Print 48 comment(s) - last by ilkhan.. on Dec 3 at 12:31 PM

Intel partners reveal plans for new processor socket designs

In a memo copied to DailyTech, Intel partners recently discussed details regarding Intel's next-generation socket designs.

Intel's next-generation processor family, codenamed Nehalem, integrates the memory controller directly onto the processor die -- a feature already standard on AMD's K8 and K10 core architecture. 

Where companies traditionally increase pin count for new processor designs, Intel's LGA715 (also dubbed Socket H) will actually decrease the amount of pins from 775 to 715.  Since the memory controller will reside on the processor, Nehalem processors no longer need the additional signaling from the processor to the Northbridge. 

Guidance released to Japanese PC Watch claims this new desktop socket will actually utilize a 1160-pin LGA1160 design instead. Intel officials would not reveal exact pin count details.

LGA1366, on the other hand, will greatly increase the pin-count for cross-CPU communication via Intel's QuickPath Interface.  AMD increased its pin-count for server chipsets when it migrated from PGA940 to LGA1207 design last year.

Server Nehalem processors will use Registered DDR3 memory; desktop processors will utilize the unregistered variant.   While not electrically compatible with DDR2, DDR3 still uses 240 pins for signally thus eliminating the need to increase pins on account of the system memory.

As one Intel engineer who agreed to speak on conditions of anonymity put it, "We try to reduce the pin counts as much as possible to eliminate cross talk and other interference." He adds, "But we do try to leave some pins for overhead and future use."

Nehalem-based CPUs will use Intel's second-generation land grid array (LGA) design.  The use of "pins" in context of the land grid array is a bit of a misnomer as the processor interfaces with the socket design via pads rather than pins.  This LGA design is recognized by both AMD and Intel for its ability to increase "pin" density and durability.  

Corporate roadmaps from Tyan and Supermicro both detail LGA1366 designs for sampling by the end of Q2 2008.  Desktop LGA715 variants, on the other hand, won't see mass production until the second half of 2008, with a target launch of Q4 2008.

Intel guidance slates LGA1366 Tylersburg chipsets for a Q3 2008 launch.  Desktop Havendale and Lynnfield chipsets using the LGA715 design are on the record for Q4 2008.

Comments     Threshold

This article is over a month old, voting and posting comments is disabled

please change your socket design!
By Etern205 on 11/28/2007 10:47:59 PM , Rating: 2
During the days when pins we're located on the processor, users are advised to handle them with great care because if a pin is bent (it's really hard to straighten it back out even with the lead pencil trick) or broken off, then your toast.

When the LGA cpus came out I was extremely happy that the cpu became pin-less, but I had never thought that the pins are now built into the socket. And here lies the same exact problem only this time it's the other way around. This I find it annoying because lets say you got a board online and when it arrived you notice there is a bent or damage pin on the socket. Now when you try to return it, the company may denied it because they think all of their products shipped out are free of defects and you physically damaged it yourself.

Since Intel is coming out with a new socket then it would be greatful if they can redesign the type of socket design instead of using this pressure point type system.

I was thinking about something like this, a socket where its pins are made out in a solid cylinders or solid pyramids with its top cut off so it's flat.

Another is with the speed of PCIe 2.0, would it help if they can create a socket to use those like the design Intel did with their slot 1 used for thier Pentium 2 and some of their Pentium 3s?

By DerwenArtos12 on 11/29/2007 6:07:28 AM , Rating: 4
As for creating a solid pin on the socket it would create a lot of problems with seating the CPU properly not when first installed but, when being held in place by whatever retention system, the current designs squeeze the packaging against the socket and the pins stay suspended. Having a hard surface would be much more stressful on the packaging and i would almost guarantee the package would warp.

Slot designs are simply inefficient, both thermally and cost-wise. Thermally because they tend to create a dead-spot for air and because you simply cannot secure the card well enough to attach a heat sink large enough. cost wise because a socket design is much smaller, requiring less silicon and because with the number of leads continually going up the cards would have to be MASSIVE to allow for enough leads. For comparison, PCI-E x16 has just over 80 total leads, imagine trying to make that 715 or 775 or 940 and that not even server processors, how about 1300 leads for server CPU's. Just completely infeasible.

"When an individual makes a copy of a song for himself, I suppose we can say he stole a song." -- Sony BMG attorney Jennifer Pariser
Related Articles

Most Popular ArticlesAre you ready for this ? HyperDrive Aircraft
September 24, 2016, 9:29 AM
Leaked – Samsung S8 is a Dream and a Dream 2
September 25, 2016, 8:00 AM
Inspiron Laptops & 2-in-1 PCs
September 25, 2016, 9:00 AM
Snapchat’s New Sunglasses are a Spectacle – No Pun Intended
September 24, 2016, 9:02 AM
Walmart may get "Robot Shopping Carts?"
September 17, 2016, 6:01 AM

Copyright 2016 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki