nVidia building the PS3 GPU in its "entirety"

[akira888]

You're quite an entrenched PC fanatic aren't you?

If we can build a better architecture than Aaron Spink's team did, then we can talk ultimate_end. When he says that a large performance delta isn't enough to dethrone IA32, he knows what he's talking about.

 

[randycat99]

The operative term would be "when", not "if".

Also do not discount that just as Mr. Spink's background implies credibility, it also may suggest sour grapes. Just a thought...

 

[Mulciber]

The operative term would be "when", not "if".

Also do not discount that just as Mr. Spink's background implies credibility, it also may suggest sour grapes. Just a though...

nor does credibility suggest mature communication skills. comparing a computer chip with communism is so childish that im still completely shocked and baffled that no one on this board called him on it

 

[randycat99]

Very true.

Well, someone did say, "Wow"

I think the notion was pretty universal, imo.

 

[aaronspink]

Microsoft does not think any current or near future x86 design is good enough for Xbox2/Xenon. I think the logic of AMD or Intel not being interested in the project is highly flawed. It doesn't really matter if they are interested in the project or not, it is a matter of what MS wants. MS clearly thinks that IBM has a higher capability of designing and producing a more powerful chip than any x86 processor out on the market now or in the pipeline. Even if AMD or Intel weren't interested and MS wanted them to make a CPU it wouldn't be hard for MS to spark interest so to speak.

What may be interesting to another semi manufacturer isn't to AMD/Intel. Yes if MS wanted they could have got some interest from AMD/Intel, but it would have cost MS too much money. AMD/Intel actually make a profit on their fabs. IBM hasn't make a profit on its fabs for quite some time. Therefore IBM is much more willing to work on pricing as long as it makes some money while Intel/AMD needs to make more money than they would normally with the same process.

Intel/AMD make $200+ dollars per chip. There is no way that MS is going to pay that much. Hence it would be generally uninteresting to AMD/Intel.

 

[aaronspink]

The operative term would be "when", not "if".

Also do not discount that just as Mr. Spink's background implies credibility, it also may suggest sour grapes. Just a though...

nor does credibility suggest mature communication skills. comparing a computer chip with communism is so childish that im still completely shocked and baffled that no one on this board called him on it

Someone made a snide comment about Cell not being the present but the future, so I made a snide comment back about communism not being the present but the future which was actually a theme during the 30's and 40's.

The point being that saying something that doesn't exist is the future is kinda pointless. The future is vast and diverse and while cell may be part of the future computing landscape, it will ony be a part.

Aaron Spink
speaking for myself inc.l

 

[aaronspink]

First of all, The article that was linked to concerning Toshiba showed how faulty your thinking was concerning each memeber of STI's role. Secondly, That's odd that it's down; nAo, Mfa or Panajev can confirm the article. Ohh, and you crack me up -- 'bring it on', heh.

Maybe you should read up on what PR is. Everything from every company is going to take over the world and be integral to their and our future. When toshiba actually uses Cell over all comers, then I'll believe their PR.

Textbook case of an "individual" using argumentum ad ignorantiam.

Actually that would be incorrect.

First x86 is currently the prefered embodiment of a microprocessor. It controls the vast majority of the market and people are willing to pay a premium for it.

Second, it has been such for a very long time.

Third, it has had multiple challengers and overcome them all.

So therefore it has a proven track record. It is only logical to assume the status quo until such a time as proof or example can demonstrate that the assumed status quo is incorrect.

Thus to paraphase Inigo Montoya: "You keep using those words. I do not think they mean what you think they mean."

And I bet he wonders why people have problems with what he says. If your arguments aren't even logically self-consistent, which is pretty fundimental, how do you expect to get anywhere with them? You and Dave need to rethink the logic behind this argument.

Vince, I have yet to read an argument from you that is logically whole. I therefore suggest that those in glass houses don't throw stones.

Aaron Spink
speaking for myself inc.

 

[Vince]

Maybe you should read up on what PR is. Everything from every company is going to take over the world and be integral to their and our future. When toshiba actually uses Cell over all comers, then I'll believe their PR.

As I said, things are inevitable. You can have your few months of talk now, but in a few months we'll see an Intel competitor actually succeed.

Textbook case of an "individual" using argumentum ad ignorantiam.

Actually that would be incorrect.

Oh really? Lets look at what you did state:

• Unless you can demonstrate that x86 isn't "good enough" then it will always be the prefered embodiement of the microprocessor.

I rest my case, this is a cut-&-dry cause of someone stating: If you can't prove [f], it must be [t]. Your supposed "support" is all irrelevent bullshit. And it pains me to hear you quote the Princess Bride.

Vince, I have yet to read an argument from you that is logically whole. I therefore suggest that those in glass houses don't throw stones

Care to point one out? Or are you just speaking for yourself as usual?

 

[Megadrive1988]

Vince wrote:

As I said, things are inevitable. You can have your few months of talk now, but in a few months we'll see an Intel competitor actually succeed

I have my fingers crossed for Cell to rise....rise and say, "yo Intel Pentium, things are about to change around here. you've had your way so easy for far too long"


or something like that

 

[aaronspink]

I am including Sony's investment in fabs as well, because Sony has stated that these 65nm fabs will be used only for Cell IIRC. I am also referring to the investment that all the STI partners have made, as everyone appears to assume that somehow, the total spent by all three companies is all for PS3 and that somehow, Sony has orchistrated this entire project because they want a "kick ass console".

1.2 Billion (generous development costs)
3 Billion Sony's new fab
maybe at most another 300 million for "related" development work.

This includes all development for ALL partners. Total of about 4.5 billion.

Which again would most likely be on the high range.

By "general purpose" I am referring to PC-style general purpose computing and I speak relatively, because according to the patents, Cell is intended to have Integer/Floating Point and Scalar/SIMD capability which means that at high clock speeds Cell will actually perform extremely well at such applications as Word processing etc. I should have been more careful in my wording.

I can today design an 8 Ghz 386. It still wouldn't outrun a 1 Ghz Pentium M in any word processing circumstance.

I can put 32 on a chip, and it still wouldn't outrun a 1 Ghz Pentium M for general purpose workloads.

There is a world of difference between peak theoretical and delivered performance. I don't for a moment believe that Cell will run general purpose workloads with high performance.

If I was doing the micro-architecture, I certianly wouldn't target those types of workloads. The amount of branching, indirect jumps, and function calls I would expect would be minimal.

And that means exactly what?

Exactly what I said, the PC already has a large portion of the media and entertainment applications. Companies like ATI and Nvidia sell a large number of graphics cards designed for this market (the All-in-wonder and Personal cinema series).

People are already switching over to using the PC as their Audio storage and distrobution hubs within their houses. There are whole product catagories that have sprung up around this market.

People are starting to replace Tivos with PC with tuner cards. This will advance as things like the CableCard standard is implemented in the US.

"I certainly don't see the x86 market as a whole clamoring for higher media performance"

Which is completely untrue.

I agree that it is untrue. Over the past 4-5 years a lot of work has gone into improving the media experience of PCs.

What have Consoles got to do with this?

I was using Consoles merely as a reference point. Consoles haven't and don't have the lead on the trends.

As for PC's FPU, well in the overall history of the Personal Computer, I don't see it as such a long time. More importantly, if we think of media/entertainment as being the current "era", then the fact that PC FPU also belongs in this timeframe also suggests "recentness". I am merely trying to highlight the rise of entertainment/media importance in the PC. Frankly, I find your comment to be nit-picking.

The PC era is generally considered to have started with the introduction of the IBM PC. Circa 1981. If you want to go back a little farther, the Apple II was first shipped in 1977. The first PC microprocessor with integrated FPU was shipped in 1989 (the 486).

It is now 2004. That means that it has been 14 years since PCs really started shipping with FPUs. There were only 8 or 12 years that it didn't ship with FPUs (and an FPU was an option with the 386 and 286).

I would consider that more than half the time is a "long time".

And I don't think it is nitpicking. A lot of the revolutionary game and media development happened on the PC.

You need to improve your reading comprehension buddy. I meant important to "companies such as Sony".

Thats my point, I don't believe that it really is that important to sony. If it fails, sony can just use another microprocessor in its CE applications. I think it will be a fine processor for a gaming console like the PS3. But I think that are several designs that would be sufficent for a games console and that the important part is the graphics processor.

I agree that Cell is just a step on the road of computing progress, I never claimed otherwise. I do fail to see however, how you can compare Cell to PPC, which was effectively nothing more than a simple x86 alternative for desktop computing.

PPC had at the time a lot of the same hype that Cell does now. It was going to be revolutionary. Much faster than x86. It would take over the computing landscape.

The truth is that if found its niches in Apple and in some embedded areas (mostly embedded areas that were already using 68K). Power was already being used in IBM's servers and PPC is pretty much the same thing.

I think cell will find its niches in the PS3 game console and several one-off HPTC servers and workstations. But that will be about it. It won't be a failure but it will fall far short of what some fans think it will do.

LOl. You do realise that quite a lot of CPU transistors are dedicated to x86 legacy support don't you? In the past, when overall transistor budgets were small, this was a very big hindrence to peformance.

Not so much that it matters anymore. ISA is dead. It is something it took people a long time to figure out. But its true. ISA is in the power/performance/die size noise area.

Think of all those transistors that could have been used to increase performance. That's just the beginning my friend.

A couple tens of thousand transistors aren't going to increase performance that much. We have examples of it. In an equal process with better designers and pretty much the cleanest ISA you can get, it makes at most 30% difference. Which is about 3 months. No really. Thats why I harp on 2x for 3 generations to spawn a revolution.

Bottom line? Microsoft dictates the PC industry. The majority of consumers use Microsoft applications and Microsoft operating systems. Because of Microsoft, each iteration of CPU has to be backwards compatible and therefore very similar to the previous one. This has prevented innovation in moving away from the tired old x86 architecture. The same thing also applies to the overall layout and architecure of the PC. If you can't see it, then you are blind.

I believe that you abscribe too much power to microsoft. While Microsoft has enough power than it can sometimes guide the direction of the PC industry, even it is beholden to it. Not Intel, nor Microsoft, nor Dell, nor HP, nor IBM, nor AMD can control it. They can exert influence for a time, but not forever. And not over everything.

Then there are industry standards to fight over, such as PCI Express for example. That took quite a while IIRC.

It took quite a while but not so much so. It takes a long time to design new building blocks from scratch. First you have to create a spec. Even with a closed spec this will take quite a while. Then you have to test that it works, and finally you have to design the things that will use it.

These things take the time that they do purely because there are so many opposing forces in the industry. People like you label anyone with a different perspective as "idealistic", but so are the people who seem to have dillusions that the PC industry is some kind harmonious utopia, when could not actually be possible for anything that consists entirely of countless self-concerned parties.

The PC industry is a cacophony. I would totaly agree. But so are the biological processes of life, and they seem to work and progress.

Over a half billion users? If you read my comments and those of others on this thread, then this is actually irrelevant. I would also like to add that just because something is common does not mean that it is good. This is something that everyone should learn at an early age, but maybe you haven't reached that point yet.

The point being that something doesn't get to that large of an active installed base without some significant value in and of itself. A group of companies has to have a significantly better mouse trap to overcome that value.

Yes we all payed real prices. Real prices that are inflated so that every single one of a thousand bickering companies can make healthy profits.

Healthy? Most of the PC market is extremely cut throat with regard to pricing. About the only company in the PC market that doesn't suffer from pricing pressure is Microsoft. Chipset margins are razor thin. Same with graphics margins. Hard drives are extremely competitive. Don't even start talking about pricing pressures at the board level.

There are few markets in the world that are as efficient pricing wise as the PC.

You are quite comical as you seem to think that you are actually seeing some kind of value in your PC!

Pop quiz... What did you use to generate this message?

A PC that you are paying for a decade and a half of x86 backwards compatability.

A PC that is giving you a decade and a half of software and engineering and innovation. For bargin basement prices.

A PC that you will have to upgrade in 6 months because of the next incremental rise in CPU performance that Intel, AMD and whoever else decides to release, because they need to sustain their income in a market that will remain essentially unchanged for years to come.

You may choose to upgrade but no one is forcing you. Personally my PC has been the same for the last 18 months. And no one is forcing you to upgrade. People do it voluntarily.

Yeah like consumers actually care whether x86 powers their computer or not. Consumers are brand loyal to Intel, AMD, NVIDIA, ATI and Microsoft. They are not brand loyal to x86! Again, read the comments made and you will realise that the iron grip x86 has on the market is not attributable some kind of imaginary superior technology or mass appeal. The PC industry is a self perpetuating behemoth. That, my friend, is what will prevent Cell from taking over your daily computing. You are completely dilluded if you believe that the few challenges to the PC industry in the past failed because the PC is the only realistic alternative.

Some would say that compatibility is technology in and of itself. And people do care, because it runs their software. It works with their add in cards.

I am not dilluded. The PC was the only realistic alternative. Because it works with what is there. You can call that self perpetuating if you want but it is the reality. I used to have the same mentality as you. I used all the alternatives. But without a mass appeal, the alternative are destined to fail and without backwards compatibility, they will not have mass appeal.

In any case, the PC era as we know it is almost at an end. The question is, who will win out as the home server king and evolve from there? will it be the PC? Don't be surprised if Microsoft lose the battle to take over your life. If Microsoft loses, it will be the beginning of the end for your beloved PC. Everything dies aaronspink... everything dies.

The PC is dead, long live the PC. The PC is bigger than microsoft. It is a concept. This battle has been fought before.

And not everything dies. The universe doesn't die. Time doesn't die. Concepts don't die. The PC morphs, it adapts, but it is always the PC.


Aaron Spink
speaking for myself inc.

 

[aaronspink]

You're quite an entrenched PC fanatic aren't you?

Nope. Lots of things I would like to change. Lots of things I don't agree with. In some ways I'm just playing the devil's advocate. In others I'm just trying to point out the futility of it all.

BTW, "competition" isn't always quintified by the number of competitors.
If you can give me a list of reasons why all those attempts failed, then we will talk.

Number one reason why they failed? The weren't backwards compatible enough. Or they lacked the perception of backwards compatibility.

In the mean time, hows about you reply to my comments I made on your reply to my original post, instead of taking potshots at my comments to other posts.

[

Done.

Aaorn Spink
speaking for myself inc.

 

[aaronspink]

The operative term would be "when", not "if".

I have no question that someone will or already has architect(ed) something better than what I've done. But thats not really the issue, is it. The issue is whether it will be better enough than the current PC architecture/implementation to matter.

Also do not discount that just as Mr. Spink's background implies credibility, it also may suggest sour grapes. Just a thought...

I got over being bitter a long time ago. Well past the stage of understanding. It is what it is.

Aaron Spink
speaking for myself inc.

 

[Vince]

And not everything dies. The universe doesn't die. Time doesn't die. Concepts don't die. The PC morphs, it adapts, but it is always the PC

Are all Turing Complete systems tantamount to a PC? The answer is an obvious No. A PC is a specific system that's semantically described as one for it's underlying physical features, and in this context -- the correct context -- it will 'die' in the long-term, there is no question.

What will remain is computation and evolution or the variablibility of a stream of information in time, but I'm biologically inclined. What form it takes is amorphous and irrelevent; yet you don't appear to make this distinction between general computation and information theory and the much more limited and contemporary (and transient) concept of the Personal Computer.

And your attempt at a philosophical overtone was cute, although ultimately incorrect in as far as we know about time and the universe. But, you haven't been doing to well with the logic and philosophical aspects so far have you now?

 

[DeanoC]

As I said, things are inevitable. You can have your few months of talk now, but in a few months we'll see an Intel competitor actually succeed.

Are you serious, a few months? Even if Cell is the future it will take years for it to even make a dent in Intel's market.

 

[aaronspink]

[Oh really? Lets look at what you did state:

• Unless you can demonstrate that x86 isn't "good enough" then it will always be the prefered embodiement of the microprocessor.

I rest my case, this is a cut-&-dry cause of someone stating: If you can't prove [f], it must be [t]. Your supposed "support" is all irrelevent bullshit. And it pains me to hear you quote the Princess Bride.

Or stating that unless you can demonstrate that someone is guilty then they are presumed innocent.

It is logical to assume the status quo until such a point that counter-proof can be demonstrated. If you can't understand what makes something a fallacious arguement then maybe you shouldn't try to tell people their arguements are fallacious.

The point being that the arguement I made can be countered in fact if there is such fact that x86 isn't the current prefered embodiment. Such fact can be demonstrated by presenting market data that counter-acts my assertion. Since you appear incapable of presenting such market data, you therefore try to sidestep the arguement by calling it a fallacious arguement. The arguement I made is founded upon logic and is an arguement which can be countered logically which material fact.

An assertion that can be countered with material fact if said assertion is incorrect cannot be considered a fallacious arguement.

Aaron Spink
speaking for myself inc.

 

[Vince]

As I said, things are inevitable. You can have your few months of talk now, but in a few months we'll see an Intel competitor actually succeed.

Are you serious, a few months? Even if Cell is the future it will take years for it to even make a dent in Intel's market.

Success doesn't have to be painted with as broad a brush as you love to use when it comes to Cell. You do realize this, right?

One can make the case that even the existence of a viable product being produced by such a consortium of indepedent companies is a sucess where others have failed. Or one can state that launching an architecture, such as Cell, that outpreforms Intel's offerings in a numbers of ways is a success. Or that even the existence of a competing architecture on the scale of Cell is a success where many others have failed as some here can attest to.

This black or white, total and absolute victory is the only possible outcome mentality of you and a few others is really getting old fast. I never stated as much, nor have I see anyone to my recollection here.

 

[Vince]

Or stating that unless you can demonstrate that someone is guilty then they are presumed innocent.

And there are many cases in which someone is presumed innocent when they are truely guilty. What you stated isn't logic, it's a cultural norm dumbass.

The epistemology of our entire civilization is based around what Davis stated when he wrote, "Lack of proof is not proof."

And the fact being, aaron, if you'd keep up, I stated even before you did that the PC/x86 is at the equilibium point for greastest return for that niche. But, inherient in that question are variables which aren't related to superiority over the competition in the aspects that we're debating here, namely relative preformance.

What, you aaron, are incapable of realizing is that the PC may very well be at a false equilibrium point. You can't prove that it's not and I can prove that it's not the point which will maximize greastest relative preformance. This is why your argument is fallicious, but you don't understand WTF is being said. Instead you've fallen back on the quasi-freshmen in high-school level debate in which you think that your argument is consistent in as long as counter-proof can't be demonstrated. Most of us have moved past that type of thinking around the time puberty hit.

 

[Mulciber]

puberty huh!? why i oughta...

 

[passerby]

When these discussions start becoming philosophical, you know THINGS ARE NOT GOOD(TM).

 

[one]

We have to spend another 3 weeks until the CES without food