Master SuperMicro Atom C2xxx AVR54 (LPC clock) Design Flaw/Bug

[mifronte]

This thread is to discuss the Intel Atom AVR54 (LPC clock(s)) bug for users of Supermicro's Atom based boards.  For those that are not aware of Intel's latest AVR54 errata, it is concerning a bug with the Atom C2000 platform where the system will cease to boot.

This thread is for owners to share their experience and knowledge regarding this bug, especially those who have gone through the Supermicro RMA process.

I own a SuperMicro A1SRi-2758F based pfSense system that I built in September 2015.  I am approaching the 18 months of operations that has been mentioned as the threshold where the bug may occur.  I have not had any issues with the system up to this point.  The thought of initiating the RMA process is frightening especially given that Supermicro will be replacing the board with a refurbished product.  I don't know if this problem is a big issue or not, but looks like Intel may have set aside a huge chunk of money in their latest financial report to possibly prepare for the fallout.

For those who are currently using a Supermicro Atom C2xxx based board:
1. How long have you been using the board.
2. Have you had any problems?
3. Do you plan on initiating an RMA with Supermicro?

For those that have gone through the Supermicro RMA process:
1. How was your RMA experience?
2. Do yo need to be within the warranty period to be able to initiate an RMA?
3. Any advice on how to fill out the RMA form to ensure a smooth process?
4. How is the replacement board working out?

[Wolf666]

1- I own a A1SRi-2558F since October 2014.
2 - No problem so far
3 - I contacted Supermicro tech support about the problem with no reply.

[liontaur]

1. How long have you been using the board? - Since end of March, 2015
2. Have you had any problems? None.
3. Do you plan on initiating an RMA with Supermicro? Same as Wolf666, I started the RMA process at the start of Feb and haven't heard anything yet. I'm not overly concerned anyways.

Edited to say that we have the SuperServer 5018A-FTN4 with the A1SRi-2758F mobo in it.

[RobertLoblaw20381]

1. How long have you been using the board. 1.5 years
2. Have you had any problems? No
3. Do you plan on initiating an RMA with Supermicro? Already did

For those that have gone through the Supermicro RMA process:
1. How was your RMA experience? Cross-shipped, worked great
2. Do yo need to be within the warranty period to be able to initiate an RMA? Apparently not, see above
3. Any advice on how to fill out the RMA form to ensure a smooth process? Just mention the C2000 bug, they seem to know about it
4. How is the replacement board working out?  Good, see below

I got the replacement board in a cross-ship. This gave me the opportunity to compare both boards side-by-side. I cannot see any physical differences. The supposed platform workaround is to add a resistor, but I did not see a new resistor on the new board that wasn't present on the old board. The boards have the same revision number and the processor has the same stepping. I wish there was a way to get confirmation that the board I received has the fix. Has anyone been able to find the resistor?

[RobertLoblaw20381]

@mifronte:

I don't know if this problem is a big issue or not, but looks like Intel may have set aside a huge chunk of money in their latest financial report to possibly prepare for the fallout.

Are you suggesting that Intel is going to pay out the price of the board (~$450) to people who have the defective model and that it's worth not putting in an RMA for that reason?

[garyd9]

As long as it doesn't reference any netgate h/w, I think this thread is safe…

Mine is a C2558 based board.  Answers in italics

1. How long have you been using the board.  :  8 months at the time of RMA

2. Have you had any problems? : no, it was a wonderful board.

3. Do you plan on initiating an RMA with Supermicro?:  Already did (see below)

1. How was your RMA experience? : Annoying (and still not actually complete.)  First board, a cross ship, was flaky.  Some perfectly sane BIOS settings caused the board to lock up on POST.  Had a couple of kernel panics.  Eventually did several CMOS resets (with battery pulls) and got pfsense completely re-installed and then I had onboard ethernet adapter drop outs.  (As if someone was unplugging/plugging the ethernet cables.)  The board also ran 6 degrees celsius hotter than my previous one.  Finally got on the phone and yelled at supermicro.  They email'd me another RMA slip and a pre-paid shipping label… I sent that board back, they overnighted another one to me.  The second replacement seems great.  My original board is in transit back to them (and they should get it Friday.)

2. Do yo need to be within the warranty period to be able to initiate an RMA? : Technically, yes.  However, at least one person reported (in a now locked thread) that they purchased the board in Aug 2015 and they were still able to perform a cross-ship RMA.

3. Any advice on how to fill out the RMA form to ensure a smooth process? : In the problem description, provide a link to some reputable websites that discuss the issue, as well as link to the Intel document describing AVR54.

4. How is the replacement board working out? : see above…

[mifronte]

@RobertLoblaw20381:

@mifronte:

I don't know if this problem is a big issue or not, but looks like Intel may have set aside a huge chunk of money in their latest financial report to possibly prepare for the fallout.

Are you suggesting that Intel is going to pay out the price of the board (~$450) to people who have the defective model and that it's worth not putting in an RMA for that reason?

I am not suggesting anything.  I am only stating that Intel had set aside a reserve to deal with product quality issues and that this could be an indicator as to the seriousness of the defect.

[ivor]

Sorry, this has nothing to do with pfSense so I'm locking it. Similar thread went out of control last week, forcing us to lock it.