Intel's Sandy Bridge chipset problem explained

Call it naivety but one expected Intel to be a little more communicative with its customers. In the light of the announcement that the company has found flaws in its new 6 series chipsets we hit the phones, asking around a variety of manufacturers, distributors and retailers to find out just what the plan was to deal with the admittedly faulty products being sold.

It is pretty clear that no one had warning that this was coming. The motherboard makers we have talked to are still waiting from some sort of response from headquarters in Taiwan – delayed by the combination of different time zones and Chinese New Year.

This of course means that the distributors and retailers of affected products are also in the dark. Major retailers are still listing motherboards and laptops based upon the Cougar Point chipsets.

How this could affect users
One could argue that because the problem with Serial ATA is one of long term degradation that there will be no one effected for now.  In the press release Intel points out that “consumers can continue to use their systems with confidence, while working with their computer manufacturer for a permanent solution”.

But this permanent solution will invariably involve the replacement of a motherboard, which isn’t likely to be a quick process. In a conference call explaining the move overnight Intel said that it was working with OEMs on replacement, and that the US$700 million mentioned will be spent replacing chipsets and motherboards.

One other thing that the conference call outlined, and the press release didn’t, was the sheer speed at which the issue was identified and fixed. The chipsets had passed Intel’s internal testing with no problems, but when OEMs underwent stress testing of the chipset the problem emerged. This kind of testing involves subjecting motherboards, systems and laptops to high temperature and increased voltages. Under these conditions (which simulate long term wear)  failures were observed.

When this happens OEMs send stuff back so Intel can investigate. In this case Intel engineers identified the issue early last week. They then spent several days testing and investigating to work out if it was indeed a silicon issue and if it would actually be a problem for end users. The decision to pull the current version of the chipset and ship a fixed one was made yesterday, and Intel started talking to its OEMs then.

What went wrong, in a nutshell
Since then the specifics of the problem have been outlined on Anandtech. For those without a working knowledge of semiconductor fabrication the TLDR version is that Intel built a circuit out of metal that required low voltage, then ended up supplying too much voltage. This meant that the circuit would be subject to leakage, where power seeps out from the circuit, effecting surrounding circuits. This problem just gets worse over time, which in turn causes the predicted failures.

Reengineering such a small part of the chipset would cause major issues, so Intel’s fix is to disable the problematic circuit completely. This won’t impact performance, but it will stop the problem dead in its tracks.

Despite the fact that the problem is now documented and public, there has been very little time for OEMs to be contacted and formulate responses. Which is why we are hearing ‘we’ll get back to you when we hear from HQ’ from every laptop and motherboard manufacturer contacted. It also explains why distributors and retailers are in the dark over this.

We suspect that Intel employees are hitting the phones pretty hard right about now in order to get the word out. It is great to discover that Intel jumped on this problem so quickly. But it would be nice to be able to advise something beyond holding off on buying a Sandy Bridge based system for now.