new abilities, etc. The majority of game programming now isn’t the result of a small company with a few people, but million dollar budgets with several departments, and normally a major publisher at the core investing the money.

At high resolution, bad work is even more apparent than ever. I’m not sure, but I can guess pretty well that most games made just barely sustain the bottom line if at all, while a few select ones will generate money. Calling it easy to make games that will run a business is almost insulting. The programming for these new consoles will be done somehow--I’ve little doubt of that, but content is now key. Programming can be outsourced, done by anyone. Creativity with the power of the new platforms will be more difficult to extract. In a simple example, you can clearly see that the designs for FF10 armor are much harder to create than FF7. Similarly the move from 2d to 3d fighting games increased the difficulty of making one to a whole new level.

(ok, I wrote too much...)

[fastcart]

Re: New 360 Pic

Capt. Kirk: Too...much...nerd talk... Head...will...explode... ^^

lol. Sorry. This thread has gotten really off topic. I just felt like beig random w/o adding anything. d:

[fuz]

Re: New 360 Pic

You bumped it >.>

[Kailea]

Re: New 360 Pic

forget all that techno crap, and list to my plain and simple explination......

X-box360 version will be the same thing as teh PS2 and PC versions, only diffrance is the HD suport. The new expansion will be release alittle after the release of the X-box360

The three new systems, PS3, X-Box360, and Revolution are going to preform the same, and have basicly the same capabilities. The gaming industry has reached that point in where there really is no diffrance in the systems, but just the games.

[kuu]

Re: New 360 Pic

For multi-core vs single, it's not about giving them more than what they want. They gave multi-core at a penalty to single thread performance. There is a fixed amount of die real estate, and instead of making 1-2 cores, they went with a lot of small ones. This isn’t like RAM, or memory in general, it’s about the more effective computing setup given a similar number of transistors.

I don't know if this is the future of computing, but do I think they did it just for the sake of doing it.

Not true again. You're getting to the technical side of Cell. Cell has a lot of "Stuff" that make this a minor issue. Sure it's a an issue, nothing beats not having to do the work, but the cell and 360 will have many things to address this. The Cell has some even more interesting tibbit to address this is that there is an front end(one of the spe?) that is like a manager of all the other spes that keeps everything running nicely.

Suffice to say, you're 5 yrs too late in your arguement, it was problematic when for pioneers like transmeta(they crashed and burned).

And just to further elaborate, you got the whole case backwards as a whole. 2 on a single core is BETTER then 2 chips apart. The closer they are, the better they are to work with(not to mention saving bunches of material and heatsink). While sharing branch prediction/cache/etc have been addressed long ago said above.

At high resolution, bad work is even more apparent than ever. I’m not sure, but I can guess pretty well that most games made just barely sustain the bottom line if at all, while a few select ones will generate money. Calling it easy to make games that will run a business is almost insulting. The programming for these new consoles will be done somehow--I’ve little doubt of that, but content is now key. Programming can be outsourced, done by anyone. Creativity with the power of the new platforms will be more difficult to extract. In a simple example, you can clearly see that the designs for FF10 armor are much harder to create than FF7. Similarly the move from 2d to 3d fighting games increased the difficulty of making one to a whole new level.

Completely wrong... the more pixels the better. You aren't much of a graphic person are you? BLOWING UP(otherwise known as scaling. to make a 1" into 4") a lack of pixels makes it look worse, because you can't make something from nothing. This is not the same as adding more pixels. a 1" will always be a 1". You can trick the human eye by using filters like "blur" and advance stuff.
BUT
having more is always better. Especially when when you add filters to it. Basically the more pixels a person OR program has to work with, the better. example -> ffxi in super high rez.

Take a couple of photoshop lessons first if you want to complain about that .

The three new systems, PS3, X-Box360, and Revolution are going to preform the same, and have basicly the same capabilities. The gaming industry has reached that point in where there really is no diffrance in the systems, but just the games.

Not true. They all have different limitations, ideals, and technologies. The rev looks like the most underpowered of all, and of course lack of HDD.

You're falling into the trap like Fuz did about "not needing". The future always look vast till you reach it. Who need 64bit address space ha! no one needs more of ram then that. Well we'll need it very soon.

[Kailea]

Re: New 360 Pic

Not true. They all have different limitations, ideals, and technologies. The rev looks like the most underpowered of all, and of course lack of HDD.

acualy the Revolution does have an HDD, it is how they are doing the whole NES SNES and N64 emulation, and it is supost to be capable of doing this out of the box, but you know how that can be :p

and never judge a system by its info, look at what Capcom did with RE4....they acualy have to dumb it down for the PS2 version that is coming out :p

[pearlsea]

Re: New 360 Pic

^so ffxi is posible on the revolution?

[kuu]

Re: New 360 Pic

acualy the Revolution does have an HDD, it is how they are doing the whole NES SNES and N64 emulation, and it is supost to be capable of doing this out of the box, but you know how that can be :p

and never judge a system by its info, look at what Capcom did with RE4....they acualy have to dumb it down for the PS2 version that is coming out :p

Spec as scarce for the Rev as nintendo is keeping hush hush. But by all articles of hints, it wil not have a HDD.

NES and SNES games can be played on a memory card. With this generations memory can be so cheap, expecially if you use some mainstream type like CF or SM, that it's more then enough. A NES is about 500k at most, and SNES is about 3megs. a CF 512 can easily do it.

And finally Rev is suppose to be the size of 4 dvd cases...I don't know about you, but those dvd cases better be big enough to have to hold a HDD AND the system AND the drive bay. Unless they are using insanely expensive HDDs(Ipod mini level, if there is enough enough supplies of it with apple buying it all up)

The 2nd option is some kind of external HDD like PS2slim. And that shall we say is going to be problematic. It's nowhere near IDE level....though it should be enough for emulation games, but not FFxi or any mmo I think.