Seems like AMD was right to go to SOI.

zephyrprime

Diamond Member
Feb 18, 2001
7,512
2
81
I was just reading about Prescott's power problems and it just occurred to me that that signifies that AMD was correct to go to SOI at the juncture that it did. Then again, maybe not since it seems to have greatly delated the hammer.
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
Do you mean power or heat problems, or both??? What is your proof or reasoning why you think SOI is better then strained silicon process Intel is using??? I mean technically how did you come to realize your statement???
 

Wingznut

Elite Member
Dec 28, 1999
16,968
2
0
Originally posted by: zephyrprime
I was just reading about Prescott's power problems and it just occurred to me that that signifies that AMD was correct to go to SOI at the juncture that it did. Then again, maybe not since it seems to have greatly delated the hammer.
You aren't taking into account that the A64 runs at 2.0ghz, while PSC runs more than 50% faster (clockspeed.) Also, reports are that Hammer's yields are quite low. I'm purely speculating, but if it's true, then SOI could very well be one of the causes.

All that being said, there really isn't a "right" or "wrong" here... Just different ways of doing things. :)
 

MDE

Lifer
Jul 17, 2003
13,199
1
81
There's reports of problems with Prescott, so strained silicon could be one of the causes. But, (gasp) Wingznut has it right,
All that being said, there really isn't a "right" or "wrong" here... Just different ways of doing things. :)
 

zephyrprime

Diamond Member
Feb 18, 2001
7,512
2
81
Do you mean power or heat problems, or both???
power == heat.

SOI is better then strained silicon process Intel is using
I was under the impression that SOI makes a bigger impact on power consumption than strained silicon. I did a little search on google and turned up:

"The current generation of partially depleted SOI technologies result in power saves of 40% to two times than of bulk silicon-based designs." http://216.239.41.104/search?q=cach...power+heat+soi+better+compared&hl=en&ie=UTF-8
http://www.ibm.com/ibm/environment/news/LegacyBrochurePg2.pdf

And

" Intel claims strained silicon leads to a 10-20% growth of the drive current "
http://www.xbitlabs.com/articles/editorial/print/tech-process.html
Which I assume means that such a process would consume only (1/1.1) ~ (1/1.2) times the power of a non strained process.

You aren't taking into account that the A64 runs at 2.0ghz, while PSC runs more than 50% faster (clockspeed.)
Yeah, I forgot about that. You're right. That would make a big difference.

All that being said, there really isn't a "right" or "wrong" here... Just different ways of doing things
I can't agree with this. This is technoly not poetry. Choices made aren't a matter a taste. There are quantitative measures of success.
 

Accord99

Platinum Member
Jul 2, 2001
2,259
172
106
Originally posted by: zephyrprime
Do you mean power or heat problems, or both???
power == heat.
"The current generation of partially depleted SOI technologies result in power saves of 40% to two times than of bulk silicon-based designs." http://216.239.41.104/search?q=cach...power+heat+soi+better+compared&hl=en&ie=UTF-8
http://www.ibm.com/ibm/environment/news/LegacyBrochurePg2.pdf
40% to 2X? Does that mean some SOI chips actually convert ambient heat to electricity?

 

Jeff7181

Lifer
Aug 21, 2002
18,368
11
81
Originally posted by: Accord99
Originally posted by: zephyrprime
Do you mean power or heat problems, or both???
power == heat.
"The current generation of partially depleted SOI technologies result in power saves of 40% to two times than of bulk silicon-based designs." http://216.239.41.104/search?q=cach...power+heat+soi+better+compared&hl=en&ie=UTF-8
http://www.ibm.com/ibm/environment/news/LegacyBrochurePg2.pdf
40% to 2X? Does that mean some SOI chips actually convert ambient heat to electricity?

I took it to mean 40-80%

BTW... I heard Athlon-64's and Opterons run on 1.2 volts... can anyone confirm that?
 

Wingznut

Elite Member
Dec 28, 1999
16,968
2
0
Originally posted by: zephyrprime
All that being said, there really isn't a "right" or "wrong" here... Just different ways of doing things
I can't agree with this. This is technoly not poetry. Choices made aren't a matter a taste. There are quantitative measures of success.
Maybe I should've been more clearer in what I meant...

Of course there are right and wrong ways of building semiconductors. I just wouldn't consider using SOI as being that black and white. It has its upsides and it has its downsides. It's just way too early to call it a success or not.

Also, take note that the "40% to two times than of bulk silicon-based designs" is a very overzealous (imho) statistic that came from IBM. Since they developed much of the technology, they obviously have a vested interest in it becoming a more used technology.

 

Accord99

Platinum Member
Jul 2, 2001
2,259
172
106
Hmm... I'm pretty sure the Opteron runs at 1.5v as well, in fact original Opterons were up to 1.55v. The 1.2v mentioned in the link suggests that its the Hypertransport signals that are at 1.2v.