Intel Chipset Design Error!!!

Basically, the simplified explanation is that there are 2 disk drive controllers on the P67/H67 motherboards, controlling a total of 6 "SATA" disk drive ports (usable for Hard Drives, Solid State Drives, and CD/DVD Drives).

2 of the ports are higher speed SATA-III (6 Gigabit/second) and are unaffected by this issue.

4 of the ports are on lower speed SATA-II (3 Gigabit/second) and will slowly degrade over time. Degrade in this case being defined as "will show data transfer error and retry, resulting in longer and longer data access times, until eventually the drive on the controller will drop off as a detected device". The DRIVES themselves are not damaged, they can be installed and recognized and used as normal if shifted to a working controller/port.

The fault appears to happen in 5-15% of the affected motherboards (depending on usage) over a 1-3 year period. Or much more often as is being reported by RAID users (where the drives are strung together to make one virtual drive, with a much higher read/write activity quotient).

Official response from various PC vendors has been "Keep using your system, move to a SATA-III port if you can, and once the supply chain catches up around about April, return your motherboard for an exchange".

Unless you're a heavy data I/O user, RAID user, this issue is unlikely to affect you at all in the short term.

(Disclaimer, I am not a Maingear, Intel, Newegg, or other retailer employee, I'm a volunteer tech over at the Newegg customer tech support forums)

There's a very strong knee-jerk return trend being observed in the field, that's why I'm trying to stress that the average user/owner will not certainly encounter this issue during a normal PC service lifetime. This is NOT a formal recall of all parts, but pc part vendors are generally honoring returns in good faith because there are no current parts in the pipeline that are free of this issue. If you have some patience, some technical common sense, and not a little bit of faith, you can make sure you are unaffected by this issue in the short term - and in the long term, all will be made good. Yeah it's a hassle, yeah nobody likes it, but if you have a working system, it's not going to just magically go poof and die on you overnight. You WILL see speed degradation over time as a symptom. You CAN switch ports to work around the issue (a $25 PCIE SATA controller works pretty well too). This is by no means a fatally flawed platform.

Relevant Data: http://news.cnet.com/8301-17938_105-20030052-1.html?tag=river
 
Iching said:
I am returning my bundle to MC. I am glad I did not sell my i5-760 I bought on ebay for $145. SB was a nice experience until now.
Click to expand...

Doing this today? Lmk if they accept your cpu, assuming that you've opened it up already!
 
According to Anand, you'll start seeing SATA errors as an early indicator of failure. Where are those reported? Is there a way to check and see if you are getting errors?
 
CheapAsianGamerGuy said:
Basically, the simplified explanation is that there are 2 disk drive controllers on the P67/H67 motherboards, controlling a total of 6 "SATA" disk drive ports (usable for Hard Drives, Solid State Drives, and CD/DVD Drives).

2 of the ports are higher speed SATA-III (6 Gigabit/second) and are unaffected by this issue.

4 of the ports are on lower speed SATA-II (3 Gigabit/second) and will slowly degrade over time. Degrade in this case being defined as "will show data transfer error and retry, resulting in longer and longer data access times, until eventually the drive on the controller will drop off as a detected device". The DRIVES themselves are not damaged, they can be installed and recognized and used as normal if shifted to a working controller/port.

The fault appears to happen in 5-15% of the affected motherboards (depending on usage) over a 1-3 year period. Or much more often as is being reported by RAID users (where the drives are strung together to make one virtual drive, with a much higher read/write activity quotient).

Official response from various PC vendors has been "Keep using your system, move to a SATA-III port if you can, and once the supply chain catches up around about April, return your motherboard for an exchange".

Unless you're a heavy data I/O user, RAID user, this issue is unlikely to affect you at all in the short term.

(Disclaimer, I am not a Maingear, Intel, Newegg, or other retailer employee, I'm a volunteer tech over at the Newegg customer tech support forums)

There's a very strong knee-jerk return trend being observed in the field, that's why I'm trying to stress that the average user/owner will not certainly encounter this issue during a normal PC service lifetime. This is NOT a formal recall of all parts, but pc part vendors are generally honoring returns in good faith because there are no current parts in the pipeline that are free of this issue. If you have some patience, some technical common sense, and not a little bit of faith, you can make sure you are unaffected by this issue in the short term - and in the long term, all will be made good. Yeah it's a hassle, yeah nobody likes it, but if you have a working system, it's not going to just magically go poof and die on you overnight. You WILL see speed degradation over time as a symptom. You CAN switch ports to work around the issue (a $25 PCIE SATA controller works pretty well too). This is by no means a fatally flawed platform.

Relevant Data: http://news.cnet.com/8301-17938_105-20030052-1.html?tag=river
Click to expand...

Appreciate the summary.
 
So has anyone seen an official comment from Gigabyte, Asus, MSI, etc., regarding what they're going to do?
 
Forceman said:
According to Anand, you'll start seeing SATA errors as an early indicator of failure. Where are those reported? Is there a way to check and see if you are getting errors?
Click to expand...

Like most drive and controller errors, you will notice at the exact moment the one file you absolutely need ASAP gets scrambled and turned to shit :p
 
IsLNdbOi said:
So has anyone seen an official comment from Gigabyte, Asus, MSI, etc., regarding what they're going to do?
Click to expand...

Really for current users this is the most important information.

No hassle/questions asked cross ship/advanced replacement would be the best for the consumer. However if you have to ship your board there first and we are looking at a turnaround time in weeks (if not longer) and having to show that the controller is indeed currently affecting your board, that will be quite frustrating to say the least.

Regarding the add in card solution, each add in card used would affect the number of PCIe lanes available correct?
 
IsLNdbOi said:
So has anyone seen an official comment from Gigabyte, Asus, MSI, etc., regarding what they're going to do?
Click to expand...
No, because they found out about it at the same time we did, and since they're all in HQ'd in Taiwan and were already closed for the day when the news broke they're probably only just now starting to figure out what they're going to do.

I'm sure they'll have lots of questions for Intel that need answering before they go announcing and plans or programs publicly.
 
Filter said:
only 5-15% will even have the issue.
Click to expand...

CheapAsianGamerGuy said:
The fault appears to happen in 5-15% of the affected motherboards (depending on usage) over a 1-3 year period.
Click to expand...

I think we should stop taking Intel's word for it and wait for independent confirmation. Plus these 5-15% and 1-3 years figures were mentioned by Intel before they announced the issue was a faulty thin transistor, it's 100% of all chipsets that were designed this way, not 5-15%. The number of failures reported to Intel also makes that 1-3 years figure extremely suspicious. It looks more like the failure can happen at any time, given the right voltage conditions or other triggers.
 
limitedaccess said:
Regarding the add in card solution, each add in card used would affect the number of PCIe lanes available correct?
Click to expand...
Also it would not likely offer RAID, you couldn't do a 6 drive RAID array, it'd likely have lower performance.

An add in card is not a viable solution IMHO.
 
Filter said:
im not worried about it. people are freaking out over nothing really only 5-15% will even have the issue.
Click to expand...

And how are you supposed to diagnose whether or not you're one of the 15%? Push the port with heavy use until you BECOME one of the 15%? The way they've described the error, the rate will eventually be 100% with heavy use on any of the SATA ports marked 2 through 5 due to the design. Hence the massive recall.
 
Chimel said:
I think we should stop taking Intel's word for it and wait for independent confirmation. Plus these 5-15% and 1-3 years figures were mentioned by Intel before they announced the issue was a faulty thin transistor, it's 100% of all chipsets that were designed this way, not 5-15%. The number of failures reported to Intel also makes that 1-3 years figure extremely suspicious. It looks more like the failure can happen at any time, given the right voltage conditions or other triggers.
Click to expand...
It's statistical analysis. There are manufacturing tolerances. That means that there's only a certain percentage of them that are vulnerable to failing in say 3 years under say a worst case usage pattern. The others will simply fail in a time longer than 3 years under that same worst case usage pattern and are omitted. Then factor in that most people won't have a worst case usage pattern. Factor in temperature and other conditions and you can get a 5-15% failure number even though 100% are susceptible to the failure.
 
I am a little curious to see what happens with the MC deal. They do not except returns on CPU's

Non-returnable and non-refundable items
Special buy or clearance items are not returnable unless otherwise indicated on the label attached to the product. Service parts, motherboards and processors may only be exchanged, and only if defective, within 30 days of purchase if in "like-new" condition. Labor, installation services and any shipping charges are non-refundable.
 
The right triggers are 24/7 high I/O traffic in a higher than average temperature higher than average voltage environment (how Intel artificially reproduced the issue). That worked out, roughly, to equate to 5% hardware failure rate over a simulated 1 year wear down, and a 15% failure rate in a simulated 3 year wear down of a PC at "average use".

If you're a 24/7 high volume user constantly thrashing disk I/O on a RAID setup, you'll likely encounter RAID errors in the short term.

Single disk users on normal data I/O loads follow the formula above as stated.

This issue has not been proven to "scramble data", the disk I/O starts lagging over time until the disk is no longer available, THAT IS THE SYMPTOM.

My previous C.Net link contains (updated as they respond) responses from several major Hardware OEM's/Vendors.

I have seen several Newegg customers have RMA's approved no questions asked. (As referenced above).

Either way, you can take your fair chances on your existing system and work around the issue, or you can try to return it for a refund. No replacements will likely be in the supply chain until Late Feb/ Early March at best, and not in quantity until April. This is still not a formal recall, and customers with working systems are encouraged to work around the issue and deal with their vendors for best long term resolution.
 
I want my board replaced with the new Z68 chipset in February, a free dinner to a restaurant of my choice, and a new car. OK, maybe not the car...
 
niconx said:
Wow I see so many people saying/hoping this does not affect them. Cougar Point is *all* 6 series chipsets. This problem affects every motherboard that has socket 1155 on it. It also covers *all* laptops that have Sandy Bridge CPU's. Cougar Point is the only chipset that is even available for Sandy Bridge motherboards and laptops. Either get a couple sata controller pcie cards or get the motherboard replaced. Those are your options.

http://news.cnet.com/8301-13924_3-20030070-64.html
Click to expand...

Well it sounds like Intel Screwed the pooch on this one. :rolleyes:

Come On Bulldozer!
 
Stereodude said:
That wouldn't exactly explain the recall would it. ;)

How many other recalls has Intel done?
Click to expand...

i think he means a fatally flaw with the Pentium chip back in the day.

lets see there was the pentium 3 1.13ghz issue.
the pennitum chip back in the early 90's around 94 i think it was.
and then this SB chipset issues is the only ones i can think of off top of my head.
 
Just out of curiosity sake, is the Marvell SATA 6.0 faster than the Intel SATA 3.0? I'd assume so since it's a 6.0 but I read others commented about Marvell being slow so I'm not too sure.

If the Marvell's aren't that bad, wouldn't those that have 4 hard drives or less not be affected as much since there's 4 SATA 6.0 ports?
 
There are *4* SATA II ports that are affected.
There are *2* SATA III ports that are unaffected.

Re: "More than that..." - Performance still degrades over time, so you will see when/if/how you are affected and you can DO SOMETHING about it (switch ports, get a 3rd party controller card). I agree it's a SERIOUS issue, but it's not a FATAL one that would cause customers to lost systems/data en masse.
 
Well.. I had buyers remorse when I first bought this new Sandy Bridge rig. This shit just confirms my initial doubt/guilt. I just got off the phone with NewEgg.. I'm returning my motherboard, RAM, Case, PSU and CPU that I bought just for this build for a full refund. $1000 worth of hardware.

At least the SSD, 580 GTX and WD 6G hard drive I also bought a few weeks before Sandy Bridges release will work splendidly with my Q6600 that I didn't sell (thankfully).
 
Hmm PCI SATA II and PCI SATA II Raid cards are prominent on newegg.
Unlikely question, I assume using a SATA II device with a SATA II cable on a SATA III plug would be fine.
Anyone compared or know if cable quality would effect chip voltage?
Chipset fans anyone?
 
We just got emailed by intel telling us to stop selling any sandybridge board immediately and contact distribution to find out what to do with ones already sold.

Intel Recommends Channel Customers Stop Building and Shipping Systems based on Intel® 6 Series Chipsets

As a result of the design issue recently identified with the Intel® 600 Series Chipset, we are recommending that system builders stop building and shipping systems based on this chipset until additional information becomes available.

If you have already shipped systems to your customers based on the Intel® 6 Series Express Chipset and the C200 chipset and believe you may be affected by this issue, please contact your place of purchase for specific information about next steps.

For additional information about the nature of the issue, the cause and scope of the issue, and other general information,
Click to expand...
 
brom42 said:
1994(Wasn't recalled, but a flawed CPU): http://adage.com/article?article_id=88999
Click to expand...
Exactly my point. They didn't even recall the flawed Pentium chips, but they recalled this chipset, but we're supposed to believe it's no big deal. :(

I'm not saying it's the end of the world. I moved the two HD's to the P67 SATA-III ports on mine, and am deciding what to do next. Thankfully I still have my Q6600 based system and the SB system is just being played with.
 
LordCalin said:
We just got emailed by intel telling us to stop selling any sandybridge board immediately and contact distribution to find out what to do with ones already sold.
Click to expand...

I got the same email. I am a little concerned but to be honest I hope the RMA process is a smooth one.
 
I wonder how this will affect the price of sandy bridge? if it means the 2600k will go down in price while intel get's it's shit together then I might consider picking one up and then getting a MB when they are fixed.
 
Yes, a PCI SATA II would work around the issue just fine.
Yes, a SATA II cable will fit and work in a SATA III port at SATA II speeds.
Yes the situation sucks. I wish it didn't, but wishes don't count for much. Hope everyone is dealing with their own systems / parts appropriately.

Newegg pulled all the CPU's because selling 1155 CPU's when there are no known good compatible motherboards without this flaw on the market is kinda pointless. Same with most other vendors.
 
DON'T PANIC. Use your board and enjoy it. If you have a problem with it down the road exchange it, you'll be taken care of.
 
BlueAqua said:
DON'T PANIC. Use your board and enjoy it. If you have a problem with it down the road exchange it, you'll be taken care of.
Click to expand...

Bad advice.

Prolonged use might cause data loss, no ones knows for sure yet.
 
Last edited:
CheapAsianGamerGuy said:
There are *4* SATA II ports that are affected.
There are *2* SATA III ports that are unaffected.
Click to expand...
Even on the P8P67 PRO? It says it has 2x Intel SATA 6.0, 2x Marvell SATA 6.0, and 4x SATA 3.0. Obviously the 3.0's are affected but what about the Marvell's? Should they be fine as well since they are also 6.0?

Still, I don't know what I should do. I was planning on building a new rig Wednesday with the P8P67 and 2500K, among other items, but I'm not sure if I should proceed or not. I guess I have the option since I didn't put anything together yet but should I keep it or return it? Decisions, decisions...
 
You must log in or register to reply here.
Back
Top