what's with socket chips

[skimzzz]

I just upgraded to a 1.2 mHz 266 FSB thunderbird socket A from a 650 mHz slot A CPU. I'm looking at the core (where all of the heat is supposed to flow to the HSF) and it is tiny.

Why did Intel and AMD move from the slot design to the socket design when faster clock speeds mean cooling is more crucial (and a larger surface area like the slots are better)? I hope it's not just to cut costs.

 

[Kelledin]

Mainly just that the slot design is extremely unwieldly. It doesn't fit on SBCs very well, or in laptops, and it makes it very difficult to line up multiple CPU slots without having at least one CPU's HSF blocked. The fact that it's more expensive--more materials, greater shipping costs, greater storage costs, etc--also has a lot to do with it.

Kelledin
<A HREF="http://www.linuxfromscratch.org/" target="_new">LFS</A>: "You don't eat or sleep or mow the lawn; you just hack your distro all day long."

 

[OldBear]

To add to what my colleague from south of our Red River stated; smaller die size means less core voltage, supposedly cooler temps and perhaps cheaper production costs.

<font color=blue>Remember. You get what you pay for. All advice here is free.</font color=blue>

 

[Crashman]

Mostly to cut cost, also to cut complexity and mounting height, also a Socket doesn't have the long conductor of the Slot which can cause RFI. The only reason they used Slots in the first place is for a handy place to mount off-die cache, cache in on-die now.

What's the frequency, Kenneth?

 

[girish]

The main purpose of introducing slot was to keep off the competition, particularly AMD and Cyrix which made processors compatible with intel sockets. That is what was told when they introduced the Klamath. Then they said it was to keep the heat off the motherboard and make cooling better, and a nice place to put L2 cache on board.

This was particulary nasty for us Indians who enjoyed a 5% import duty even on these P-IIs, until intel stripped of the plastic housing on the Celerons and custom officials at New Delhi airport spotted the Celeron PCB and then, classified it as a PPCB (Populated/Preassembled Printed Circuit Board) rather than a Semiconductor Package and ramped the duty to 22%!!! We Indians in particular did not have too good a time with the slot! ;-)

Anyway, I remember having written somewhere back in 1997 when they introduced the PII that eventually intel will fall back to sockets, and they have!

Now due to improved manufacturing processes it is now possible to integrate the cache ondie itself, and reduce the conductor lengths so as to achieve higher FSB. And still, they continue to keep competition at bay by introducing a new socket every year! ;-)))

girish

<font color=red>No system is fool-proof. Fools are Ingenious!</font color=red>

 

[lhgpoobaa]

and given the increasing complexity of CPU's, increasing thermal output requiring more extensive HSF's, and shorter trace paths, the socket really was the only way to go.

a slot wouldnt stand a chance of mounting a P4 or large socketA cooler.

i forsee heatsinks becomming bigger. things like the 80mm x 80mm socketA monsters becomming the norm, and the socketA chipset or its sucessor having the 4 holes in the motherboard being used as standard or some mechanism like the p4

Excuse me for a moment. I need to drive my ergonomic wheely chair over a sheet of bubble wrap!

 

[girish]

well, it seems so!

some socketA boards are already having 4 holes around the socket to mount more powerful HSFs.

The microPGA478 P4 was one attempt to shorten the wires from the pins to die which will assist in increasing the FSB. Smaller fabrication technologies will make them smaller and consume less power which will in fact make today's bulky HSFs history.

But as it stands, the HSFs are becoming bigger and bigger with smaller processors! Now air alone isnt enough! Probably tiny refridgeration units will cool these hot chips sometime in the future!

girish

<font color=red>No system is fool-proof. Fools are Ingenious!</font color=red>