But how safe is it? Intel doesn’t really do ‘Total Recalls’ like this. It’s scary stuff. The company that effectively made the global market for microprocessors is such a galactically huge corporation, with advanced R&D the like of which has never been seen before, that no one ever really expected it to get knocked off track. Not even for a quarter. That’s what happened, so what can be done now?
Intel’s processors work perfectly. The company guarantees that this is true. However, every CPU needs to sit on a mainboard. Apart from providing comfortable living accommodation, the mainboard provides the meeting point for the CPU to link up with the other parts of your system.
That mainboard is, pretty much, divided down the middle into slow and fast. On the very quick side of the board, you have memory and graphics. Over on the slow side, you have hard drive controllers. Looking at the hard drive controllers, you get a further break down into slow(er) and fast(er). While many old kit gurus will remember transfer rates in the Mb per second range, those speeds are now so slow that we only use them for the Internet (cue the creaking of the rocking chair and shuffle of slippers).
All H67 and P67 boards have two sets of SATA ports. Four of these support 3 Gbps operation and two support the faster 6Gbps speeds. Each of these sets requires its own PLL source. This problem in the chipset has been traced back to a transistor in the 3Gbps PLL clocking tree. This transistor has a very thin gate oxide which allows you to turn it on with a very low voltage. The mistake that Intel made was biasing the transistor with too high of a voltage which caused a higher than expected leakage current. Each transistor will operate slightly differently, so over time the leakage current can cause a failure on the 3Gbps ports. Luckily the fact that the 3Gbps and 6Gbps circuits have their own independent clocking trees helps ensure that the problem will be only limited to ports 2-5 of the controller.
OK. Now we know where the problem is – we need to ask “How much of an issue is it ?”
.For Intel, the answer is simple. The product is outside the specification which they set, which means it is not correct and must be cancelled/returned/buried in a pit and forgotten about.
But the real world does not work that way. IT is a business and companies need to sell. Checking around, no one in the UK channel seems to have missed much of a beat.
So the real question is, “Can you safely sell/buy/use a Sandy Bridge system?”.
The answer is, quite obviously ‘Yes – as long as you are aware of the issues and you’re sure they won’t affect you’. Especially if you choose a mainboard from one of the top vendors like Asus.
KitGuru has no specific brand preference when it comes to mainboards, but what looked like ‘Asus overkill’ a month ago, may well turn out to be a life jacket for the channel over the next few months as Intel scrambles to re-build its reputation.
Here’s a question: In an average system, how many SATA devices does the normal user have?
Here’s the KitGuru reply: Booting from an SSD and then loading programs onto a big system drive from a DVD/BluRay is about as many attachments as the majority of people will make. Add in a second hard drive (either internally – or through an eSATA port on the front of your system) and you have solution that probably works for >95% of the world’s users.
That means as long as you use the correct 4 headers, you should not only enjoy carefree data transfers for the life of your system – those transfers will also be as fast as (presently) mechanically possible.
This image is being passed around the industry right now and it helpfully shows you where up to 4 drives can be attached on an Asus board with no risk whatsoever.
So there you have it. As long as your system builder, local store or KitGuru knows what they are doing – you can build a perfectly good Sandy Bridge system, without ever experiencing an issue.
AMD will, no doubt, be creating some new, faster Phenom II X6 SKUs for launch in the next 4-8 weeks and they also have Bulldozer coming soon. Competition is a good thing, so we hope AMD gets these new launches right. Them’s your options.
KitGuru says: If you really want to buy a Sandy Bridge product before it has been fixed by Intel, then that’s your call. You will be paying full price for a board which has a documented problem and which the original chipset manufacturer has recalled. That said, deals seem to have been reached with Asus, Gigabyte, MSI and Intel that say the boards can be sold and, if you like, swapped out later.
AMD also has solutions – fully functional and guaranteed.
Choosing a pre-assembled system from a ‘known’ company who offers a full warranty, is a smart way to go. In most cases, the cost of buying assembled will be close to buying yourself and building. With pre-assembly, you get to blame someone else if it goes wrong (although your data is still your responsibility).
If it’s Intel you really want, KitGuru suggests waiting for the fixed boards. But at least you know the risks of buying sooner.
Comments below please – or unload in the KitGuru forum.Working around Sandy Bridge problem,