CPU0704 CPU1/CPU2 Machine check error detected on Front LCD Panel

Kaizenphilo, 

Would you clarify the processors you have installed? Are they the Intel E5-26xx V4 or Intel E5-26xx V3  versions, if so would you verify in the BIOS if C states are also enabled?

Let me know and we can go from there. 

Hi Chris,

Thanks for the response: They are the Intel E5-2643 v4 and we're using the default Performance per Watt (DAPC) System Profile setting -- C states are enabled.

Thank you. There is a BIOS update it sounds like is in the works to resolve this, but there is a temporary work around that we can try temporarily until the update comes out. 

Try the following;

1.        Change BIOS profile to Performance

2.        Select Custom profile, set C State to disable and C1E to enable.

Let me know if that, for the time being, resolves the issue for you.

KaizenPhilo,

We are going through this exact same issue, on a T620 with dual Xeon E5-2630's. We are getting CPU0704 CPU2 Machine check error. I've have updated BIOS and firmware with no change, as well as change the performance setting in BIOS. It is rebooting very frequently...about every 40 to 90 minutes. Did you ever get a resolution on this? I've contacted support and sent a support assist log. They suggested swapping the CPUs to see if the error followed the CPU. I have not done this yet. But may have to if no other solution is known. Any information you have would be appreciated. Thanks in advance!

KaizenPhilo or Copper,

Do you have a solution now?

 At the moment we have 5 servers (PowerEdge T640) with the same problem. "CPU machine check error" randomly on CPU1 or 2 about once a week, at random time. The server restarts without Windows logging. BIOS / Fireware / drivers are all up-to-date. Is a Windows 2016 server with Hyper-V enabled. BIOS profile = Performance (C-state is disabled).

Dell support has replaced three system cards. The problem exist.

It looks like the error is sometime come up when a virtual machine is starting up in the morning. I can’t reproduce the problem manual.

Dell is searching for a solution, but it is slow. Someone any idea?

I'm having/had the same issue on an R740, Server 2016 Standard with Hyper-V role. CPU Machine check error, OEM diagnostic events, and Intel Management Engine warnings.

2+ months into service call and we've replaced the motherboard twice, the PCI card (BOSS NVMe riser) twice, the cables once... Flea drains, cold boots, AC cycles, you name it, we've tried it.

We've cut down the CPU machine check errors but the OEM diagnostic and Intel M.E. events still come along together, and it looks like it's happening when the system hangs. I've seen it now several times where during a restart it will hang on Shutting down service: Hyper-V Virtual Machine Management.

Support just informed me of a new BIOS, 1.4.5, so I'm running my restart script on it and see what happens... the results just in, updating the BIOS re-generated the CPU machine check error (does this mean it didn't go away? Is it a throwback reference? Who knows) and then the system stopped on trying to shut down H-V VMM service.

Hi Sierra,

We also update the BIOS to version 1.4.5 for five days now. Update also Windows 2016 to KB4103720. We have not seen the CPU machine check error till now. But it is too early to cheer.

Do you have the possibility to reproduce the problem manual? Reboot the machine is in our environment no problem. The CPU machine check error coming up only random on a running machine. 

I know the waiting time for 'Hyper-V Virtual Machine Management' service to stop. But it is an event I have seen on a lot of machines.

When I have news, I will post it.

All servers are up for three weeks since the BIOS update 1.4.5
I must slowly believe that the BIOS update solve the problems.

Sorry for the delayed reply, it didn't notify me and there's been so much going on :D

So, we never did get a fix for this - I have no idea what's caused it, but this continued to be a problem. Eventually they caved, got us a new server dispatched, and we have sent the old server in for their Dev team to rip apart and find out.

If I ever find out what the cause was, I'll happily share it with the world - there was no need for this to have taken me well over 3 months to have a brand-new server that works. But, I'm just happy it does...

We have a stack of R620's with 2X E5-2667 v2 that almost all seem to have this problem.  What I don't understand is when running memtest or stress it runs for days without any problems.  Once we try loading up Windows 2016 again it starts to crash with the CPU0704 errors on CPU1 & CPU2.

Setting the Profile to Performance in the BIOS appeared to help on some of the servers but not all of them.  I thought maybe it was just a Windows 2016 problem and was going to try loading CentOS but it looks like someone already tried that and was still having problems.

Update the iDRAC/LCC and the BIOS and other system firmware? That is the most common reason for those types of errors. Any third-party cards/hardware?

Everything is updated. Nope, no extra hardware/cards. We already pulled the whole server apart, cleaned everything, re-seated CPU, ram, all the risers and cables. Swapped CPU 1 & 2 around but still having the same problem. Extended stress tests and Dell Utilities don't find anything wrong.

We have a M1000e full of M620 blades with dual Xeon E5-2650 v2 (Model 62 Stepping 4) and have this problem on all of them.

They are all running Windows 10 and Bios, CPLD, DRAC, etc is all latest versions. I know Win 10 is unsupported but we need it. The blades is just part of a render farm and Win Server is overkill for that.

It first started about a year ago. And it seems to happen in waves and linked to when Microsoft pushes out updates. Last time was in august, and now again in the middle of December. Restarting a machine without installing updates is mostly working just fine.

The only way to get a machine back up is to just keep on trying to restart it and hope that it will start, or reinstall Windows completely.

Setting system profile to Performance helped a little bit but it still happens.

This is so **bleep** frustrating. Please Dell, fix this.