ITPC 2005 (and some other Cell-related events)

[one]

ITPC 2005, 21st Annual SEMI International Trade Partners Conference, for semiconductor/FPD industry senior executives was held in Oct. 30 - Nov. 2 at Maui, Hawaii, with golf tournament and lessons... (agenda)

http://techon.nikkeibp.co.jp/article/NEWS/20051101/110328/

Kutaragi gave one of the keynote speeches. First, he talked about the 1 PFLOPS Cell cluster which was mentioned also at the Tokyo International Digital Conference and emphasized the objective of the development of Cell by STI is computing. Then he talked about the importance of the manufacturing scheme of LSI makers, citing the lost opportunities in Playstations caused by the inability of the 90nm logic LSI manufacturing resources to catch up with the demand of Playstations. Also in future, Playstations need the massive amount of memory, and Kutaragi said "our computer entertainment business coexists with the semiconductor industry."

More interesting than this talk is those charts:

It is clear that PS3 needs more chips than PS2, but the mass production of the memory chips seems to start later than Cell & RSX. PS2's memory chips may be combined into less chips, but the console is expected to sell in 2006 as much as in 2005.

In November 1-3, 2005, A-SSCC 2005, Asian Solid-State Circuits Conference 2005, was held in Hsinchu, Taiwan. STI gave a session about the power efficiency of CELL.

I-I-3 11:40-12:05 The Power Conscious Synergistic Processor Element of a CELL Processor

Scott Cottier(IBM Systems and Technology Group, Austin, TX), Osamu Takahashi(IBM Systems and
Technology Group, Austin, TX), Sang Dhong(IBM Systems and Technology Group, Austin, TX), Brian
Flachs(IBM Systems and Technology Group, Austin, TX), Koji Hirairi(Sony Computer Entertainment
of America, Austin, TX), H. Peter Hofstee(IBM Systems and Technology Group, Austin, TX), Brad
Michael(IBM Systems and Technology Group, Austin, TX), Hiromi Noro(Toshiba America Electronic
Components), Dieter Wendel(IBM Systems and Technology Group, Austin, TX), Michael White(IBM
Systems and Technology Group, Austin, TX)

Keywords: CELL, Synergistic Processor, Low Power, Architecture

Abstract: A 4-way SIMD Streaming Processor of a CELL Processor is developed in a 90nm SOI
technology. CMOS static gates implement the majority of the logic. Dynamic circuits are used in critical
areas, occupying 19% of the non-SRAM area. ISA, micro-architecture, and physical implementation are
co-optimized to achieve a compact and power efficient design.

http://techon.nikkeibp.co.jp/article/NEWS/20051102/110353/

http://techon.nikkeibp.co.jp/article/NEWS/20051102/110353/Cell01.jpg
http://techon.nikkeibp.co.jp/article/NEWS/20051102/110353/Cell02.jpg
http://techon.nikkeibp.co.jp/article/NEWS/20051102/110353/Cell03.jpg

Finally, Embedded Technology 2005 is going to start from Nov. 16 in Yokohama, Japan.

http://www.jasa.or.jp/et/english/conf_p.html

Conference Center501/502 The Engineers Who Created Cell
-The Impact of Multicore Technology-
- Moderator -
Naoki Asami
[Nikkei BP, Inc.]
- Panelist -
Suzuoki Masakazu
[Sony Computer Entertainment Inc.]
- Panelist -
Tago Hiroyuki
[Toshiba Corporation]
- Panelist -
Koichi Awazu
[Toshiba Corporation]

 

[Carl B]

Good find One!

 

[Shifty Geezer]

Those graphics - are Sony being optimistic of demand for PS3 or do these represent more components (CPU+GPU)? Starting with twice as many chips per month as PS2's midlife before the single-chip solution, yet saying PS3 will be expensive...sounds like a recipe for unmoved stock.

 

[TekkenMaster]

Can someone explain to me what logic LSI are?

 

[Platon]

I don't get it, why isn't ther any demand for 05? I am reading those charts wrong? Will they not start production in 05?...

 

[London Geezer]

I don't get it, why isn't ther any demand for 05? I am reading those charts wrong? Will they not start production in 05?...

2006 Financial Year started in either April or July 2005. Makes sense.

 

[one]

I don't get it, why isn't ther any demand for 05? I am reading those charts wrong? Will they not start production in 05?...

I think the part drawn with dotted lines in the Logic LSI graph represents the PS3. Doesn't it?

 

[archie4oz]

Can someone explain to me what logic LSI are?

Processors...

 

[mckmas8808]

Did anybody else notice that in 2006 the PS2, PSP, and PS3 will be using 65nm tech?!? So will Sony rep some savings from the PS3 in 2006 when do this? Hold didn't l-b say that the 2006 year actually started in April of this year? What going on guys? I'm damn confused. I though the PS3's CELL was going to be 90nm. Can someone explain?

Thanks

 

[Shifty Geezer]

Financial 06 starts April next year. These'll be approximate times I'm sure. Sony were originally hoping to launch on 65 nm, so it shouldn't be long before they switch process and save some money. If they're fabbing both Cell and RSX and both make the process shrink, that'll probably have quite an impact.

Shame they can't process shrink the BluRay drive too!

 

[avaya]

Did anybody else notice that in 2006 the PS2, PSP, and PS3 will be using 65nm tech?!? So will Sony rep some savings from the PS3 in 2006 when do this? Hold didn't l-b say that the 2006 year actually started in April of this year? What going on guys? I'm damn confused. I though the PS3's CELL was going to be 90nm. Can someone explain?

Thanks

Sony's Nagasaki 2 will be coming online for 65nm in 2006.

 

[version]

cell at 4 GHz, rsx with 500million transistors

 

[mckmas8808]

Honestly speaking

Sony's Nagasaki 2 will be coming online for 65nm in 2006.

Then why didn't Merrill Lynch add this into their cost estimation that leaked a couple of days ago? Do you guys think being that 65nm is right around the corner that Sony might price the PS3 around this happening?

 

[Carl B]

Then why didn't Merrill Lynch add this into their cost estimation that leaked a couple of days ago? Do you guys think being that 65nm is right around the corner that Sony might price the PS3 around this happening?

Well I mentioned that in the Merrill thread actually, that Sony would reach 65nm before MS and thus their chip costs go below MS' sooner than the contrary.

Still, I don't think it's worth dragging that thread into here - best to keep this clean and technical.

 

[Powderkeg]

Well I mentioned that in the Merrill thread actually, that Sony would reach 65nm before MS and thus their chip costs go below MS' sooner than the contrary.

TSMC (Who manufactures MS's chips) will be producing consumer level .65nm chips THIS YEAR.

Unless Sony is producing .65nm right now, MS will have access to it first.

 

[Carl B]

TSMC (Who manufactures MS's chips) will be producing consumer level .65nm chips THIS YEAR.

Unless Sony is producing .65nm right now, MS will have access to it first.

Riiiight. And tell me Powderkeg, which company's are going to be putting out 65nm chips this year besides Intel and RAM makers? Plus since MS is on a contract, they don't automatically get upgraded when new processes come around; there has to be renegotiation. Obviously Sony doesn't have any contractual constraints to fab at it's own fabs.

If MS beats Sony to the punch, hey, more power to them.

 

[SentinelQW]

cell at 4 GHz, rsx with 500million transistors

So, Xenon at 8 GHz, Xenos with 1000 million transistors 'minimum'...

 

[one]

TSMC (Who manufactures MS's chips) will be producing consumer level .65nm chips THIS YEAR.

Unless Sony is producing .65nm right now, MS will have access to it first.

You have to take chip size in consideration.

 

[Platon]

Is a process shrink always associated with lower costs? Doesn't atleast initially many times the yields become worse and it takes time until they get the same yield as the process before. And sometimes certain chips don't even make the transition all together...

 

[Powderkeg]

Riiiight. And tell me Powderkeg, which company's are going to be putting out 65nm chips this year besides Intel and RAM makers? Plus since MS is on a contract, they don't automatically get upgraded when new processes come around; there has to be renegotiation. Obviously Sony doesn't have any contractual constraints to fab at it's own fabs.

If MS beats Sony to the punch, hey, more power to them.

I didn't say MS would be using the 65nm process so soon, only that TMSC will be. IF MS choses to go with 65nm, they will have that option before Sony, but I rather suspect you'll find that the process switch doesn't translate into instant profits.

There is an investment that has to be paid off before you'll get a significant price drop, and in Sony's case they have a huge investment while MS is going through TMSC who can spread that cost out across many different chip types.

And as for what chips would be using it, I would like to point out that every modern electronics device of any kind in the world uses one or more chips. Everything from a digital watch or cellphone to a car or radio uses chips, and they aren't all made by RAM makers or Intel either.