i7 reliability

[peterguthrie]

I have a couple of i7 980x based render nodes, mostly they are fine, but quite often i will remote desktop to them to find that 3dsmax has crashed. Never an actual bsod, just that 3dsmax quits (cant remember the exact message, will copy and paste it in here next time it happens.)

My question is, how to i diagnose the problem? I have no idea if it is the psu, memory issues etc.

They ARE overclocked, running at 4.3ghz, and never go higher than 70degrees C.

[tricky]

I guess an obvious test would be to not OC them for a while and monitor if things improve. We have always had things like this happen on our i7-based farm (4core and 6core machines). Its one of the reasons (for workstations at least) that we are moving over to Xeon-based systems to try and improve stability.

Are they running spawner, backburner or both? I have a script on my local machine whereby at the click of a button, I can restart spawner on all our slaves. Similarly I can restart the machines at the click of another button. Would that help you?

[peterguthrie]

I did for a while, and it wasn't any better, so thought f%^& it might as well run em overclocked.

Considering selling them or scrapping them and just sticking with xeon based machines, but many people seem to get on completely fine with i7s.

[AndresAhumada]

TEmps?
Issue whit dust?

Maybe a drama code whit max and i7

[Neilg]

I've seen this when nodes are running slightly different versions of plugins - sometimes if the versions are close enough instead of erroring out in backburner it'll pick the frame up and crash with no feedback.
Also sometimes on scenes that use 15.9gb of ram (16gb nodes)
They should be easy things to check before you dig around with the processor.

[tricky]

We just take random crashes as a fact of life. Xeons are certainly more stable in our experience.

[Companioncube]

something people rarely do is check if there are bios updates for the motherboard. system stability can greatly increase with a simple bios update.

do you know the motherboard make and model number? you can use cpu-z (free program) to find out

[peterguthrie]

the motherboards are fairly old, but will check for any updates.

attached is the type of max crash i get.

[AndresAhumada]

Reading in various Microsoft forums. He says the problem could be on SYNC Autodesk, which conflicts with some updates on windos 8 and 7.

Most likely it will be that ... but so many things that are chained, there is nothing certain.

[AndresAhumada]

The StackHash is a module error message you get when Windows Vista, 7 and 8, does not allow the execution of any file for prevention. That is, DEP is enabled. What you need to do is set it can open the one you want.

"If that file or application you want to open is your full knowledge and confidence, and also has no virus, malware, trojans, etc ... or is not harmful to your computer"

[AndresAhumada]

sometimes it can be a problem of the motherboard voltages. This can be either the power supply or wear capacitors motherboard, voltage controllers or just wear and electromigration.

[peterguthrie]

Hi Andres, regarding the first two, I'm afraid I'm not technologically savvy enough to understand your suggestions, could you explain in a 'for idiots' style?

and the last one, someone else suggested dodgy voltages, perhaps replacing the psu with a really good more modern one would be something to try??

Peter

[tricky]

What AV software are you running?

[AndresAhumada]

Hi peter ...

Based on first choices: I mean prevention system for malicious software or "recognized as melicioso" causes a crash.

This blockage causes basal autodeks files can not run and generates the problem.

This can happen in any system executable if for any error recognized as a dangerous program.

One option is to completely disable the system of prevention but this is dangerous. We live in a technological world full of virus.

http://windows.microsoft.com/en-us/w...ngs?SignedIn=1


In the last point.

You could check voltages directly on the BIOS or motherboard UEFFI. Whenever you generally indicates three voltages

12v
5v
3v

These can vary at most 0.5% in negative or positive. If you pass that value creates instability in Max system loads.

regards Peter

[Morbid Angel]

I found this, maybe helpful?

never mind.