Please register or login. There are 0 registered and 551 anonymous users currently online. Current bandwidth usage: 326.30 kbit/s December 07 - 04:59am EST 
Hardware Analysis
Forums Product Prices

  Latest Topics 

More >>


  Socket-775, twenty insertions and counting 
  Jun 28, 2004, 07:30am EDT 
By: Sander Sassen

Granted, I've been mocking Intel's new socket for a while now, making fun of the frailty of the needlepoint sized pins that connect Intel's latest processor to the motherboard. In reality we've done more than just mock, we've had a few motherboards in the office for a few weeks now which are undergoing rigorous testing as you read this. One of the things we'll be evaluating is how fragile the new socket is and whether there's truth to the rumor, or common sense impression, that the hundreds of small pins that form the socket will break easily. Intel has actually just stated that twenty - yes, that's 20 - insertions is about the maximum the socket can handle before showing defects. They pitch it in a way that they make it look like the old socket-478 was also specified for that number of insertions. We can tell you from personal experience that a quality socket-478 will easily last over a few hundred insertions and still work flawlessly.

But there's more to the story, as by moving to socket-775 Intel has also conveniently found a way to get rid of one cause of processor returns - broken or bent pins. Now that the processor is void of any pins and is using contact pads instead, the motherboard is the part that will need to be replaced if a pin, or a number of pins, on the socket get bent or broken. I'm quite sure that the motherboard manufacturers won't be too happy about that, and from what we hear from some of them they're considering revising their warranty policies if returns are high due to damaged sockets. That'll make for an interesting situation much like what you'll see with cars nowadays. For example you can change the oil yourself, put on new tires or even replace the air- and oil filter but the moment you start replacing the crank shaft you're in risk of losing the warranty as those replacements are only allowed to be made by the dealer. It wouldn't be too far fetched to say that some retailers might even decline giving you a warranty if they haven't installed the processor in the socket themselves and secured it with a warranty sticker.

Am I bearing a grudge towards Intel? No, I'm not; it is just that I don't think that another socket change is needed especially when the new socket is so fragile and doesn't have any benefits over the old one. The move from socket-7 Pentium to slot-1 Pentium II was logical given the fact that cache chips couldn't be included on the die at a reasonable price. The move back to socket-370 with Pentium III was also understandable as cache chips were old-fashioned and slow compared to on-die cache which had now also become affordable. The move to socket-423 was also expected - a new architecture that isn't compatible with the old one should not be based on the same socket as people might just exchange parts that cannot be exchanged. Intel should've thought socket-423 through a little further though as shortly after its introduction socket-478 arrived which is now to be replaced by socket-775.

Where I'm getting at with this story is the simple fact that the only technical requirement I can see for the new sockets stems from Prescott's excessive power requirements - it could very well be that socket-478 has a insufficient number of pins to route enough power to the hungry processor core. Again the fact that Prescott is not living up to expectations now pushes Intel in a certain direction. Intel however also wants to make sure Prescott sells well; development costs for 90nm have been substantial, regardless of its problems, and thus introduces a new socket for their new chipsets. Why? Simply to make sure that no older processors, or the Celeron for that matter, can be used with them.

So apparently the move to this new socket is not purely for technical benefits or preparing for higher clockspeeds as Intel has recently stated that they will be focusing on dual core and not high clockspeeds from now on. This move to a new socket seems to be motivated by economical reasons mostly; Intel simply wants to make sure people buy Prescott regardless of its problems and move away from socket-478 as quickly as possible and it does this by introducing a whole new platform based off of it. They expect all of the industry to follow suit like they've done all along and thus for Prescott and socket-775 to become a success. In the meanwhile we'll be continuing our evaluation of the new socket - we've passed twenty insertions already and we're hoping we'll make it to fifty soon. We'll let you know when we get there and how.

Sander Sassen.


 Last Post 
Re: Socket-775, twenty insertions and counting ad asdf 4 replies Aug 27, 2004, 01:56am EDT


  Voice Your Opinion 
Start New Discussion Topic


  Related Articles 

A weekly newsletter featuring an editorial and a roundup of the latest articles, news and other interesting topics.

Please enter your email address below and click Subscribe.