AMD mobile S1 socket features dual-processor support

[YKhan]

As was discussed in a previous thread, AMD's Socket S1 for laptops not
only reduces the number of pins from 754 to 638, it also adds
dual-channel memory vs. single-channel. Now it looks like it also has
support for dual-processors, which I was not aware of before! This
article also contains a preview of the upcoming Bulldozer redesign of
next-generation Turion.

AMD Turion 64 X2: A Closer Look - LaptopLogic.com
"Amazingly, Socket S1 features a reduced pin count compared to the
original Turion 64 socket - 638 vs. 754 - while adding dual channel
DDR2 and dual processor support. AMD was able to accomplish this feat
mainly by reducing the number of power and ground pins. "
http://www.laptoplogic.com/resource...=3&PHPSESSID=f5c4f18bd0ab8dceb2243f3c2256703f

Yousuf Khan

 

[Tony Hill]

As was discussed in a previous thread, AMD's Socket S1 for laptops not
only reduces the number of pins from 754 to 638, it also adds
dual-channel memory vs. single-channel. Now it looks like it also has
support for dual-processors, which I was not aware of before! This
article also contains a preview of the upcoming Bulldozer redesign of
next-generation Turion.

By "dual-processor" support they are only talking about dual-core, not
two sockets. The product brief for these chips is up at AMD's site
now:

http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/41407.pdf


Of note, there is only a single Hypertransport link. So unless AMD
has some method of splitting these HT links, they're single-socket
chips only. Really that only makes sense anyway, what would be the
point in a dual-socket/dual-core laptop?

 

[YKhan]

By "dual-processor" support they are only talking about dual-core, not
two sockets.

It's possible, but why would they get dual-core mixed up with
dual-processor? Laptoplogic has been around for a bit of time now, they
should know the difference.

The product brief for these chips is up at AMD's site
now:

http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/41407.pdf


Of note, there is only a single Hypertransport link. So unless AMD
has some method of splitting these HT links, they're single-socket
chips only. Really that only makes sense anyway, what would be the
point in a dual-socket/dual-core laptop?

I think we can safely ignore AMD's white papers here, there's lots of
hidden features in Socket AM2 that aren't listed in their white paper.
They've been revealed by the 4x4 initiative. I think AMD is probably
still hiding tons of stuff in Socket S1. Most of those power and ground
pins are very likely hidden features.

Yousuf Khan

 

[Tony Hill]

It's possible, but why would they get dual-core mixed up with
dual-processor? Laptoplogic has been around for a bit of time now, they
should know the difference.

Just because they *should* know the difference doesn't mean that they
DO know the difference. Apple *SHOULD* know the difference between
two dual-core processors and a "Quad Xeon", but their website shows
that if they do, they obviously don't care about making such a
distinction.

I think we can safely ignore AMD's white papers here, there's lots of
hidden features in Socket AM2 that aren't listed in their white paper.
They've been revealed by the 4x4 initiative. I think AMD is probably
still hiding tons of stuff in Socket S1. Most of those power and ground
pins are very likely hidden features.

I *HIGHLY* doubt that. AMD obviously put a MAJOR effort into
minimizing power and ground pin requirements for this chip, and for
good reason, each pin costs money. Why go and waste all that savings
by having unused hidden features? If you aren't going to sell it, why
spend the money to make it?

If I were you I would stop counting on extra features from either
socket S1 or AM2. Companies just don't go around wasting money unless
they've got a real good reason to do so. As for 4x4, until we get ANY
kind of meaningful tech info, I wouldn't go around making any
assumptions about it just yet.

 

[nobody]

Just because they *should* know the difference doesn't mean that they
DO know the difference. Apple *SHOULD* know the difference between
two dual-core processors and a "Quad Xeon", but their website shows
that if they do, they obviously don't care about making such a
distinction.

Tony,
You forgot *who* is the targeted audience. Most Mac users *can't*
tell 'the difference between two dual-core processors and a "Quad
Xeon"'. They will swallow any marketing bull$hit Steve decides to
dump at them. They believed into superiority of G4/G5 until the very
moment Steve declared previously inferior Pentium to be the most
superior CPU on Earth, even though back then Intel badly lagged AMD in
both price (usually not of concern for Mac users) and performance.
;-)
NNN

 

[Yousuf Khan]

Just because they *should* know the difference doesn't mean that they
DO know the difference. Apple *SHOULD* know the difference between
two dual-core processors and a "Quad Xeon", but their website shows
that if they do, they obviously don't care about making such a
distinction.

Fooling a typical Apple user is not the gold standard to go by.

They've been under the cloud of the reality distortion field right from
the beginning. They'll buy anything if Steve Jobs sells it.

I *HIGHLY* doubt that. AMD obviously put a MAJOR effort into
minimizing power and ground pin requirements for this chip, and for
good reason, each pin costs money. Why go and waste all that savings
by having unused hidden features? If you aren't going to sell it, why
spend the money to make it?

Because quite obviously it doesn't cost *that* much money to put extra
pins in. They're already in the multiple hundreds of pins range, putting
more pins is not going add that much to the cost of the package. Also
it's not going to cost that much to route those extra pins within the
motherboard as most of them are simply hanging off the same power or
ground rails. Very simple routing in that case.

And of course the payoff of keeping extra pins is for future expansion
without having to change the infrastructure.

If I were you I would stop counting on extra features from either
socket S1 or AM2. Companies just don't go around wasting money unless
they've got a real good reason to do so. As for 4x4, until we get ANY
kind of meaningful tech info, I wouldn't go around making any
assumptions about it just yet.

What other meaningful tech info do you need? AMD said 4x4 is based on
Socket AM2. AM2 was originally classified as a single-socket-only
package, and now it's classified as a dual-socket package without making
any changes to the pin definitions whatsoever -- i.e. hidden features!

Yousuf Khan

 

[Tony Hill]

Fooling a typical Apple user is not the gold standard to go by.

They've been under the cloud of the reality distortion field right from
the beginning. They'll buy anything if Steve Jobs sells it.

True enough. But if a major corporation can (intentionally) mix these
things up, what's to stop some small website from doing the same?

Because quite obviously it doesn't cost *that* much money to put extra
pins in. They're already in the multiple hundreds of pins range, putting
more pins is not going add that much to the cost of the package. Also
it's not going to cost that much to route those extra pins within the
motherboard as most of them are simply hanging off the same power or
ground rails. Very simple routing in that case.

It doesn't cost that much more, but the profits are pretty darn thin
to begin with.

And of course the payoff of keeping extra pins is for future expansion
without having to change the infrastructure.

That might fly for servers, but not for laptops. Changing the
infrastructure of laptops is not nearly as big of a problem or as
costly.

What other meaningful tech info do you need? AMD said 4x4 is based on
Socket AM2. AM2 was originally classified as a single-socket-only
package, and now it's classified as a dual-socket package without making
any changes to the pin definitions whatsoever -- i.e. hidden features!

What tech-info I need is some indication of just HOW this is being
done, of which there is none. There are no (publicly available) tech
docs for Socket AM2 so none of us have a clue as to what the pin-out
is.