[jqp]

Intel allowed a new chipset onto the market with a small flaw, now it will lose an estimated $1Billion.

The new Sandy Bridge Architecture core-i chips were released reccently (core i-5 and i-7, the ones in the new MacBook Pros). I was looking forward to building a desktop PC with one of these chips because of the Quick Sync technology which really speeds up video transcoding. Also the new chips are faster than the previous generation core-i chips.

The chipset with the flaw was for use with these new processor chips, and is mounted on the motherboards that have sockets for the new Intel processors.

Motherboards for the new processors appeared on the market in January, priced normally in the $100-$300 range for various features, and were also manufactured by OEMs for the PCs they would supply.

Then this flaw was discovered that affected the SATA 2 ports on the otherboards - if you have a current hard drive it is likely a SATA 2 drive. What happens is that the drive will transfer data, then after a while it will bog down, and eventually disconnect! If you use a SATA 3 port you were not affected, but not many SATA 3 drives (6Gb/sec) are in use. Or you could just plug your SATA 2 drives only into the SATA 3 ports (2 SAta 3 ports supported by the chipset, 4 SATA 2 ports) and waste the SATA 2 ports...but that will not fly...

Bottom line is that ALL the P67 and H67 chipsets and ALL the motherboards that already have them mounted have to be recalled. I went to look the other day and all the motherboards are gone from the internet vendors!

I think it will be 750 million to retool and remanufacture, and 250 million to process the recalls (or is it the reverse?).

And the flaw was a simple oversight iin the engineering process! Intel's revenue I think was 8 Billion last year?

[JerryM]

If you're going to screw up bad enough to lose your job, make sure to go down in flames. 

What a huge f@#k up. At least the unemployed guy has a cool story to tell. 

Have fun,
Jerry

[Pez]

that's unreal. It makes me sick just to think about. This is one of those disasters that should have never happened.

[JLM]

Most engineering screw ups look simple in hindsight.  It's usually due to the human "need for speed" (gotta have it yesterday and it's gotta be better, slicker, faster, funner than snot).

This kind of thing is no big deal to me (it's only money and inconvenience).  The recent collapse of a concrete parking deck while under construction in Texas (that somehow only killed two) is the kind of thing that gets my attention.  We'll probably find out it was due to usually cold weather slowing the curing of the concrete (as they keep building higher levels while the ones below are still curing and some still have the forms in place).

[jqp]

Pez - I am sure there were a lot of sick engineers and managers. At least they seem to be doing the right thing by completely removing the issue - in the past AMD had a flaw in their processor and they ended up not recalling the chip - it stayed in the market but was sold as a lower performing chip. In this case the processors are fine, it is the supporting chipset that is flawed - but is still costs them $1 Billion! You can buy ( and many did) the processor but you will have no motherboard to put the processor on for a while!

JerryM - Parking garages these days seem to be seem to have very little tolerance for error...Part of the floor came down on the cars below when a car crashed into the outer wall in one shopping center here.

In this case the flaw was in 1 tiny transistor on the 3Gbps PLL clocking tree. There is a current leak because of a voltage being biased too high for that transistor. It apparently took them a few days to completely assess and confirm the implications, that disk transfer errors and eventual drive disconnections would occur. The fix means changing the gate oxide layer on the transistor, which means a new stepping for the chip (new version).

To make matters worse, due to regulatory requirements, Intel could not notify its OEMs until after it abruptly announced the recall to the public at the end of January! This was after all the reviews of the chipset were very favorable and many purchasing decisions had been made!

As someone said, at least they didn't say "you're not holding it the right way".

[audioengr]

Intel does massive simulations of all chips, but I think a deficiency is system simulation.  Computer I/O systems have become very complex and they need to improve this area.  It's a lot easier to model a chip than a system of several chips, and then you need to create the stimulus that tries every conceivable scenerio.  Mind boggling really.

Steve N.
Design Team Lead for Pentium 2

[rob-fi]

Maybe AMD can take advantage of this? 

[jqp]

Maybe AMD can take advantage of this? 

That's what AMD fans are hoping But the new Intel CPUs (the problem was with the Intel P67 and H67 supporting chipsets) have set the bar VERY high. I am looking forward to the future Z67 and other chipsets and CPUs coming out later this year.

Really if you are not a desktop enthusiast, and an overclocking hot-rodder at that, this probably does not affect you much. If you use only the new SATA6 ports you would not really be affected.