[The Reg]C2000-series Atoms (Rangeley) bricking devices

[DrMrLordX]

http://www.theregister.co.uk/2017/02/06/cisco_intel_decline_to_link_product_warning_to_faulty_chip/

So um, apparently that might have been the problem cited in Intel's Q4 2016 earnings call? I have heard that it was Atoms, but it's actually kind of bad that it's embedded stuff since they will be mostly unable to just swap the chips.

Unless they expect to people to go in there with soldering irons, Cisco and Intel will be on the hook to replace entire devices.

 

[Arachnotronic]

http://www.theregister.co.uk/2017/02/06/cisco_intel_decline_to_link_product_warning_to_faulty_chip/

So um, apparently that might have been the problem cited in Intel's Q4 2016 earnings call? I have heard that it was Atoms, but it's actually kind of bad that it's embedded stuff since they will be mostly unable to just swap the chips.

Unless they expect to people to go in there with soldering irons, Cisco and Intel will be on the hook to replace entire devices.

Oof, a swing and a miss.

 

[inf64]

The timing of this is very unfortunate. Do they know how many devices are affected?

 

[dark zero]

Wait... Isn't that Avoton chips? Oh godness... Intel wants to kill Atom for good?

 

[Arachnotronic]

Wait... Isn't that Avoton chips? Oh godness... Intel wants to kill Atom for good?

No, it's called a mistake. It happens.

That being said, this is a pretty bad one. Not quite FDIV level, but still...

 

[XavierMace]

On the Cisco side, it affects 4300 series routers and a handful of other devices. We've had to replace 4 already due to failure and we're working with them to get them all replaced.

 

[IntelUser2000]

This is quite serious. Avoton chips were quite popular. Not intentionally, this may prove to be a death of the Atom line entirely.

If they want any goodwill from customers, I suggest they do what they did with 6 series chipset bug in 2011 and do a full refund/recall when necessary.

 

[nvgpu]

https://ark.intel.com/products/codename/63508/Denverton

Intel is replacing Avoton C2000 with Denverton C3000 soon, the Atom isn't going anywhere.

 

[dark zero]

At least Intel did a good move in order to solve this delicate situation. Seems that some of them wants to save Atom.

 

[KompuKare]

If they want any goodwill from customers, I suggest they do what they did with 6 series chipset bug in 2011 and do a full refund/recall when necessary.

In the Register story there is talk of the affected OEMs not 'wanting' to name Intel as the supplier of the faulty part. This sounds to be like they are under some corporate NDA from Intel because the OEMs are in effect taking the hit only their reputation while letting Intel off the hook.

 

[DrMrLordX]

Apparently there's a board fix for units that haven't failed yet. My guess is that it reduces clockspeed/voltage to slow down (or prevent) chip burnout.

If the chip is already burned out, it would require a device refurb that would basically mean pulling the entire chip + board since it's BGA. PITA all around.

 

[nathanddrews]

That does suck, I was looking at building an Avoton server, but went with a used Xeon setup instead. Close call. I don't doubt that Intel will do everything it can to satisfy their clients and mitigate brand damage, but this is a pretty bad situation for them given the timing.

 

[Nothingness]

So it turns out not only mobile Atom are useless... Sorry could not resist, I have developed an allergy to the Atom brand

 

[Rifter]

This makes me happy i went the celeron route for my pfsense box. Wow, atom really does just want to die doesnt it.

 

[sm625]

It is amazing that a decade old recycled CPU with multi-decade old SuperIO could have a flaw like this.

 

[Rifter]

It is amazing that a decade old recycled CPU with multi-decade old SuperIO could have a flaw like this.

Me to, it seems intel has not only gotten lazy lately as far as innovation goes they have also gotten horrible at QC.

 

[krumme]

http://www.anandtech.com/show/11110/semi-critical-intel-atom-c2000-flaw-discovered

"As a result we finally have some insight into what the issue is with C2000. And while the news isn’t anywhere close to dire, it’s certainly not good news for Intel. As it turns out, there’s a degradation issue with at least some (if not all) parts in the Atom C2000 family, which over time can cause chips to fail only a few years into their lifetimes."

"Cisco, whose routers are impacted, estimates that while issues can occur as early as 18 months in, they don’t expect a meaningful spike in failures until 3 years (36 months) in."

 

[coercitiv]

Atom, it's not that Intel doesn't want you, it's Lupus!

 

[krumme]

I have a bt nas fortunately not c2000 series. Man. Its not the error you want on your nas device. The c2000 series is the atom series where this should not happen. Not good - they should be 100% stable and last 6 years easy. But hey its fragile stuff so it happens but qc should be tougher for such a series.

 

[itsmydamnation]

i wonder if this effects the Dell 4048 DC switches, i just deployed a whole heap of them as TOR switches. They also use Rangeley..... O'well i guess this is why you design A-B paths in datacentre environments.

 

[IntelUser2000]

It is amazing that a decade old recycled CPU with multi-decade old SuperIO could have a flaw like this.

Atom, isn't a decade old CPU. The design is based off the 2013 Silvermont core. It's not even 5 years old.

Though, Intel seems to have problems outside of their core competency. By core competency I mean specially in CPUs, that are being used as their main bread and butter line.

 

[thecoolnessrune]

I don't know why Anandtech's article was so soft on this. This is a *massive* problem for us and big vendors like Cisco, Supermicro, and Synology that has *tons* of these units in the field.

We have 1 client alone with over 400 Cisco ISR 43xx's across North America, and we'll be spending months getting these replaced.

 

[VirtualLarry]

Hmm. Does this mean that my QNAP TS-451 NAS with an Atom 64-bit CPU is likely to die? What about my Win8.1 (w/bing! ) 7" tablets?

Does this affect CHT at all? I have a "TV box" PC, with Cherry Trail and Win10, and it ... is a little flaky. I left it running 24/7 as a HTPC, and it would get random restarts occasionally.

Edit: Hmm.

UPDATE: As of about 1 month ago the 451 will not run for more than about an hour before it locks up ... no errors populate all the light indications are normal but the NAS will be inaccessible and the only way to return it to operation is to force a reboot. I've done factory resets, rebuilt arrays swapped drives and the issue remains. I no longer trust this unit.

From a recent Newegg review. Could this be the "degradation"?

 

[krumme]

Hmm. Does this mean that my QNAP TS-451 NAS with an Atom 64-bit CPU is likely to die? What about my Win8.1 (w/bing! ) 7" tablets?

Does this affect CHT at all? I have a "TV box" PC, with Cherry Trail and Win10, and it ... is a little flaky. I left it running 24/7 as a HTPC, and it would get random restarts occasionally.

Edit: Hmm.

From a recent Newegg review. Could this be the "degradation"?

The ts451 is J1800 or J1900 so should not happen. Only c2000 line.

But hey i asumed you got a direct mail from Intel because of all your atom stuff
hell i even have some atom stuff i dont know what is.

 

[krumme]

I don't know why Anandtech's article was so soft on this. This is a *massive* problem for us and big vendors like Cisco, Supermicro, and Synology that has *tons* of these units in the field.

We have 1 client alone with over 400 Cisco ISR 43xx's across North America, and we'll be spending months getting these replaced.

Yep. They treat it friendly and like its a consumer product. Its a sensible subject and thats why Kyle is on it i guess.
The total cost of this is out of this world. A good case to show why qc on such stuf is vital.
Unfortunately testing and qc is low status work many places. But used right its certainly a most important strategic advantage.