PSX3 and PSP blamed for Sony Group's poor performance..

[DeadmeatGA]

...

why should any CE maker produce its own chips ?

Exactly, when there are low-cost foundries out there.

Samsung that make most if not almost all of the major components they use in-house.

Samsung is a leading foundry; SCEI is not. Comparing an apple to an orange.

Dell, btw... operates on quite razor thin margins as well...

At least they are not losing money even in economic downturns.

they get special deals from Intel and other CPU makers.

You don't get any deal if you fabricate it yourself.

if you buy LCD panels for 15'' monitors from Sharp, it is quite tough to make 15'' LCD monitors cheaper than the equivalent Sharp ones.

Actually you could. Go shop elsewhere. Lots of hungry Korean and Taiwanese LCD manufacturers willing to cut you a deal.

Here you are looking like a fool slamming down a console, PlayStation 3, almost 2 years before it is launched.

Did you think I was a fool when I was dissecting EE apart and exposing it as the fraud it was 4 years ago???(66 million polys/s, I still want to see it done)

Sticking a $15 dollars CELL chip manufactured in-house

$15 CELL chip???? That was funny, I thought it was a type, that you really meant $150. Call me when the EE fab cost drops below $15, 5 years into production and SCEI is still not there.

instead of a $20 chip bought from a third party vendor

Which does the job and costs 1/8th~1/10th of CELL.

Also, with all the fab space SCE has, they have the capacity of producing more chips for the rest of Sony furtherly saving money and not leaving those 90 nm and 65 nm lines used at non full potential.

Recycling older fabs is a nice idea.(Handheld engine comes to mind) But CELL is clearly either unsuitable or an overkill for 99% of consumer electronic applications and there is no need to use one. ARM didn't become the king of embedded market for no good reason.

 

[Paul]

Did you think I was a fool when I was dissecting EE apart and exposing it as the fraud it was 4 years ago???(66 million polys/s, I still want to see it done)

Your kidding right? That was a PEAK performance measure of the EE, something Sony said time and time again. I don't see you knocking on MS for claiming their Xbox could do what 150 million polygons per second raw?

 

[Panajev2001a]

With the money ARM makes ( $120 Million a year in revenue ) if there is a big shift in embedded CPU development there is the chance they could get steam-rolled: no wonder why Intel has its own ARM-like architecture, XScale which once was StrongARM.

Recycling older fabs is a nice idea.(Handheld engine comes to mind) But CELL is clearly either unsuitable or an overkill for 99% of consumer electronic applications and there is no need to use one.

There is a need and you can stop looking at CELL as a 200-1,000 GFLOPS monster and think how it could fit small Consumer Electronics devices.

CELL is an architecture, one which was designed to scale from small devices to large scale ones.

Also, SCE is not that uncomparable to bigger foundries like Samsung, especially not in the contest of CELL.

2 big fabs with 300 mm wafers and 65 nm ( and then 45 nm ) technology and 200 mm wafers and 90 nm technology are not exactly antiquated.

Other two have 130 nm technology and can be upgraded to 90 nm and 65 nm, since PlayStation 2 production is moving abroad ( except the 90 nm parts ) it is likely we will see the upgrade taking place.

Plus Sony has more fabs and it is building others: LCD plant agreement with Samsung.

Sony is currently taking a very midle-road approach between fabless Consumer Electronics makers ( for which it is becoming more and more difficult to push the technology boundaries... this is a concern common to fabless companies in that field ) and guys like Samsung and it is not paying off ( Sony Consumer Electronics is seeing the harsh reality of it ).

Pushing for consolidating their Semiconductor R&D and strengthening their flagship SCE is a good way for Sony to conserve its own identity and to re-organize and move towards higher profitability.

The current approach is not paying off and the best portion of Sony is the one who has steered clear of that approach.

There is not only the fabless way to go in Consumer Elecronics, especially for the market Sony usually cathers to: if they moved to that approach they would die quite quickly and you know it: this is the reason why I am not surprised seeing you advocate for such a scenario like when you wanted SCE to re-design PlayStation 2 and delay it for a couple of years and launch PlayStation 1.5 instead.

Did you think I was a fool when I was dissecting EE apart and exposing it as the fraud it was 4 years ago???(66 million polys/s, I still want to see it done)

That is 102.5 MVertices/s for you which is the same as 116.5 MVertices/s for your beloved NV2A: maximum theoretical Vertex Transform rate.

So yes, you were a fool and that kind of comment is not the only foolish thing ( do we have to go back to your predictions about how succesful the PlayStation 2 was going to be in economical terms ? ) you said due to your loathing of SCE and Ken Kutaragi in general.

 

[Vince]

Wait... is that number with or without the extra VU1?

< ducks and runs >

 

[cthellis42]

Re: ...

$15 CELL chip???? That was funny, I thought it was a type, that you really meant $150. Call me when the EE fab cost drops below $15, 5 years into production and SCEI is still not there.

...because they will of course be putting PS3-grade chips in every device using chips designed specifically for scalability.

Tell me, do you ALWAYS take the most nonensical approach you can think of when discussing Sony? DID Kutaragi run over your dog?

 

[gurgi]

Why do I feel like I've been transported to the pre-PS2 speculation era? Now the hype has a name.. and it's called cell. Hope it pans out for everyone emotionally involved.

 

[Megadrive1988]

this is the reason why I am not surprised seeing you advocate for such a scenario like when you wanted SCE to re-design PlayStation 2 and delay it for a couple of years and launch PlayStation 1.5 instead.

Panajev, I am glad you mentioned this.


I recall, long before final PS2 info came out, back in 1996-1997, the reports about Playstation-2, or Playstation-64 (64-Bit) as it was called, mentioned that it would push 800,000 polygons a second with everything on like textures, effects, etc. making it somewhat more powerful than the M2.

http://tinyurl.com/sgef

PSX2 ON THE WAY?
Sony has openly admitted that there will be a PlayStation 2. Its being called the "Highlander Project" in house, and there is talk of 8 mb internal ram memory, anti-aliasing, z-buffering, trilinear mipmapping, bilinear texture filtering - all accomplished in hardware and able to push more than 800,000 polygons with all features on.

The details and format are not clear, but it will most likely be in the form of a quad speed or 6X speed CD ROM. Backward compatibility with PlayStation 1 software and peripherals is a feature. The machine is not likely to appear for at least 18 months.

deja/google hit return for "sony 800,000 polygons":
http://groups.google.com/groups?q=Sony+800,000+polygons&ie=UTF-8&oe=UTF-8&hl=en



- there were even reports that said the new playstation would use the PowerVR Highlander chip, though i think this might have been confution on those who reported on such things because of Sony's 'Highlander Project' the PSX2


Around 1997, magazines like P.S.X. or whatever it was, had articles on Playstation64. they were saying it would have 8~16 MB of main memory, 2~8 MB of VRAM. filtering, AA. have at least a 100 Mhz CPU. better sound. about 1 million polygons per second. they said the new Playstation would need to show a 10-fold improvement over PS1.

Then just after Sega Katana (Dreamcast) specs were known (at first 1.5mpps) there were reports that said said Playstation2 would do "a couple million polygons a second" or about 50% more powerful than Katana/Dreamcast.

I believe the Playstation2 we have today is really PS2.5 or the original PS3, because, as far back as 1995, there were rumors that Playstation 2 would be out in 1997-1998, and Playstation 3 in 2001 !

 

[Urian]

If my eyes doesn´t lie to me, I see that our friend DeadmeatGA has the limited point of view of a PC User.

The use of the PowerPC Architecture is the key, if you examine this architecture you can see a lot of advantages.

The advantage is the size of the chip, if you remember G4 and Athlon the two processors were born in 1999 and G4 is a 50% more faster and G4 has a half of the size of Athlon (10.5 milions of transistors for G4 and 22 milions the Athlon).


Cell is a third way product, created, designed and manufacturated by STI.

1) IBM is interested because they want that its PowerPC and PowerX market will growth, today the architecture is used in IBM mainframes, AS/400 servers, and Apple macintosh (Every year Apple sells 7 milions of MAC in the world). IBM wants that PowerPC becomes and estandar beyond than MAC world and IBM servers and PS3 is an excelent key.

2) Toshiba is interested because MIPS is in comma now and the dead of MIPS architecture could be soon because PPC is stronger in marketing and support and ARM is eating all the handheld devices market.

3) Sony is insterested because they will manufacturate the 100% of all the products and Cell is the key for the total independancy of Sony, the cost reduction and the free of doing prize down to its products without the fear of losing money (see MS with XBOX and the economical problems with the machine).

 

[DeadmeatGA]

....

if you remember G4 and Athlon the two processors were born in 1999 and G4 is a 50% more faster and G4 has a half of the size of Athlon (10.5 milions of transistors for G4 and 22 milions the Athlon).

Die size was comparable. It simply meant that Athlon had a larger cache. Compare Intel Dothan to Intel Prescott; Dothan has 40% more transistors yet it is 20% smaller. Why? Because it traded in logic gates in favor of SRAM gates which can be packed in at 8X density.

This is why I estimate a massive die size for EE3, its design is inherently logic heavy and SRAM light.

Cell is a third way product, created, designed and manufacturated by STI.

Save for the fact that Toshiba and IBM has no uses. Or any other businesses for that matter.

IBM wants that PowerPC becomes and estandar beyond than MAC world and IBM servers and PS3 is an excelent key.

Tired old logic. Gecko didn't make PPC any more acceptable to embedded market than before. CELL won't change that.

2) Toshiba is interested because MIPS is in comma now and the dead of MIPS architecture could be soon because PPC is stronger in marketing and support and ARM is eating all the handheld devices market.

So Toshiba should be an ARM licensee like everybody else.

3) Sony is insterested because they will manufacturate the 100% of all the products

Memory is not manufactured by Sony. Sony cannot manufacture components for its VAIO lines.

Cell is the key for the total independancy of Sony, the cost reduction and the free of doing prize down to its products without the fear of losing money (see MS with XBOX and the economical problems with the machine).

What has changed? Instead of paying royalty to MIPS, Sony will be paying IBM. Sony doesn't own CELL.

...because they will of course be putting PS3-grade chips in every device using chips designed specifically for scalability.

Even the unicore version will be at least 100~120 mm2 large and does not carry any auxiliary circut, far too large for embedded applications.

 

[DeadmeatGA]

...

With the money ARM makes ( $120 Million a year in revenue )

Meaning ARM license is not all that expensive. Better to license from ARM than to build your own.

There is a need and you can stop looking at CELL as a 200-1,000 GFLOPS monster and think how it could fit small Consumer Electronics devices.

It can't. Even the unicore version will measure at least 120 mm2 @ 65 nm. You must understand why ARM is so popular.

1. It is very small.
2. It is synthesizable.
3. The license fee is relatively cheap.
4. You don't have to pay anyone else for its use.
5. Has plenty of tools and developers.
6. It is very power-efficient.

With CELL

1. It is extremely large.
2. It is only available as a discrete device.
3. It is extremely pricy($150~200+)
4. You have to pay IBM and Toshiba in addition to SCEI.
5. It has no tools and developers are already scared.
6. It can double as your room heater.

Contrary to what SCEI claims, CELL in its present state cannot be used for portables and SCEI admitted it with the PSP design. Even SCEI won't use CELL for its own portables, so why are you expecting others to do????

CELL is a console architecture and has no other applications. Get used to the idea.

 

[Josiah]

PS3 R&D won't bankrupt Sony. They have enough cash to get it finished (they are after all in the black still). If something bad happens after the R&D (difficulty moving to 45nm/product delays/poor sales) it might affect PS4. But they seem to be "past the point of no return" with PS3.

 

[randycat99]

What sort of PPC core takes up 120 mm^2 at 65 nm?

 

[cybamerc]

Panajev2001a:

> I think they have a very good chance of survival

Of course. A company of Sony's size doesn't just go under.

> A good reason of the profit margins being so thin is that they are
> putting in advance PlayStation 3 R&D into the books:

Except it is not a recent thing and it applies to the whole corporation.

> even if you have not wasted once occasion to troll Sony in a long while.

Disagreeing with your ridiculous optimism does not equate trolling.



_phil_:

> We would expect to see Cell based products before PS3 arrives, to
> drive costs down more quickly.

We would? Why isn't it in PSP then? According to Paul there should be nothing keeping them from doing that.

 

[Paul]

Why isn't it in PSP then? According to Paul there should be nothing keeping them from doing that.

Because Sony wants it in PS3 first. PS3 is and always will be Cell's main application.

Sony is already working on a PS3 version of Cell as it is, they don't need to be diverting valuable engineers to a PSP Cell when they need them for PS3's version.

PS3 is afterall the "big thing".

 

[Vince]

Disagreeing with your ridiculous optimism does not equate trolling.

You know, you're totally correct. If all disagreements were like yours or based on a factual and plausible basis, then we wouldn't have these problems. But, DeadmeatGA here is not only the spiritual father of the PS2 (the pre-VU1 iteration before the scary Sega people frightened Ken to timetravel back and put in a VU1), he also thought up the Cell and is claiming to know things about the architecture and it's licensing model that he just doesn't. Shit, it probably doesn't even exist yet.

This is trolling in it's purest form. Yet, it's so pure it's frightening because he truely believes this utter BS. I disgress so harshly from his posts that I can't even respond to them because I'll loose all semblence of composure and just tell him that he's a retarded ass; and not because it's Sony he's talking about, but because it's an insult to the rest of us that he expects us to believe, follow, or even contimplate these insane thoughts.

Cell was clearly designed for the CE enviroment and Ando or Kutaragi or IBM has said as much in the past. Yet, he keeps saying it's not. He's to ignorant to inquire into the STI agreement and instead says that Sony doesn't "own" Cell but will be paying IBM (WTF?) - which is wrong. He keeps bashing these mystical "tools" yet he doesn't know anything, he bashes the architecture as often as possible latching on to any "current" arguing point that sticks, yet his knowledge on the matter is limited. It's not right, it's not deserving of a place here.

*The Kutaragi article is a great reread for those who don't have a clue what's going on - of which I can think of a few people.

Within it, Kutaragi talks of the product differentiation this will bring that I speak of and explains how Sony's Semiconductors and Cell differ from the fallicious comparasons to Samsung, et al that some make. Among much more, so if you have a few minutes to spare...

 

[DeadmeatGA]

...

What sort of PPC core takes up 120 mm^2 at 65 nm?

A PPC core with 9 Altivec units and 1 MB cache. Unfortunately, IBM wouldn't dare to build such a beast.

 

[Fafalada]

Cell-tronics is the lack of a Cell forum(ala DVD forum) yet. Yes! Im a firm believer of a standard. When one as widespread and common as home electronics, a standard is even more important.

x86 market is probably numbering in bilions by now and it never had a forum dictating a standard for electronics powering it - granted it has it's issues but plenty of people (including yourself) are quite fond of it just the way it is, including the respective monopolies that dictate it.

Anyway, should there be a "Cell"-Forum I would hope it has a lot better track record then DVD one, which is all but ignored by any companies that have an idea to create a new format anyhow, especially if they are members of the forum. :?

 

[randycat99]

Re: ...

What sort of PPC core takes up 120 mm^2 at 65 nm?

A PPC core with 9 Altivec units and 1 MB cache. Unfortunately, IBM wouldn't dare to build such a beast.

There are lots of different kinds of PPC's. What PPC core do you mean? Where do you get that they will be using full-fledged Altivec units? Seems to be a lot of assumptions in there...

 

[Paul]

Seems to be a lot of assumptions in there...

Assumptions? Dude, he created Cell.

 

[randycat99]

Might be amusing if he posted scans of his napkin sketches- stick figures and all...