Intel 6 Series chipset has a serious flaw?

[alex-the-cat]

Wierd, Newegg pulled the procs too. Backward time machine enabled!

http://www.newegg.com/Product/Produc...d=1&name=Intel

if there is no board to use the proc, why bother?

[Syl66]

Here is a little bit of info I read over on NCIX .

The feedback from Intel's tier 1 motherboard partners is that no one outside of Intel's lab has been able to produce this error yet despite attempts to speed up the process by increasing voltages to the affected components and subjecting them to heavy use. At this time NCIX recommends that customers continue to use the P67/H67 products until replacements are available.

[Lupine]

We've posted the official Asus response on the front page:

http://www.rage3d.com/index.php?cat=75#newsid33973552

[WhAt3v3r]

No bid deal. Jeez, they are correcting it. Just switch to the Marvel and put your DVD drive on the Intel. I'm glad I get a shiny new updated motherboard when they are out since I bought from the EGG, but what a bitch to undo my water cooling setup and do another mobo swap.

[noko]

Intel aren't covering costs, they are offering an allocation from the pot of money they've set aside. So each OEM/Partner puts together a number and justification of why they should get that much.

The partner wants the number to include punitive costs as well. They're the ones with the black eye and extra work. The number has to cover not only the cost of physical replacement, but the extra shipping, tracking, paperwork, office hours for handling it all.

Now, that $1Bn doesn't look like such a big pot anymore. So the partner wants to play both ends against the middle. They don't want to cover any cost they can pass on; shipping back to them of the defective board from the end user, advanced replacements, covering replacements of board models they don't have in stock with higher end models, step-up programs.

The end user wants to be treated like a injured party, premium concierge style. They want a free, better motherboard, without having to send the defective one back until the new one comes in (if they even want to send it back at all). Until the free, better motherboard is available, they want a free add in board that offers all the functionality that they lost. Despite having paid $150, they want to be treated like they bought a Bentley. Dammit, without them, the company they purchased from would go out of business! They're entitled to this!

The reseller/partner/OEM wants to minimize costs, while balancing PR. Depending on the market segment they service and the number of customers they have, they will approach it differently. Especially if they have nice wide product porfolio to defray costs against until they can work out passing on costs.

So large companies can put the price of all their lines up a small amount, get the money from Intel, and do right by the consumer for those affected. Smaller companies have to consider how much cash they have on hand, to pay for immediate costs like shipping, calling their suppliers to get orders cancelled, updated etc., as well as overtime/new hires for handling the deluge of emails and phone calls. They might be up against a choice of damaging reputation slightly vs. going out of business before the replacement boards arrive. And all of this doesn't factor in publicly traded companies that have to deal with shareholder and board pressure to keep money revenues high.

That's why we'll see what appear to be obviously bad mind-share and PR moves, in the handling of this recall. Intel are already doing the dance to appease the shareholders and board, by the very nature of their disclosure. Now they middle management guys have to appease the partners, OEM's and so forth, using the $1Bn the sparingly as possible.

Of course you got countless want of be users wishing for one of those defective boards too

[Bires]

Lost 4 SATA2 ports to gain all that? I'll take that deal!

+1.

I do wonder what happens in the spring when the new boards go on sale. Newegg said this:

In keeping with our commitment to our customers, we are extending the return period for your motherboard by 90 days or until replacements become available from the manufacturer, whichever is greater. Intel expects to have a new revision of the P67 & H67 chipsets out around April, at which point first-run motherboards with this issue will need to be physically replaced in affected systems.

[moshpit]

I like Anand's proposed solution as well, but I don't think it will happen. Perhaps some companies will offer 'step up' type programs to allow replacing the P67/H67 board with a Z68 for a token minimal cost.

If they can offer a Sabertooth P67 style board with Z68 for a minimal cost difference, I'd take that with a smile on my face and call this whole thing a plus. But, if the Z68 offer isn't for a board nearly identical in other aspects, including the Thermal Armor which I've decided I really like the look of, then I just want a straight up trade out for the same as what I have.

I'm also loath to give up the most forgiving overclockers motherboard I've ever owned. Auto is a worthwhile setting now! Seriously! And not just for CPU voltages, memory too! Auto seems to actually scale to exact real world needs in a way I've never had a board do before. As CPU speed is increased via multi, the vcore goes up JUST enough to keep it stable, manual doesn't shave anything off and keep stability. Same goes for vmem, this is the first board I dropped 4 sticks in, manually set speed to 1600 and timings to all 9's, 21, and 2T, and leave vmem on auto. Instantly worked fine and perfect stability.

Lost 4 SATA2 ports to gain all that? I'll take that deal!

[caveman-jim]

If Intel is covering the costs, why would any partner do any less than offer zero cost replacements and/or full refunds?

Intel aren't covering costs, they are offering an allocation from the pot of money they've set aside. So each OEM/Partner puts together a number and justification of why they should get that much.

The partner wants the number to include punitive costs as well. They're the ones with the black eye and extra work. The number has to cover not only the cost of physical replacement, but the extra shipping, tracking, paperwork, office hours for handling it all.

Now, that $1Bn doesn't look like such a big pot anymore. So the partner wants to play both ends against the middle. They don't want to cover any cost they can pass on; shipping back to them of the defective board from the end user, advanced replacements, covering replacements of board models they don't have in stock with higher end models, step-up programs.

The end user wants to be treated like a injured party, premium concierge style. They want a free, better motherboard, without having to send the defective one back until the new one comes in (if they even want to send it back at all). Until the free, better motherboard is available, they want a free add in board that offers all the functionality that they lost. Despite having paid $150, they want to be treated like they bought a Bentley. Dammit, without them, the company they purchased from would go out of business! They're entitled to this!

The reseller/partner/OEM wants to minimize costs, while balancing PR. Depending on the market segment they service and the number of customers they have, they will approach it differently. Especially if they have nice wide product porfolio to defray costs against until they can work out passing on costs.

So large companies can put the price of all their lines up a small amount, get the money from Intel, and do right by the consumer for those affected. Smaller companies have to consider how much cash they have on hand, to pay for immediate costs like shipping, calling their suppliers to get orders cancelled, updated etc., as well as overtime/new hires for handling the deluge of emails and phone calls. They might be up against a choice of damaging reputation slightly vs. going out of business before the replacement boards arrive. And all of this doesn't factor in publicly traded companies that have to deal with shareholder and board pressure to keep money revenues high.

That's why we'll see what appear to be obviously bad mind-share and PR moves, in the handling of this recall. Intel are already doing the dance to appease the shareholders and board, by the very nature of their disclosure. Now they middle management guys have to appease the partners, OEM's and so forth, using the $1Bn the sparingly as possible.

[Lupine]

Well, yes.... that is what Intel's $1Bn write is for, paying for the refunds and costs of handling the refunds, returns and replacements. Everybody expected that, where we need to look is to see which partners/customers of Intel don't offer full refunds or zero cost replacements of products.

I like Anand's proposed solution as well, but I don't think it will happen. Perhaps some companies will offer 'step up' type programs to allow replacing the P67/H67 board with a Z68 for a token minimal cost.

If Intel is covering the costs, why would any partner do any less than offer zero cost replacements and/or full refunds?

[caveman-jim]

From the above:

Samsung, however, said there would be no financial impact on its business as total payment will be funded by Intel.

As I expected, so end users should see no financial cost here, regardless of who supplied the faulty hardware.

Well, yes.... that is what Intel's $1Bn write is for, paying for the refunds and costs of handling the refunds, returns and replacements. Everybody expected that, where we need to look is to see which partners/customers of Intel don't offer full refunds or zero cost replacements of products.

I like Anand's proposed solution as well, but I don't think it will happen. Perhaps some companies will offer 'step up' type programs to allow replacing the P67/H67 board with a Z68 for a token minimal cost.

[Lupine]

Samsung is issuing full refunds

From the above:

Samsung, however, said there would be no financial impact on its business as total payment will be funded by Intel.

As I expected, so end users should see no financial cost here, regardless of who supplied the faulty hardware.

I like Anand's resolution the best though ...

I maintain that the best gesture of goodwill on Intel’s part would be to enable motherboard manufacturers to replace P67/H67 motherboards with Z68 boards for those users who want them.

http://www.anandtech.com/show/4143/t...point-sata-bug

[caveman-jim]

Samsung is issuing full refunds

[caveman-jim]

I don't know, Intel's stock buyback happens the same day as their stock price is depressed because of the announcement? Not only that, but on the 24th - when they knew about the problem of CP chipsets - they increased the authorization for share buyback. And then do the buyback on the same day they halt trading, announce the $1Bn write down, and recall. SEC, wherefore art thou?

[aop]

Charlie @ S|A doesn't like that explanation:

http://www.semiaccurate.com/2011/01/...old-any-water/

I wonder what kind of educational background does Charlie have? Has he ever read any electronics or does he even know how a transistor works or how is it made? I'm no expert and have only studied some electronics in University (including basics of manufacturing technologies) but Intel's explanation sounds much better than Charlies rant:

The problem in the chipset was traced back to a transistor in the 3Gbps PLL clocking tree. The aforementioned transistor has a very thin gate oxide, which allows you to turn it on with a very low voltage. Unfortunately in this case Intel biased the transistor with too high of a voltage, resulting in higher than expected leakage current. Depending on the physical characteristics of the transistor the leakage current here can increase over time which can ultimately result in this failure on the 3Gbps ports.

http://www.anandtech.com/show/4143/t...point-sata-bug

edit:
This is how I interpreted Anandtechs article:
Transistor gate needs voltage X to open the channel. Instead of correctly biasing the transistor Intel used voltage Y which causes the channel to be atleast partially open all the time and thus current leaks through it more than expected. The exact reason of transistor failure was not mentioned in the article but this may cause heat degradation of this transistor and eventually failure.

[Hapatingjaky]

Charlie @ S|A doesn't like that explanation:

http://www.semiaccurate.com/2011/01/...old-any-water/

The voices in Charlie's head told him this.

[caveman-jim]

Motherboard companies will be suing for damages shortly.

Unlikely. They will try to leverage this in preferential terms for future shipments and designs. And they will look to AMD to spread their business and be less reliant on one company.

[Razeus]

Motherboard companies will be suing for damages shortly.

[Skynet]

[noko]

In this case, I don't think quick is better. Not when you're dealing w/ user data.

If ports are disabled, other ports are good, user data all good
Except when hard drive fails . Still I rather it be my choice instead of someone else thinking they know better then I in what I need. I would never use half of the good ports as it is. In fact probably 99% of all users will not use all the available good ports.

Anyways I am hoping for a cut down or stripped version (as in dissabled ports maybe with plastic inserts in them with a big red warning label saying these ports are not used or something). At a reduced cost though .

[Lupine]

Maybe Intel will get their Head out of something to give a better quicker solution, I won't hold my breath though.

In this case, I don't think quick is better. Not when you're dealing w/ user data.

[Ristogod]

Charlie @ S|A doesn't like that explanation:

http://www.semiaccurate.com/2011/01/...old-any-water/

Wow. Long read.

This is really going to hurt the mobo companies though. Not only does it really mess with their planned productions, but just think of all the wasted materials that has to be brought back.

And how long before we see these boards back in action? My guess is maybe April?

[noko]

hehe, yeah things really happen quick at Intel and presto solution found.

Ranting so disreggard if it is offensive to you

Anyways I like to decide what I will buy as in if 4 SATA 6 ports and 2 SATA 3 ports are enough for me or not. Also if I need raid or not. Wait the SATA 6 ports are raid capable anyways . Hell if others can decide that then maybe I need even more PCIX 16x slots, three may not be enough, so then all cheapo motherboards also need to be recalled. The idea that some bloc just maybe dissappointed that those 4 sata ports are unusable, buys the board anyways knowing they are unusable is justification to prevent all shipments is utter nuts . If that is the case that is. Lets screw everyone because one person may get screwed or could get screwed mentality.

I guess the .32 micron Sandy Bridge production should stop as well since no reason to keep making them and storing them in some warehouse or build another great wall of China. Those folks should be layed off I guess until this is all sorted out.

Now if Nvidia had this standard we would be saying now "Nvidia who?".

Kinda fitting though in a way, Intel stingyness on chipsets, now there are no alternative chipsets to use Sandy Bridge with so all those CPUs can now gather dust .

Maybe Intel will get their Head out of something to give a better quicker solution, I won't hold my breath though.

[caveman-jim]

Charlie @ S|A doesn't like that explanation:

http://www.semiaccurate.com/2011/01/...old-any-water/

[moshpit]

No, because the design separates the two different sets - read Anand's update I posted above.

I guess that's what saved our bacon then. Okay.

[caveman-jim]

Drive, singular. The whole controller doesn't go belly up, only a single port with the remaining ports remaining functional was my understanding. This is a port by port issue, not whole controller. If it was whole controller, wouldn't the Intel 6Gb ports be affected too?

No, because the design separates the two different sets - read Anand's update I posted above.

[moshpit]

Plus a BIOS update to disable the affected ports would be good. The SATA card would need to be RAID capable and include tools for moving Intel RAID sets on it, which means it's not going to happen.

As far as interim solution, there is none - this is a metal layer problem, the chip will fail and cause the drives to disappear. This will happen intermittently during operation before final failure. There is no way to fix this without removing the chipset and putting an 'A' stepping or new stepping in place (boards shipped with stepping B).

Drive, singular. The whole controller doesn't go belly up, only a single port with the remaining ports remaining functional was my understanding. This is a port by port issue, not whole controller. If it was whole controller, wouldn't the Intel 6Gb ports be affected too?

[caveman-jim]

Sending out a sata2 card WHILE they work on the fix would be a nice interim solution, but still wouldn't remove the need to fully make consumers whole w/ 100% functional hardware.

Plus a BIOS update to disable the affected ports would be good. The SATA card would need to be RAID capable and include tools for moving Intel RAID sets on it, which means it's not going to happen.

As far as interim solution, there is none - this is a metal layer problem, the chip will fail and cause the drives to disappear. This will happen intermittently during operation before final failure. There is no way to fix this without removing the chipset and putting an 'A' stepping or new stepping in place (boards shipped with stepping B).

[Bires]

I don't know...so I have to move my spinning HD to the marvell, but while I wait for a new mobo...4.5GHz Quad core at 63C! I'm so glad I bought early. (so far)

[Lupine]

Sending out a sata2 card WHILE they work on the fix would be a nice interim solution, but still wouldn't remove the need to fully make consumers whole w/ 100% functional hardware.

[noko]

5% failure rate, other options available and likely to be used instead, a.k.a sata3. I don't know, what I've read so far would not stop me from purchasing a board and cpu, in my case I would probably never use the sata 2 ability anyways. So why not disable the Intel sata 2, give a sata 2 card out or incorporate another sat 2 device on the mobo which is a mute point since problem has been fixed I do believe with the newer 6 series chipsets. Got to hand it to Intel though for their high standards, also if this is the only problem after high volting it then Intel release one hell of a chipset in the end.

Seems stupid to have all these workable motherboards and cpus sitting around with the motherboards scheduled to be trashed. I can't see retrofitting a new chipset back on the motherboard but who knows.

So how long before Sandy Bridge plateform will be available again?

Ahmmm sell me one of those defective (possible and unlikely) boards for half price, I will remove it off your hands and save you all the trouble

[caveman-jim]

Anand's update on the problem - it's design issue, engineering 'oversight'

http://www.anandtech.com/show/4143/t...point-sata-bug

The problem in the chipset was traced back to a transistor in the 3Gbps PLL clocking tree. The aforementioned transistor has a very thin gate oxide, which allows you to turn it on with a very low voltage. Unfortunately in this case Intel biased the transistor with too high of a voltage, resulting in higher than expected leakage current. Depending on the physical characteristics of the transistor the leakage current here can increase over time which can ultimately result in this failure on the 3Gbps ports.

You can coax the problem out earlier by testing the PCH at increased voltage and temperature levels. By increasing one or both of these values you can simulate load over time and that’s how the problem was initially discovered. Intel believes that any current issues users have with SATA performance/compatibility/reliability are likely unrelated to the hardware bug.

[caveman-jim]

Ask yourself this...

Didn't Intel JUST report record earnings for the quarter?

Do you see the coincidence of giving this information AFTER the quarterly report?

This is the same Intel that has FTC oversight inside the business - onsite, in meetings, remember?

[Razeus]

Intel isn't that stupid. Please explain the upside?

Ask yourself this...

Didn't Intel JUST report record earnings for the quarter?

Do you see the coincidence of giving this information AFTER the quarterly report?

[caveman-jim]

Intel isn't that stupid. Please explain the upside?

With a tin foil hat on, I would say financial year end reporting.

[noko]

Go figure, bought the DDR3 ram, Win7, just ordered the Sata 3 hard drive and then this comes out . Was going to purchase the mobo/cpu next week.

[Lupine]

This will hold up my Macbook Pro purchase that I'm sure Apple had in the works for a refresh.

Something tells me that Intel let this one through the gate with full knowledge.

Intel isn't that stupid. Please explain the upside?

[Syl66]

NCIX where I purchased my board from has already stepped up and made this announcement.

Intel announced this morning that there is an issue with their P67/H67 chipsets that can cause the SATA 3.0Gb/s controller to fail over time. NCIX has put in place the following measures to ensure that our customers who have already bought NCIX PC systems or standalone motherboards are appropriately cared for.


These Notices Are For All Intel LGA1155 "Sandy Bridge" Customers:
1. Please note that this issue does not affect the LGA1155 CPU itself, nor does it affect any other Intel chipset products (such as P55 or X58).
2. The chipset is believed to be MORE LIKELY to fail with heavy use. To reduce the risk of data corruption or data loss, NCIX recommends that you move all drives to either the Intel SATA3 6.0Gb/s ports or any other 3rd party chipset SATA3 6.0Gb/s ports (check your motherboard manual). It is also recommended that if you must use any of the SATA2 3.0Gb/s ports, you should use them with your least important drives or ones that access read-only media (ie DVD burner).


These Notices Are For All NCIX PC Customers With LGA1155 Systems:
1. Customers who have already purchased their system will be credited with 5000 NCIX Rewards points.
2. Systems that are currently in the assembly phase will have to be approved by the customer before they will be completed & shipped.
3. All systems will be assembled using only the available SATA3 6.0Gb/s ports whenever possible.
4. NCIX will perform complimentary motherboard exchanges once replacements are available from the various manufacturers. Once updated boards are available, please schedule an appointment online or with your local NCIX retail store. If your order was shipped, then NCIX will cover the shipping charges both ways to return the system or motherboard to us for replacement.
5. The 12 month NCIX PC warranty will be extended on ALL Sandy Bridge systems to 18 months to compensate our customers for any inconvenience that is caused by being without their system during a motherboard exchange.


These Notices Are For All NCIX Customers With Standalone LGA1155 Motherboard Purchases:
1. Customers who have already purchased their motherboard will be credited with 2500 NCIX Rewards points.
2. We are currently working with all of our partners to find a solution for customers who have purchased Sandy Bridge components. Please check back to this forum thread for the most up-to date information. Again, we suggest that to reduce the risk of data corruption or data loss, please move all drives to either the Intel SATA3 6.0Gb/s ports or any other 3rd party chipset SATA3 6.0Gb/s ports (check your motherboard manual). It is also recommended that if you must use any of the SATA2 3.0Gb/s ports, you should use them with your least important drives or ones that access read-only media (ie DVD burner).


Please check back periodically in this forum thread as we will be updating it once we have more information. Subject to change without notice.

I've moved my SSD and Apps drive over to the Intel Sata3 ports. I'll leave my burner and general storage drive on the Sata 2 ports as I don't really anticipate any major problems until the board can get replaced.

[Razeus]

This will hold up my Macbook Pro purchase that I'm sure Apple had in the works for a refresh.

Something tells me that Intel let this one through the gate with full knowledge.

[sandorski]

No sense Selling a CPU with no available Motherboards. Especially for the less informed who throw the CPU in the Shopping Cart then start E-Mailing about not being able to find the Motherboards for it.

[Hapatingjaky]

Wierd, Newegg pulled the procs too. Backward time machine enabled!

http://www.newegg.com/Product/Produc...d=1&name=Intel

Yeah I don't see the sense in pulling the processors. Intel can still make some monies off of it, purchase the processor now and wait for new boards to be available.