If you were as excited about Intel’s Sandy Bridge processors as we were earlier this month when they soft-launched, then today’s news will come as a shock (especially if you already bought one of the second-gen Core i5 or Core i7 desktop processors available online; Core i3 still isn’t selling).
In essence, Intel identified a problem with the SATA 3 Gb/s on its Cougar Point chipsets last week. SATA 6 Gb/s ports are unaffected. The issue is hardware-related and requires a silicon-based fix at the metal layer, which of course means that all of the currently-shipping P67- and H67 –based motherboards are affected. It’s severe enough, especially for the enthusiast community most likely to be populating multiple SATA ports and pushing heavier I/O workloads that we’d want to recommend holding off on Sandy Bridge-based builds until boards with a fixed version of the chipset ships out. This will happen within “weeks,” according to Intel, as motherboard vendors will start getting updated core logic in late February for a full volume recovery in April.
Intel’s Steve Smith, vice president and director of PC client operations and enabling at Intel, says that the specific problem occurs over time, and is affected by temperature and voltage. It’s more likely to manifest in configurations with lots of data being moved across the SATA 3 Gb/s ports—that’s why OEMs are encountering a problem now. The company says it would have expected roughly 5% of systems to be affected over a three-year period. That's a serious enough figure to compel Intel to halt shipments and incur a total cost to replace/repair existing systems of $700 million.
If you’re already a P67/H67 owner, the problem relates to connectivity between the SATA ports and hard drives. That link can degrade over time and, in a worst-case scenario, you’ll boot your machine to find attached storage simply isn’t identified at all. None of your data is at risk—anything on the drive already can’t be affected by the link degrading and ultimately failing, after all.
Why wasn’t the issue identified during validation, before Sandy Bridge launched? Intel says Cougar Point did in fact satisfy its validation procedure, and was only caught after more strenuous OEM testing. Should OEMs be the ones to catch problems like this? No. But that’s what happened here. That'll likely have ramifications for the way Intel tests its products in the future.
Unfortunately, motherboard vendors, first, and early adopters, second, are the ones to be most seriously affected here. The motherboard manufacturers are going to have to stop production and wait a month for updated Cougar Point. They’re only getting the news today. Enthusiasts won’t be as immediately affected. Boards that shipped out already, in most cases, carry a three-year warranty, offering some form of protection. Sandy Bridge notebooks haven’t shipped out in volume. And Z68 won’t be delayed, Intel says. Everyone else: you’ll want to wait until “fixed” boards start shipping in March/April.
Stay on the Cutting Edge
Join the experts who read Tom's Hardware for the inside track on enthusiast PC tech news — and have for over 25 years. We'll send breaking news and in-depth reviews of CPUs, GPUs, AI, maker hardware and more straight to your inbox.
New Chinese GPU arrives to challenge Nvidia's AI dominance but falls woefully short - Loongson unveils LG200 GPGPU, up to 1 Tflops of performance per node
Yes, you can have too many CPU cores - Ampere's 192-core chips break ARM64 Linux kernel in two-socket systems, company requests higher core count support
it should have been identified by more rigorous testing.Reply
that said, at least they aren't just going to ship it anyway. good for them for repairing the error as best they can at this point.
Wow Intel is really messing up lately. First the P55 had the cpu sockets burning out and now the P67 has this problem. That's what happens when you try to release too soon. Hopefully AMD learned their lesson with Phenom I.Reply
oh boy, this is going to attract amd fanboys like a bear to honey. anyway atleast they caught this early, through not early enough some would say.Reply
Very good for them for nipping this in the bud now. Yes - I'm sure their testing for future products will be a LOT more stringent. But with that - and them readily stopping and fixing this problem will put a little more trust in future Intel products. (We've all seen what happens when a hardware manufacture tries to "hide" flaws....)Reply
Just for clarification - this is with the MB Chipset - NOT with SB CPU's themselves.
This is bad, really bad. It means that every single Sandy Bridge motherboard so far is defective and has to be replaced (since chipsets are soldered, and short of extremely labor intensive manual rework, there is no way to replace them). And motherboard manufacturers are a real pain to deal with; good luck finding someone who can even speak English, let alone understand this recall.Reply
Worse, how will we know if a board we buy in the future has the fixed chipset? The board manufacturers aren't going to eat the losses, so I expect the defective parts already in the pipeline to continue being used until they run out. That could take the better part of a year.
Looks like I'll have to wait until 2012 to upgrade to the i5-2500K.
That's unfortunate... I don't think I've seen this issue yet, but I have my 4 data drives on the 3GB/s ports. At least the data itself is supposedly *safe*.Reply
Will they start calling the updated silicon P67A so there's a way to know which Cougar Point generation is part of the boards? Depending on how the manufacturers do it... they may or may not bump the PCB revision... since the replacement CP chip will be pin compatible... so probably no new layout. This could turn into a minor nightmare... unless they're very good about recalling bad stock, and paying cross-shipping fees, etc...
Those with SB parts will probably have to live with the bug for a month or so before the fixes come down.
Those are the breaks for those who adopt early. Hopefully GB, ASUS, others will offer replacements for those early adopters - but even if they do, what a hassle. I would just return the board and wait a couple weeks for replacements to arrive. Who wants something that "might" have a defect that is so bad they stop production.Reply
But again, I wouldn't have been eager to buy the gimped chipset in the first place. LGA1366 and it's replacement for the win.
Wow this is a serious issue, all those people that bought into Intels promise of a good CPU will be disappointed, with all the money Intel is making, you would of tough that they would take the time to test there CPU , I guess we all know know where is Intel cutting the corners to make that extra profit, well at least we have AMDReply
The one question I have is let's say my mobo has 4 SATA ports and I am only using 2 of them. When one of the ports goes bad, can't I just swap my HD to a different one?Reply
Also, while I really wish Intel would have caught this sooner, I have a hard time believing anyone that already purchased an x67 mobo will still be using it in 3 years. At least Intel didn't do the J&J thing like they did with the stealth recall of Tylenol.
Josh GThis is bad, really bad. It means that every single Sandy Bridge motherboard so far is defective and has to be replaced (since chipsets are soldered, and short of extremely labor intensive manual rework, there is no way to replace them). And motherboard manufacturers are a real pain to deal with; good luck finding someone who can even speak English, let alone understand this recall.Worse, how will we know if a board we buy in the future has the fixed chipset? The board manufacturers aren't going to eat the losses, so I expect the defective parts already in the pipeline to continue being used until they run out. That could take the better part of a year.Looks like I'll have to wait until 2012 to upgrade to the i5-2500K.I would be pretty sure that part of Intel's $700M losses are eating the chips already in the OEMs hands... so there will be no 'old stock' (chipsets) making it to 'new' boards. It sounds like the 'old' mobos already in end user hands should be RMA'd. I'm hoping they eat the delivery costs too... or make it so I can get an immediate exchange at Micro Center... For anyone around at retail stores... have they cleared the shelves of LGA1155 boards?Reply