Slashdot Mirror
Intel Resumes Shipping of Faulty Sandy Bridge Chip
arcticstoat writes "After causing chaos among motherboard makers by revealing a flaw in its 6-series motherboard chipsets, Intel has announced plans to recommence shipments of the faulty silicon, before the fixed chips have even started shipping. Intel claims it decided to start reshipping the chipsets after lengthy discussions with computer manufacturers. "As a result of these discussions and specific requests from computer makers,' says the company, 'Intel is resuming shipments of the Intel 6-series chipset for use only in PC system configurations that are not impacted by the design issue." The announcement follows Intel's recent exposure of a well publicised design fault that affects the 3Gbps SATA ports (typically ports 2 to 5) in Intel's P67 and H67 chipsets. As such, we assume that the new systems based on the faulty chipsets will either come with a separate SATA controller card, or that they will only use the two (unaffected) 6Gbps SATA ports provided by the chipset."
This will confuse people and make them wary of Sandy Bridge based machines for years. "Is this box tainted? I don't know, and the manufacturer won't tell me. I guess I'll buy something else." A nice clean break of recalling *all* defective machines and shipping only good silicon would have been better.
I strongly suspect that laptops will be the big one here.
Virtually all laptops, excluding on a few high end workstation/gamer beasts, are 1HDD and (still common; but getting rarer) 1 optical drive. And, in a laptop, there isn't exactly much room to monkey with the shipping configuration...
Given that Intel has held the crown for reasonably high performance at laptop-friendly TDPs, I'm assuming that laptop makers would really like to get their hands on the latest silicon so that their roadmaps stay mostly accurate.
Small form factor and very small form factor desktops may also want in, for the same reasons. If you can only physically fit 2 drives in the case, only having 2 ports isn't a huge issue(Joe Tweaker who wants to put one of those 4-2.5inch-trays-in-one-5.25inch-bay devices in place of the optical drive will have to suffer; but nobody else will care...
It will be more interesting; but less certain, to see if production of standard motherboards resumes. By all accounts, the built-in intel SATA ports are(when working) competitive or better than most outboard ones cheap enough to integrate on a mass-market motherboard, plus they don't eat PCIe lanes. From a design perspective, it'd be easy enough to not solder headers for the faulty ports, and leave people with just the 6GB/s chipset ports and 4-6 provided by a 3rd party controller; but it remains to be seen if that will be acceptable to enthusiasts...
when companies did this stuff and didn't tell us? When XP hit those upgrade installs were blowing up because the big manufactures stuck bad RAM into Win98 boxes knowing it would never be used (Windows 98 won't used RAM past 256M unless you hack the registry, it'll use the page file instead). Well, the XP install copies the whole disk into RAM before copying it out to disk, so BOOM, there goes your XP install. Usually couldn't recover.
At any rate, this is just great. I'm sure the lower end manufactures will be just pleased as punch to make sure those broken ports don't get used. You know, if it made it into production it must work just well enough to blame the problems on the OS when you call for a warranty swap.
Hi! I make Firefox Plug-ins. Check 'em out @ https://addons.mozilla.org/en-US/firefox/addon/youtube-mp3-podcaster/
If a Laptop uses a faulty chipset, but is only configured to use the two 6GB SATA ports, it will be entirely unaffected by the bug, as it only effects the 3GB SATA ports. Since there is really no way for the consumer to actually use the 3GB ports, it will never have the bug problem.
So yes in cases like that, it makes sense to keep shipping. Those laptops are perfectly fine.
When I read the title I was a bit leery until I thought about it for a second. I know when I buy my new desktop one eventually, I don't want there to be a chance I get a faulty one!
Given that discrete SATA upgrade cards are in the 20$ish range, I'd assume that adding an SATA controller directly to the motherboard would run maybe half that. I assume that any motherboard makers shipping will just leave the faulty ports without the headers soldered on, and tack on a 3rd party SATA controller(something that many were already doing).
Unfortunately, that will(in some ways) be worse and more confusing than the straight crippling. With the chipset ports, basically all motherboards of a given chipset will get the same performance out of those ports. With a 3rd party controller, performance will be substantially variable; based on how many PCIe lanes they give the controller, and who makes it(anybody who remembers the god-awful JMicron[seriously, what is it with JMicron? their IDE controllers sucked ass, and then so did their SSD controller chips...] IDE controllers that some motherboard makers started using when Intel's chipsets went SATA-only should be getting nervous right about now...)
For 1-2 drive only systems, like laptops and very small form factor systems, no problem. The two good chipset ports will do just fine. For motherboards purporting to offer more, though, you'll have to really do your reading before you buy....
Although us geekier types read, "recommence shipments of the faulty silicon," and scream, "Well that's a fine idea of how to get rid of a warehouse of faulty chips!"
Didn't we have this with Intel already, with floating point division? Oh, yeah, here: http://en.wikipedia.org/wiki/Pentium_FDIV_bug .
And Devo did a song about it, years before it happened:
"When chip bug comes along . . . you must ship it! Ship it! Ship it good!"
I wonder if the sales kid at your local super-computer store will inform you, "Oh, by the way, this model has a faulty chip." Or, maybe a sticker on the computer: "Faulty Intel Chip Inside!" That should do wonders for sales.
I remember that once the floating point division problem got mainstream press coverage, folks got all ornery, despite statements from Intel that most users would never see this problem. Most folks don't even know what floating point is. Intel eventually bought off the math prof who discovered the bug, by giving him testing contract. He deserved it, because he did a damn good job tracking down the bug. He is really, "a geek's geek."
Schroedinger's Brexit: The UK is both in and out of the EU at the same time!
This is highly relevant to my interests as I embarked upon an upgrade crusade about a week ago to replace my aging PC
I'm very happy with my four core Phenom II. Powerful, quiet, cheap - pick all three.
Have you got your LWN subscription yet?
Do we really have to keep calling this a Sandy Bridge issue? This isn't a sandy bridge issue, the name Sandy bridge is for the CPU. The issue is NOT with the CPU, it's with the chipset Cougar point. The Sandy Bridge is (so far) perfectly fine, and has no issues at all. Of course, I guess "Intel Resumes Shipping of Faulty Cougar Point chip" doesn't seem as catastrophic.
"replace my aging PC (circa 2008 tech)"
Yeah , 2008 , thats like totally ancient dude. Not.
Christ , no wonder we have an electronics waste mountain and all its associated pollution issues when people like you bin perfectly servicable and upgradable machines.
Who said he's throwing it away? Or even that he's replacing every part of it?
Did you wake up on the wrong side of the bed or something?
What? 2008 is "aging tech"?
I've recently replaced a 2006 processor with 2009 processor (per date stamp on the chip casing itself) - AM2 Athlon64 X2 with AM3 Phenom II 820. It even fit in the same socket of my "aging" 2006 ASUS board.
So what is the point? This isn't 1995 anymore. You are not doubling performance every 2 years, heck, single threaded performance has been about the same for the last 5 years (more or less). 2008 is only 2 years old - today's chips are about the same performance as they used to be, you get more cores now. CPU is a commodity item thees days. For the last few years, it's at the "good enough" and the "wow factor" is gone.
Unless Intel's controller woes are substantially worse than so far disclosed, and somebody read "LVDS" as "Large Voltage Differential Signalling", I'm guessing that Joe Tweaker is sad; but safe enough...