Congratulations to Vince
- Thread starter bbot
- Start date
- Status
Phil said:
Ok, so they are calling it the BE, and the BE is in the PS3, but they are not the same. Gotcha.
I was clearly talking about Desktop apps. Consumers and small businesses (or even large business) do not want to spend money on new software and training. Now workstations and server racks--that is another issue. Companies will make risks in these areas because the performance issues can be well worth it. But the entire Wintel talk is pure silliness. Until STI present a clear path on how to support software they are not going far. This is not to say a PS3 could not become some type of convergence device, but breaking apart Wintel is more than just getting your foot in the door. I remember when RISC was the next big thing and how the PPC was going to change computing. I remember how everyone would have these cheap Oracle computers and run everything off large super computers in datafacilities. WebTV? Check. PS2 and the great Emotion Engine? Check. The list goes on and on and on.
Yet you are placing a bet on a plan Sony has never officially announced. You have no clue what they are going to do. At least I am confident the PS3 is, and will be, a gaming console. Will it do anything else? Sure. Is it poised to topple Wintel? Heck, is it even prepared to get its feet wet?
The fact is many challenges have been made to the x86 throne and they all fail. Accusing me of being "gaming centric" when we are talking about a gaming device is quite silly. Until the PS3 proves to be anything different I stand by that. As it was already pointed out PS2 was supposed to be this huge hub--never happened. I remember the EE hype well. CELL may live up to the EE hype but you guys buy into the anti-Establishment promises too easily because all you ever see is the fall of the market leader. The fact is Sony has never claimed half the stuff you guys want it to do--you are pushing your own pie in the sky hopes on the machine. This is the same thing with the 1TFLOPs number. Go back and READ my reasons why this will be difficult from a FAB perspective. I never said Sony could not do it, but from STIs own showing it is not looking great. And Li Mu Bai gave some excellent financial reasons why this could be an overly risky move.
Oh well, you challenge some statement like "The PS3 will have 1TFLOPs" with some sound reasoning from STIs own info and you get all kinds of resentment. I am still waiting for you guys to explain how they are going to stick 4 CELLS into a PS3 and keep it cool, quiet, and affordable (that includes good yields) while not skimping on RAM and the GPU. Surely a convergence device of the scale you are discussing is going to need more than 256MB of RAM.
Acert93 said:Ok, so they are calling it the BE, and the BE is in the PS3, but they are not the same. Gotcha.
I'm not sure why everything must be repeated multiple times, but CELL (as it seems Broadband Engine == 1 PPE with 8 SPUs) is a scalable architecture. We know CELL variant will be in PS3 but the configuration (name it what you want) is unknown (unless you care to point me to some offical statements).
Acert93 said:
And yet, there are people that are clearly not sharing your limited angle. Do a search on CELL and Vince and you may dig up some posts explaining some of the advantages you are missing. Those posts also include corresponding links to articles with quotes that indicate what they're doing.
BTW; I'm not claiming a TFLOP processor is viable (that all depends on the purpose it serves) - I'm saying WE don't know what purpose in Sony's strategy the PS3 is going to play - but we should certainly not limit its performance estimates on the ground it just being a pure gaming device.
(Sorry, I don't have the time to dig them up myself right now, but seeing how much time you invest in writing replys, I'm sure you'll find them)
Phil said:
You assume I do not understand that. Yes CELL is scalable. Everything so far is pointing to a 256GFLOPs CELL. As far as I know BE is what Sony was naming the PS3 incarnation. Are you saying that is incorrect?
I have read Vince's posts. The question is not the hopes and dreams of the "have nots" or the anti-establishment. The question is has Sony stated clearly that this is their goal. You do not overtake the market and replace it by not having a clear plan in place with developers. If STI is trying to go this road themselves then it is a failure already.
And lets be totally honest here: The same stuff we are hearing about CELL is the same hype from EE. Whether you accept that or not is up to you, but the fact is until STI actually provide a product and then followup with the services and abilities they are claiming are possible it is all a pipe dream. It is one thing to have a machine capable of such things, it is another to have a machine that actually accomplishes those things. It is easy to promise the world--now show us.
Instead of stating the same thing over go back and LOOK at the reasons why a 4 CELL BE is not likely right now. It has nothing to do with "Oh it is a gaming box so there is no way they stick 4 CELLs in there". Gaming box or not, you still have to deal with issues like heat and power consumption. Those are not going away anytime soon. Die size and yield (aka cost) are always a factor. 1 CELL on 90nm (200+mm^2) would be over 400mm^2 at 65nm with 4 CELLs. Die size, yields, heat, power--those issues wont go away. Put cost aside, yields could make availability very tight. And like it or not they still got to fit this all into a $300 unit. Any way you cut it there are a lot of hurdles to get to 1TFLOPs now. There is more than just wishing or saying "Well maybe they are looking bigger?" There are limits. Intel cannot slap a 5GHz chip on the market with will power alone. Manufacturing limitations and cost always catch up--even to the big boys.
And while you are quick to criticize my supposed narrow view, the "gaming machine" angle is what gets the PS3 in the door. Don't forget that. Without that STI has no chance for the consumer market. If you bump it up to $500 or $600 it no longer gets the market penetration needed to accomplish this "CELL World". So it is important to have the low sticker price and it is important for it to play games because without games Sony has no inroads.
So (a) the limitations I see are not restricted to the fact the PS3 is a game device but are also technical issues to overcome and (b) the fact STI is banking on the fact the PS3 is a gaming device and will have market penetration dictates a "low" sticker price and that it not be unreasonably big in size. Sony is not MS. There is only so much they can lose on this device, and you guys are suggesting take a brand new chip at over 200mm^2 on 90nm and cutting it down to 65nm by Spring of 2006 (1 year from now) and sticking 4 CELLs on a die over 400mm^2. If Sony does that there will be HUGE availability issues and a lot of hurdles to overcome technically. So it may not be impossible (never said it was) I say look at the facts: the PS3 CPU will dwarf the Xenon and the PC in FLOPs with 1 CELL. You consider this irrelevant, I do not. It still is competing in the market and there here and now. While STI may want this to be more than a console that is all it currently is. And I am still at a loss on how giving it 4 CELLs over 1 CELL makes it "better". To reach the goal of the CELL World is more than just more power--it is about better design, better apps, and more convienence. Throwing power at the problem is not the solution. Look at the Xbox--was more power enough? Of course not. PS3 has the power, the question is not whether it needs more but what it will do with the power it has.
Unless the technical hurdles can be overcome in 1 year with high yields and in an affordable range I see no way this is possible. I am not saying it is impossible, but I am saying STI would have their work cut out for them. And I really doubt Sony has $4B--every company has their limits, regardless of their goals.
If the PS3 won't have more than 1 CELL chip (not 2 cores per chip), why have the 75GB/s FlexIO port? That seems to suggest that they will put two CELL chips atleast in the PS3.
They could have easily made a server version later with FlexIO and a consumer "PS3" version, without the highspeed interconnect. I don't buy that the CPU<->GPU channel is the major reason for this design.
They could have easily made a server version later with FlexIO and a consumer "PS3" version, without the highspeed interconnect. I don't buy that the CPU<->GPU channel is the major reason for this design.
Panajev2001a
Veteran
DemoCoder said:
The FlexIO port will connect more than the CPU...
.GPU, I/O chip, etc...
http://docencia.ac.upc.edu/ETSETB/SEGPAR/microprocessors/emotionengine%20(mpr).pdf
i found this apr report on EE after ISSC and ps2 unveil. 2mths later ps3 one should follow.
cell is more scalable then EE and with nivdia, ps3 should do much better than ps2.
though no doubt new stuffs alway get people excited, let see how it will work and last better.
i wonder which CG film sony will referance this time.
i found this apr report on EE after ISSC and ps2 unveil. 2mths later ps3 one should follow.
cell is more scalable then EE and with nivdia, ps3 should do much better than ps2.
though no doubt new stuffs alway get people excited, let see how it will work and last better.
i wonder which CG film sony will referance this time.
Spidermate
Regular
Acert93 said:DemoCoder said:
I disagree. The issue is not pure power but price
I never said not to increase the power at all, but consider: Top of the line P4 is in the high 20's gof GFLOPs (that was off a website, IGN I believe) and XeCPU is rumored to be ~80ish GFLOPs (no on who knows is saying, so pure speculation). If 1 CELL is 256GFLOPs you are looking at 8x the performance of a top of the line desktop and 3x the performance of the XeCPU (although comparisons at this point are sketchy).
So I ask: Why would they need to make it even more powerful? Would not the cost for more CELLs be better spent on other areas? It is like the body builder with big arms and chicken legs. He might impress a few people, but when it comes to the competition or sports the well rounded athlete kicks his butt.
I agree we need more power for AI, physics and general believability but 1 CELL is a good step toward that. 1TFLOPs is a totally arbitrary number. Wny not 2TFLOPs? or 16TFLOPs? 1 CELL is substantially faster than what is going to be available, so why 4? It will cost more AND most cross platformed games wont take advantage of it.
If the concern was "massive amounts of processing to realistically simulate physics, AI, sound, etc" on a $300 machine I would look to spend some more resources on memory. This is just me, but a 256GFLOPs CELL and 512MB of XDR is more appealing to me than a 1TFLOP machine with 256MB of XDR although we do not know enough to compare pricing. A 1TFLOPs/256MB PS3 may cost a lot more than a 256GFLOPs/512MB CELL so the comparison would not be fair.
Anyhow, good design is more important than putting all your eggs in one basket. 1TFLOPs is just a number. The rest of the system needs to be able to feed that CPUs. And at a $300 price point, I would think if you were beating your competition by 3x in CPU performance you would look to put more money in other areas. You have to remember if they spend too much on CELL and neglect other areas that leaves room for competitors to move in. A great CPU with a weak GPU would be worse than a solid CPU and GPU. Balance is more important than pure power.
Did you make those retail numbers up yourself or were they officially announced? What you seem to be comparing here is dated history. How do you know sales won't increase next-gen? To say that Sony, Microsoft, or even Nintendo will be shooting for a $300.00 mark is silly? Who would have expected Sony to launch a handheld at $185 to $250 to the DS's $150? Although this is a price war, any one of these consoles could still launch at $350 or $400 to shave cost.
Also, would you mind telling me what the NES launch price was compared to what we are paying for a console now? :? [/b]
Megadrive1988
Veteran
if you wanted a game and 2nd controller with your Neo Geo cartridge system, it was $600~$650 8)
Megadrive1988
Veteran
Acert93 said:
so, it's going to be a 1 Cell-CPU, or maybe a 2-Cell CPU at most, but not a 4-Cell CPU. I am okay with a 1-2 Cell 256-512 Gflop CPU. I mean, developers should be ok with that much. besides, what we see on the screen will be upto the Nvidia GPU. adding 2 or 3 more Cell PEs to PS3 won't automatically make the PS3 better, or what we see on screen better. so basicly, I am agreeing with you.
Acert93 said:
I'm not going to bother anymore. Lets put it this way: Link me to an official statement where BroadbandEngine == PS3 CPU. Unless you can do that, I'll just assume that it was merely popular belief derrived from the patent and has nothing to do with the PS3.
Acert93 said:
Depends on your definition of "clearly". For some it is, for others, it will never be. The truth is out there - if you choose to believe it or not is up to you. There are certainly clear hints in what direction it's heading and this all in relation to CELL and the hundreds of million in investment bound to it. There are also interviews and quotes from the past that are sending out a very clear message - and the latest revealings are adding up slowly. A clear all-in-one message for the non-believers will surely follow in due time.
Acert93 said:
Now? Who's expecting PS3 with x TFLOP to launch now? I think you're confused. There's no need to go back and read the obvious - I am very well aware of the issues surrounding a TFLOP version (which is why I'm not hoping for one). I will say however that performance is obviously very bound to the timeframe it's launching in and at what price. Those two factors are very dependend on what purpose the PS3 should serve. Ultimately, a dedicated gaming-device will have a different cost/timeframe than a more living-room centered device would. I'm not entirely sure PS3 will launch first thing in 2006 nor in which form its performance figures will be. It also isn't bound to having 4 PPEs on one die - perhaps 2 external chips? I'm not even going to rule out the possibility of seeing a PS3 that is upgradable in some kind of form. Maybe Sony is showing its best cards now... maybe they still have something up their sleeve. There are just too many things uncertain at the moment that I simply won't take a 1 PPE in PS3 for granted. I do see however that the PS2 business in running extremely well, which makes it more likely for PS3 to launch later regardless of what competition is doing.
Acert93 said:
You've already misquoted once, no reason to do it again. While a 400mm^2 chip is certainly out of the question, it doesn't rule out multiple CELL chips on the same mainboard (not on the same die) nor is Sony bound to the timeframes you're suggesting. We also don't know what it effectively costs Sony for each PPE and how good the yields are/will be. These are all going to be factors.
Acert93 said:
As I said, I doubt Sony is bound to your prefered timeframe(s). Perhaps 1 year (March 2006 for Japan) is the most likely, but it certainly isn't their only option. All depends if Sony will play it safe or be willing to take a gamble - and that depends again on various other factors - factors you're still downplaying.
Just to express that Sony may not be interessted in taking X86 market ( form what we see from Aplee it, it is not that good to that), but if you see Sony TVs like WEGA and those products that starting to appear which have a price really really high due to big circuits and so. There a 4 PU cold do a lot because it is a WEGA TV that I want, not a new TV only a bit better, which I can not buy, so if they can put those prices low enought then they can give a reason to people start buying new TVs beyond -Dam my TV is brook I need a new!
Cell could be a way to renew their own market.Remember that Cell is, foremost, a multimedia processor.
People do not want, do not have time or even money to star their learning, colections ....again and that is the reason for Wintel dominance!
But in high end off line renders and those markets meybe they have a shoot.
Cell could be a way to renew their own market.Remember that Cell is, foremost, a multimedia processor.
People do not want, do not have time or even money to star their learning, colections ....again and that is the reason for Wintel dominance!
But in high end off line renders and those markets meybe they have a shoot.
Sony or any other manufacturer won't cut in in the PC market anytime soon simply because whatever ship they make won't be compatible with x86 code.
Until a chip is made that can run on an OS that has as much software available to it as Windows does, no one's taking over anything.
They can go the OSX route, but then they would only take over the Mac market, which is not really huge. Good enough, and definately a good starting point, but really, there is a reason why there are billions of Windows compatible PCs in the world, and availability of software for it is the main one. How it got to that point is another discussion altogether.
Until a chip is made that can run on an OS that has as much software available to it as Windows does, no one's taking over anything.
They can go the OSX route, but then they would only take over the Mac market, which is not really huge. Good enough, and definately a good starting point, but really, there is a reason why there are billions of Windows compatible PCs in the world, and availability of software for it is the main one. How it got to that point is another discussion altogether.
pc999 said:
The question I have here is for what Multimedia applications and when? In its current form its a fairly large chip, and in widescale consumer applications it will still be so in a shrink or so's time.
- Status
Similar threads
