Sandy Bridge

[Deneb]

No matter how you slice it, Intel’s new 2600K is a cold-blooded killer

People will look for a lot of reasons to hate Sandy Bridge: Overclocking is limited to the K parts, you have to buy a new board, and the graphics core is switched off once you install a GPU. But once you get to raw, ripping performance, it’s hard not to gush over Sandy Bridge. Frankly, it’s an astounding amount of performance for
the money.

The top-end Core i7-2600K smashes every other quad-core Intel chip by healthy margins. This is aided by the new microarchitecture, the ring bus, and other magical stuff, we suppose, but we see no reason to buy any other CPU for the money. Even the once-powerful Core i7-975 Extreme Edition is flatly punched in the nose by the 2600K. While the 975 is long gone, you can extrapolate that the 2600K will outgun the Core i7-950, i7-930, and the poorly priced i7-960. Against non-Intel chips, it’s no contest. AMD’s hexa-core Phenom II X6 1090T, which was already getting beaten up by existing Hyper-Threaded Core i7 chips, also takes a serious thrashing from the Core i7-2600K.

Even the mighty Core i7-980X loses a few benchmarks to the Core i7-2600K chip. These are mostly in benchmarks that can’t exploit the six-cores of the 980X, and where the Turbo Boost 2.0 gives the Sandy Bridge part a key advantage.

Certainly, overclocking the older Intel parts and the Phenom II can help, but the 2600K, we must add, also overclocks like a champ. So, haterz, set aside your hate. If performance is what you care about (and you don’t want to shell out for a $1,000 hexa-core), Sandy Bridge, particularly the K versions, should be on your radar.

http://www.maximumpc.com/article/features/sandy_bridge_washes_ashore

[Deneb]

A billion-dollar mistake: Intel recalls a supporting chip for popular Sandy Bridge platform

“The tough part for Intel is that it can’t fix the chip with a simple software update. It has to remake the chips in its factories, and that causes a delay of weeks for computer makers who want to ship their machines now. Since Intel won more than 500 designs for Sandy Bridge, the delay could affect sales for the entire PC industry.

Intel said it discovered the flaw in the recently released support chip, code-named Cougar Point, and has had to redesign the silicon. That could mean a delay of 12 weeks (the typical time it takes to get a chip through a factory) in getting corrected chip sets to customers. Cougar Point had flaws in its Serial ATA (SATA) ports, which means that SATA-linked devices such as hard drives and DVD drives might not work properly over time. Cougar Point is a companion chip set for Sandy Bridge, which itself was not affected by any design problem.

Intel said it expects to deliver the updated version of the chip set to customers in late February and expects a full recovery of production volumes by April. The company said it would accept the return of the Cougar Point chip sets. The systems with the bad chips have been shipping since January 9 and there are relatively few consumers who already have flawed systems in their hands. The potentially affected consumers are those who bought second-generation Core i5 and Core i7 quad-core systems.”

http://venturebeat.com/2011/01/31/a-billion-dollar-mistake-intel-recalls-a-supporting-chip-for-popular-sandy-bridge-platform/

[RickyTick]

Gigabyte announces a motherboard replacement program.
http://www.anandtech.com/show/4146/gigabyte-announces-6series-motherboard-replacement-program

[zburns]

New report by 'The Tech Report' dated this morning regarding the Sandy Bridge Problem: http://techreport.com/discussions.x/20326

There is some new information in this report that is reassuring to the Sandy Bridge overall program. The Tech Report article has a 11:43 am update; the sixth para down in that update refers to the location of the problem being in a chip fabricated on the older, and therefore, very familiar 65 nanometer process, not the very new 32 nanometer Sandy Bridge process. The report quotes Intel as saying to the effect that the problem is: "very straightforward" and it has "very high confidence" that the fix will be effective.

EDIT, Feb2, 9:30pm. Here is a Tom's Hardware link about the Sandy Bridge mess. http://www.tomshardware.com/news/cougar-point-sandy-bridge-sata-error,12108.html
It is worth reading, as is the Tech Report and the Anandtech article (above Ricky Tick). Bottom line replacement P67 chipsets will delay mobo shipments until April according to worse case senarios.

[piikea]

How will anyone know they're buying one of the "fixed" chipsets in the future & NOT getting one w/ the flaw that was still floating around in the supply chain somewhere??

[zburns]

For one thing, the mobo manufacturers will recall all mobos currently out at distributors and replace them or refund $ for them - something like this. All the mobo's have identification on the board itself, so identification by the mfg, distributor and the end user should be relatively easy. Newer motherboards with the revised chip will likewise have identification that would clearly show or prove they do have the correct chip on them. The motherboard manufacturers will be the most 'rigorous', I would think, in getting all the 'bad' stock back.

[meg]

Yesterday, I received a great e-mail from Newegg, where I purchased my ASUS mobo. I found the you-tube video very helpful, and joined the forums there for additional information. Below is what I received in the e-mail:


Thank you for giving us the opportunity to serve you. Newegg has recently become aware of a design issue that is affecting recent models of Intel Sandy Bridge platform motherboards. We are working with Intel to identify the exact nature of this problem.

As always, Newegg remains 100% committed to our customers' total satisfaction. 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.

From a technical standpoint, the design issue can be bypassed fairly simply by not using the Serial ATA (SATA) ports that are affected. Your motherboard’s manual should identify your SATA ports by number, and at a minimum you should see ports 0-5 (6 ports in total) listed. Ports 0 and 1 are Sata Rev. III (6Gbps), and do not appear to be affected by this problem. Ports 2-5 are SATA Rev. II (3Gbps) and should not be used. For a thorough explanation of this hardware work-around, please refer to our video on YouTube:
http://www.youtube.com/watch?v=tJcE2alQPvY

If you choose to use the hardware work-around option, there is no need to contact us at this time. We have your information on record and will email you as soon as the replacements become available. If you would like to discuss this with our tech community or read up on the latest updates, please visit our EggXpert forum:
http://www.eggxpert.com/forums/thread/682006.aspx

If none of the above options are suitable to your needs and you wish to return the board at this time for a full refund, please email us at intelsandybridge@newegg.com and include your sales order number so we can help you out with your return.


Hope others find this valuable, as well.

[IlliniGuy]

Sorry if this isn't the proper board, but I purchased a Gigabyte Mobo that was affected. This is my very first build. I haven't had time yet to switch my HDs to the two unaffected SATA ports, but yesterday my machine shutdown while watching a movie. It powered back on fine though. My question is, could this be due to the Intel bug or do you think I have another issue with my build?
Thanks!

[zburns]

If you have a rare one time shutdown, forget about it; there are too many things that could cause it. If you have this on an ongoing basis, then try to see a pattern both to what was happening before the shutdown and any pattern to it 'powering back' up. If you were physically involved right before the shut down, then it is a good bet that something you did that contributed. If it shuts down and stays down, that is different; something definitely out of kilter. If you can see a pattern in the way a problem occurs, that does help figuring out what the possibilities are. For example I can shut my Sonata III down instantly by pulling out and reinserting my USP printer plug into the front panel while getting out of my chair at the same time -- unequivocally a static discharge problem -- from the chair and also, making the mistake of letting my finger touch actual metal on the USB plug or the metal inside of the front panel USB female connector.

If you do have the Sata problem, it will continue to happen in what seems a random manner. But the time the machine is on until the problem occurs would be an indicator. If the amount of time the machine is on before failure is roughly the same, it is probaly the Sata chipset problem; the intel problem has to do with current leakage in a transistor circuit. That current leakage occurs over a period of time, maybe only when that SATA outlet is being used. But again, if a consistent pattern emerges, I would assume the SATA problem. Anyway, you can change the SATA ports and get away from the problem until get your replacement.

"That current leakage occurs over a period of time, maybe only when that SATA outlet is being used." This would be the key to identifying the problem, from the time you start to use that Sata port, the problem would occur consistently in the same amount of time or most of the time.

[RickyTick]

All they really need to do is swap the ICH chip, desolder it and solder the new one. I'm sure they'll test it before sending it back out. I don't think I'd worry about getting the same motherboard back. I could be totally wrong. They may be doing a completely new motherboard swap.

Here's the latest from some of the major board makers.

Gigabyte (http://techreport.com/discussions.x/20344)

Asus (http://techreport.com/discussions.x/20345)

MSI (http://techreport.com/discussions.x/20349)

[zburns]

Hey Ricky, Not sure about your solution. If the board is returned to the manufacturer of the board and they have a 'specific setup' to machine 'desolder' the chip on the board and lift out the chip, put a new one and do the equivalent of 'wave solder'; then they run a full QC test on the board -- that works for me. Not sure they can do that. I just hope the mobo manufacturers can remove the chip and put a new one in without any degradation to the board or to the new chip. Ok, what I did not say is that all these boards all of us use in computers we build are mostly assembled by machine and soldered by machine -- they are not done by hand.

But if they have a crew of folks with solder irons, doing the replacement desoldering and putting in the new chip -- bad idea -- no controllable Quality Control using this method. In essence the chip has to be replaced using a method that results in the repaired board looking identical and having the replacement chip resoldered in place with no more heat on the chip than occurs in the normal new manufacturing process.

[RickyTick]

Hey Ricky, Not sure about your solution.
Neither am I. :)

I was sort of speaking in jest. I really don't know what they will do. I feel bad for everyone that bought one of the motherboards. Especially the ones that I encouraged to go the Sandy Bridge route.

[zburns]

Hey Ricky,

You were right to make the comments above. I respect the mobo builders and their expertise; I just sort of expect them to do the right thing. But there is nothing in print yet. Without asking any purchaser of a Sandy Bridge mobo about how they feel about their board being repaired with a replacement P67 chip (if that is the chip), I think they all assume they wind up with the equivalent of a new board with the same new warranty. My point about how the board is repaired is simply asking the question, What is the senario that unfolds now that mobos must be replaced. If I presently owned a two week old Sandy Bridge motherboard and had to return it for a 'repair' here is what I would expect to be shipped back to me:

1. My same motherboard repaired with not only a quality control statement regards the retesting of the mobo, but I want the repaired mobo to look physically the same as the brand new one I sent back for repair (as if I took comparison photographs).

2. As important, I do not want someone else's board because I have no knowledge of how that mobo was handled by the previous owner; this comment goes to static control and long term damage due to poor static control in the assembly and disassembly process.

3. Unless the mobo manufacturer can guarantee my points (1) and (2), I want a brand new board with a brand new warranty.

In point (1) my comment about looking physically the same goes to my earlier statement of 'repair by hand', wherein the bad chip is desoldered by hand, replaced by hand and resoldered by hand. The repair would look much different (my opinion) than a automated or semi automated repair by machine (assuming the manufacturer can do it by machine).

This is not a small chip regards the number of pins on it. It is made on the 65 nanometer process with 'X' amount of pins surrounding the perimeter of the chip; I am not even sure it can be replaced by hand.

Given that this is a mistake by Intel and violates the trust given Intel by end user purchasers, they are entitled to a repair equivalent to a new board or they are entitled to a new board altogether. Many or all of these motherboards are only weeks old.

Also, I have not said one word about the 'delay factor' and the so called implied cost to the end purchaser as a result of the chip problem and the inconvenience of waiting until April for something to happen.

So, my point, the more you stare into the problem the more problematic it gets!

[RickyTick]

And #4. I don't want to re-install Windows or go through the hassle of reactivation of a Windows license.

[zburns]

Sandy Bridge Motherboard Replacement

This post is for those of you who have a motherboard involved in the P67 replacement problem. In my posts above I made reference to potential problems related to 'hand replacement and hand soldering' of the replacement chips. This comment came about due to my prior experience some years ago. However, in looking at my motherboard and video card inside my case, I can see enough large chips and their pinouts (pins surrounding the chips) to say that I do not think it possible to replace these chips by what I refer to as "by hand" and, therefore, those comments of mine should be disregarded.

However, it is still prudent for those of you who have these boards to get a firm statement that the replacement boards you get back with the repair will either be brand new or have a repair that looks to the naked eye exactly like a brand new board. I would prefer a statement from the manufacturer stating the full senario regards the replacement of boards to you -- ie. brand new or repaired 'totally like new' with extended warranties, etc.. Now is the time to push the manufacturers to treat your problem with the seriousness it deserves. Do not wait until the last minute.