PlayStation III possible details...

Status
Not open for further replies.
Its not a problem at all. It makes me think i should go and edit the post I did in another section though. they might think I'm fully serious.
 
mech said:
What exactly are these holes?

Lithography advance and transistor budget allowances.

You can't be any more right than I am, because the thing's not even finished yet.

Every console which has had multiple CPUs to program for has been deemed a "programming nightmare". Saturn, PS2, Jaguar, all these machines were very difficult to program for because of the multiple cores that required managing.

Um, when you make comparsons such as this, I'm inclined to think that I am *more* right.

Instead of trying to win a point against me with an off-hand remark like "go do some research", why not try and explain

Ok, but this has been covered many a times and I felt that going over it yet again is redundant - which is why having you go do some background reading is logical.

Why you think multiple CPU setups are going to be the way of the future?

Because, unlike historical solutions, which are strictly CMP or SMT, cellular computing has many advantages over the conventional approach.

-They're high effecient in silicon usage. The design is built around a core with the bare minimum instruction set and features, yet yeilds (as shown in IBM studies) upwards of 70% the preformance of a conventional solution (von Neumann) at a fraction of the die space/tranistor count.

-The cores are a SoC design that included a pluratly of embedded memory that allows the core to function and sustain preformance near it's theoretical peak. It could fundimentally change the idea of a cache hierarchy and access speeds.

Basically, it's more effecient in it's silicon/preformance ratio, can achieve much higher preformance threw concurrency, and it the future in the broadband enabled world where processing is secondard to the movement of data.

I still think there are too many problems with this setup, especially the problem of parallelizing many game tasks - several of which are really quite difficult to parallelize effectively. So how are they going to overcome this problem?

Dave Barron once said something that's quite relevent. It basically went, it doesn't matter how the hardware guys get something done aslong as it works right.

I obviously don't have an exact answer, but It appears (big surprise) that your not the only person to think of this:

Knowing this, the three chip partners have so far set a goal of crafting Cell as a system, creating operating system and application software alongside Cell hardware.

Aswell, as a dev here (forgot who) said that SCE's R&D has been working on something along the same lines.

With the advent and widespread adoption in the recent past of HL languages throught the entire pipeline masking all low-level or architectural functions, I think the answer is buried somewhere in there. Just let the hardware guys do their job and look at the upsides of the hardware and start thinking of how it can be done, not the mindset that it's impossible.
 
Vince If you have anymore info on this or a good article can you post a link so i can catch up a little on this ? Thanks in advance
 
jvd said:
Vince If you have anymore info on this or a good article can you post a link so i can catch up a little on this ? Thanks in advance

I need to start labeling my CD's :) I had one with saved links on it, I'll look for it tomorrow, I need sleepy now...

But, here's a good place to start concerning the fundimentals of cellular computing

http://www.research.ibm.com/journal/rd/462/nair.html

NOTE: It's from 2001 and has .10um speced for 2005, but as shown above, IBM currently is aiming at 0.065um with SOI and Strained Si. So, a more do-able comparason is to look at the 2007 and 2011 plans when the lithography is comparable.

According to this projection of densities and clock speeds, and averaging between 0.07 and 0.05um, it would appear I might owe Mech a dearest appology - which if this pans out, then I most certainly will give. But realise, these projections are on a die thats huge.. have to break down the numbers tomorrow, just giving fair warning that i may be wrong.

Oh well, atleast I don't make bets like *other* people.. hehe
 
Vince said:
mech said:
What exactly are these holes?

Lithography advance and transistor budget allowances.

Que?


Vince said:
You can't be any more right than I am, because the thing's not even finished yet.

Every console which has had multiple CPUs to program for has been deemed a "programming nightmare". Saturn, PS2, Jaguar, all these machines were very difficult to program for because of the multiple cores that required managing.

Um, when you make comparsons such as this, I'm inclined to think that I am *more* right.

Sorry? What exactly are you "more right" about? And what am I making a comparison to? I made a statement about consoles with multiple CPUs on a single core (such as the PS2).


Vince said:
Instead of trying to win a point against me with an off-hand remark like "go do some research", why not try and explain

Ok, but this has been covered many a times and I felt that going over it yet again is redundant - which is why having you go do some background reading is logical.

Well I'm sorry I haven't read anything about it, but you didn't even provide a starting place to back up your main points. "Go research it" - mmmright.

Vince said:
Why you think multiple CPU setups are going to be the way of the future?

Because, unlike historical solutions, which are strictly CMP or SMT, cellular computing has many advantages over the conventional approach.

-They're high effecient in silicon usage. The design is built around a core with the bare minimum instruction set and features, yet yeilds (as shown in IBM studies) upwards of 70% the preformance of a conventional solution (von Neumann) at a fraction of the die space/tranistor count.

-The cores are a SoC design that included a pluratly of embedded memory that allows the core to function and sustain preformance near it's theoretical peak. It could fundimentally change the idea of a cache hierarchy and access speeds.

Basically, it's more effecient in it's silicon/preformance ratio, can achieve much higher preformance threw concurrency, and it the future in the broadband enabled world where processing is secondard to the movement of data.

That's all well and good for specific embedded applications, but what about issues in allowing the parallelization of game tasks?

Vince said:
I still think there are too many problems with this setup, especially the problem of parallelizing many game tasks - several of which are really quite difficult to parallelize effectively. So how are they going to overcome this problem?

Dave Barron once said something that's quite relevent. It basically went, it doesn't matter how the hardware guys get something done aslong as it works right.

I obviously don't have an exact answer, but It appears (big surprise) that your not the only person to think of this:

Knowing this, the three chip partners have so far set a goal of crafting Cell as a system, creating operating system and application software alongside Cell hardware.

Aswell, as a dev here (forgot who) said that SCE's R&D has been working on something along the same lines.

With the advent and widespread adoption in the recent past of HL languages throught the entire pipeline masking all low-level or architectural functions, I think the answer is buried somewhere in there. Just let the hardware guys do their job and look at the upsides of the hardware and start thinking of how it can be done, not the mindset that it's impossible.

Again, this all very vague and mysterious. "Let the hardware guys handle it" - what if they can't? What if there is in fact no simple way around this? Maybe that's why this method hasn't been used much in the past?

At any rate, we'll see what happens when it comes out hey? Obviously we have differing opinions on the effectiveness of parallel computing, but this is hardly something to get snooty about - we are allowed to have our own opinions.
 
what i think he was trying to say is that the main problem with both Saturn and Jaguar wasn't really the fact of having multiple cpu's...

it was because of BADLY IMPLEMENTED multiple cpu's.........

like when ps2 first came out, everyone was saying how it would have failed just because of the high parallelism needed to get some good performance out if it... and how it would have been abandoned by devs who wanted an easier work environment... :rolleyes:

of course the best on both gc and xbox are visibly better than the best on ps2, but i think some cool effects used in some ps2 games makes u wonder what nintendo and microsoft have spent their time on, when designing their consoles apart from more textures and more *conventional-typical* advances... after all they are almost 2 years more advanced technologically :LOL:

i must be the only one who thinks that Sony's approach to graphics technology and technology as a whole is the only interesting thing about this console war, and the next one(s)...

gc and xbox are more of a pc-graphics-card type of update on features, resolution and (imo) minor aspects instead of focusing on new and radical approaches....

oh well...
 
london-boy: PS2 is still a major pain in the arse to program for. People are just getting used to it now.

If anything, you hear devs praising the development environment of the Gamecube and Xbox, but you never hear someone praise the simplicity of programming the PS2. The thing's a pig to use well - the power's there, but hard to get at.
 
Though it is of course entirely speculation, and will never be proveable to any significant degree, I am of the strong opinion that the PS2's developer support is almost entirely due to its brand and percieved future market support. That this perception and resulting developer support resulted in a more or less self fulfilling prophecy as far as the success of the PS2 is concerned is another issue (more below).

The point is, just because developers work with the PS2 doesn't mean they prefer its architectural approach. The likely reasons are financially motivated, as are most business decisions. While the techies may or may not enjoy working with the platform, the pencil pushers generally call the shots with most publishers, and the PS2 was and is seen as the market leader (rightfully so, by now). Marketing, sales and management generally work with brand strength and value, not technological approaches, and this is how the PS2 was (IMO) largely evaluated for development focus.

Most of the previous examples of poorly supported multi-processor platforms in the console space were either not produced by market leaders, or had other significant negative factors in their lack of adoption. So I don't believe any conclusion can be drawn from the lack of support for previous multi-processor consoles vs. the PS2's large support base. Imagine for a moment if the PS2's exact hardware architecture was used by Sega, and the Dreamcast architecture (or an updated version, comparable with the technology of the time of the PS2's release) was used by Sony, which do you think would have been more popular and succeeded? I'd bet heavily on Sony's product, regardless of the architecture. Branding *is* that strong, unfortunately.

As far as the issues with the Jaguar and Saturn, poor hardware implementation or not, they could have been made better development platforms with good middleware and better tools, just as the PS2 could have at introduction. Only in the last year to year and a half have good tools been widely available for PS2 development. But good development environments not withstanding, the point I made about brand strength still stands, and neither the Jaguar nor the Saturn would likely have been significantly more successful with better development environments. Well, maybe the Saturn, although there were other negative factors that severely hampered its success.

Which brings us to the sad reality that developers will basically eat a bowl of excrement at work just so they can afford to eat a bowl of food when they get home. In other words they are most concerned about their percieved bottom line which, while it makes theoretical sense from a business perspective, often does not tell the whole story. And when the developer won't, their publisher will usually make the decision for them, again based on the same often irrelevant or incomplete factors.

In my opinion the DC, GC, or XB could all have been leaders with enough developer support; any one one of them could have trampled the PS2 if the developers simply refused to deal with the headaches Sony imposed with their lack of tools and support (which was the primary problem with the PS2, not the architecture per se, which *is* innovative and powerful). The developers/publishers ultimately choose the success of a platform, but since they don't make decisions with that power taken into consideration, and since they can't exercise power as any kind of unified group with cooperative decision making ability, all their power is essentially worthless in practice. Developers should realize the power they have and start leveraging it to encourage development systems, platforms and environments that they favor, rather than simply bowing to the arbitrary and caprecious whims of the uninformed masses. Ultimately the consumers will go where the games are, so if the developers can get the same or better results for cheaper or in less time on one platform vs. another, both sides are better off in the end.

Given the above then (assuming you accept my assertions) and the current success of the PS2, it seems likely that, no matter what direction Sony goes with the PS3, they will succeed. An extremely parallel architecture may be quite difficult for developers to target and take advantage of, especially if Sony isn't there with good tools as before, but developers/publishers will still choose the platform and thus guarantee its success simply because of its brand strength. Seems a stupid way to decide on a development platform to me, in the long run, at least given the fact that, as I said before, developers ultimately seal the fate of any platform with their support or lack of it. But that's the way it is. God bless the bottom line.

- JavaJones
 
In my opinion the DC, GC, or XB could all have been leaders with enough developer support

Hehe that's wrong, consumers are the ones that decide the leader. There are just too many game developers now days, their support is not a force that dictates the market anymore.

Developers are easily replaceable, consumers on the other hand are harder and more expensive to replace.
 
How do you justify that V3? The developers are the major provider of actual value in the equation. Without them, there is no content. It would be like a TV network without TV shows. You have your TV, you have your Cable connection, but no programming.

Sure you could easily argue that each element in the equation is vital to the whole: platform, developers and customers. One without the other doesn't work, plain and simple. The questions that must then be asked, as a result of that consideration, are which area has more options available to it and is more maleable? Where will the bottom line be hurt the least by focusing efforts? Which of the 3 areas of this equation in consideration - hardware manufacturers, software developers and consumers - can exercise the most power over the others, and in what ways? Where is success ultimately decided, or if not in any one area (as is most likely), then where is the power of decision most focused? I contend it is in the hands of the software developers, the middle of the console market chain.

I find your assertion that consumers decide the fate of a platform to be poorly defended and focused too far down the value chain. The consumer market and opinion is maleable enough that changing something further up the chain is more justified for its effects on later parts of it than simply bowing to the pressures of the *end* of the value chain, in an area in which they (the consumers) have no particular expertise or real knowledge to wield. In other words, you're looking at it bass-ackwards IMO.

The consumers will never make rational, informed decisions that will benefit the industry. It is up to the industry to do this for themselves, and "the industry" in general refers (IMO) to the software developers, for they are the largest and most diverse and financially strained element in the equation. All of the hardware producers basically have money to spare and are generally rolling in capital, whether they make a profit on that division or not. The consumers will buy whatever appeals to their taste, or ultimately whatever is available (consumers are easily fooled, manipulated, etc. just look at Furby, Beanie Babies, the Star Wars franchise, etc.). Certainly there are limits to this, but I'm not suggesting software developers stop producing quality titles by any means. Rather the opposite.

The developers are the only ones who really have solid and defensible needs from a financial and functional perspective, and they're the only ones who are in financial and structural danger as a result of pressures from the two other ends of the game development value system - hardware producers and software buyers. They are in between and ultimately have the most leverage, if only they work together to take advantage of it. Otherwise they are at the mercy of the consumers and hardware manufacturers, both - for all intents and purposes here - essentially large, simplistic factors to consider.

A union consolidates the power of its working members to leverage power to negotiate advantages for its worforce by the manipulating the commodity those workers provide to industry: work energy. It is even simpler for game developers. Without games, no platform will succeed, no matter how strong the brand. No one will buy a Playstation 3 without GTA5, or Tekken 6, or what have you. It's all about brand power, and the developers ultimately control the key brands, and due to their ability -because of the nature of software - to switch platforms if they desire (as Square did with FF, to great effect), their position is functionally most able to make choices that shape the industry.

The reason the Playstation brand is so powerful is because it is synonymous with good games. If, in the next generation, that were rapidly proved to not be the case, I am quite certain the Playstation brand would suffer a swift and dramatic reduction in value. The early PS2's lack of real quality games not withstanding, as there was not a strong and well supported alternative at the time. Sony essentially had what ammounted to a near monopoly for that period. History shows this to be the case in the electronics industry in any case, so I don't need to rely on my own predictions or assumptions. Though technology has changed significantly, the power of branding has changed much less so, so from that perspective, brands that have failed in the past where they were formerly successful because they stopped providing significant enough value, provide a perfect model for the current hypothetical situations.

Surely you can't deny that if, somehow, MS made an exclusive deal with *every* major game developer tomorrow, and halted all production on games currently unreleased for any other platform but the XBox, that people would begin to buy XBox's in record numbers, and PS2 and GC sales would drop off dramatically, especially within a year, where I would predict sales of the two others would be almost nil. That says to me quite clearly that the true value in the industry is the product of the developers, the games themselves. And since the developers *do* have a choice, free will to decide what platform to develop for, regardless of brand, and since they will take the primary value factor with them in that decision (game brands), it stands to reason that responsibility lies with them to encourage positive practices on both ends of the value chain that they lie between.

Ultimately console software developers control their destiny, at least in so far as what platform they will develop for and how successful that platform will be. Obviously this does not extend to being able to control the fate of individual products. As I've said, they simply need to cooperate to some degree in order to actually leverage the power they have. Without some consensus, as much power as they have, it is almost totally useless.

- JavaJones
 
JavaJones - I agree . That and the amount of tittles a system has out . Look at it this way. Say the xbox has a 1000 games out and 900 games suck and the ps2 has 300 games out but only a 100 games suck. A parent goes into best buy and see the ton of games for the x box and the few games for ps2 what do you think the parents will buy.

The main reason why the psx did so well is the amount of money sony threw at it . Fom advertising to buying exclusives. They were able to kill off the dreamcast (which imo is the better system game wise) with hype , money and a metal gear clip which didn't look better than games on the dreamcast. For all microsoft is worth they have a few years of pushing to catch up on and unlike sony , microsoft doesn't have a system meant to win the 32bit 2d wars (saturn) and a what 2 year late n64. I figure the next round microsoft and sony will be split in market share or very close to it. Nintendo well as long as they make a console it will do okay and they have the gameboy to fall back on which will be years to destory
 
Or, to sum up the admittedly rather lengthy and rambling post above:

Consumers are easily led. They also provide no input other than money into the equation, and money is platform and product independant. So, with point 1 in mind, the money can be put where it needs to go by putting the product in the hands of the consumer, regardless of how it is delivered. Ultimately the consumer cares about the product most of all.

Meanwhile, developers invest the majority of the work into a plaform. If a hardware design team invests 10 man years into designing a processor for a console, the software development industry as a whole meanwhile invests hundreds or thousands of man years over the life of a console. Their time can be spent better or worse depending on their development environment, which is my whole point here. They can choose the development environment because, with point 1 and 2 above in mind, the consumers will basically follow their choice in distribution path (development hardware) for their value commodity, because that is essentially the sole compelling reason the consumer participates in this value equation.

- JavaJones
 
lol. That's one place you definitely *don't* want to ramble, too. Especially if it's some kind of in class essay.

From looking at my post history here (and on other forums), you'll see conciseness is not one of my strengths. My ability to summarize and make my points quickly is directly proportional to the amount of time I have to edit though. Were I to spend more time going over these posts before sending them, they'd definitely be shorter. And I know that, at this length, they risk not being read by many, which I feel is a shame as I think I do have some interesting perspectives to impart, whether I'm right or not. It's something I'm always working on...

Anyways, I'll shut up before I ramble even more. ;)

- JavaJones
 
Heh, and some people still believe that the reason why developers flocked to the PS2 platform was because of the underlying technology which provided developers an attractive challenge :LOL:
 
How do you justify that V3? The developers are the major provider of actual value in the equation. Without them, there is no content. It would be like a TV network without TV shows. You have your TV, you have your Cable connection, but no programming.

There are alot of game developers now days.

I think you got me wrong. It would be like TV networks, with too many shows, consumer is the one that will choose.

If there are fews than it would be harder for consumer to choose. Then these few shows has more bargaining power, on which TV networks they stay on. Not the case, when there are too many good shows available over several TV networks. Than consumers will have the bargaining power.

I find your assertion that consumers decide the fate of a platform to be poorly defended and focused too far down the value chain.

Well in today market, they do, consumers decide the fate of a platform.

The consumers will never make rational, informed decisions that will benefit the industry.

But consumer decision is the decider.

No one will buy a Playstation 3 without GTA5, or Tekken 6, or what have you. It's all about brand power, and the developers ultimately control the key brands, and due to their ability -because of the nature of software - to switch platforms if they desire (as Square did with FF, to great effect)

I am talking about now. This is a different market to the past. The future might be different yet.

Say developer of GTA or FF12, jump the gun, from PS2 to GC exclusive. Consumers won't easily jump the gun. There are alot of inertia. This is more so, since there are alot of other developers that are willing to fill in.

So what I am saying, if GTA jump the gun, there will be other franchises that consumer can choose to be the next popular thing. GTA doesn't even need to jump the gun and consumer can choose the next popular thing.

Consumer decides, its no longer developers.

Without games, no platform will succeed, no matter how strong the brand.

See, this is not the case anymore. Previously developers are scarce, not now. There won't be shortage of games today. We are at different stages of the market already.

To sum it up, Currently, developers are competing for consumers, it is NOT consumers competing for developers.
 
Status
Not open for further replies.
Back
Top