Can the North Bridge Crash a Computer?

[Shakis]

I am helping a friend diagnose his PC problem.

When under load the PC will power down after about 5-15 mins and will need to stay powered down for a few mins before it will even power back on.

I have an almost identical set up, for all intents and purposes the only differences are PSU, motherboard and the HDD

I have stress tested the cpu and gpu in both towers, only the DS3P mobo system powers down.

I updated from the F1 bios to the only other bios available (a beta bios) and the problem remains.

Windows has been reinstalled several times and on a second HDD, problem remains.

Installed all the latest drivers after clean installs of windows... problem remains.

WhoCrashed reported a few errors regarding corrupt RAM address and told me that nvidia driver caused the problem but I  have reinstalled them multiple times and on fresh OS installs. I replaced the RAM and stress tested them with no issues.

From some reading I done today I discovered that the North bridge controls communication between cpu, gpu and memory. So could an overheating northbridge corrupt data being sent to/from GPU and Memory.

Right before a power down during a stress test all temps (taken from CPUID HW monitor) were normal but north bridge temps aren't reported. When the PC powered down it would not start back up again and when I touched the NB heat sink it burned my fingertip. Is this normal or have I found the culprit of crashes?

Mobo: GA-970A-DS3P
PSU: OCZ 750w 80+ Bronze
OS: win7
CPU: FX-8350
GPU: Gigabyte Windforce GTX 770

[Vezina]

Your problem shows some overheating protection kicking in.The NB overheat can t shut down the PC.The NB chip is designed to take 95 Celsius and 125 in worst case scenario ,as per AMD data sheet

What you are having is a CPU VRM thermal protection kicking in.
Most probable the CPU is not supported for real on that board and you should thank God that you didn t blow up something.

The motherboard is a budget board and you should not use high end CPU-s on it.

FX 8350 should be used with high end motherboards.

What is the other motherboard used ?!
 

[Shakis]

Hi, thanks alot for the quick reply.

I thought it was some sort of thermal protection by the way it just powered off and then wouldn't turn back on for a few mins (I assume because it is still too hot or is on some sort of time out).

The other mother board (my board) that works fine with GTX 770 + FX8350 is an ASUS M5A78L-M LX V2

Should I be worried? I have had my FX-8350 for quite some time now without any issues. I've even used my rig for Dogecoin mining on the CPU and GPU without any (obvious)  issues.

[Vezina]

The motherboard is in culprit here as by what you are stating it fails to keep that CPU up and running.
Your friend needs a better motherboard with a good CPU VRM.
Nothing more you can do.
If the protection wouldn t be there the board would blow.
The VRM (Voltage regulator) for the CPU is quite stressed with an octocore CPU.

You could set the CPU to maybe a maximum clock like 3000Mhz  and with turbo disabled or you could disable one or 2 modules in the BIOS so the CPU sucks less power and thus the VRM stays in the safe operating area.Each module has 2 cores.Disabling 2 modules would make the CPU a quad.A quad is enough for today games.
This approach could be used until he sells the board and get a proper one.

But of course this would deny the right to full performance of that CPU
High end CPU s should be used in the right mobos and this is not a marketing trick it s about real hardware capabilities.Some cheap boards shut down like this one ,others throttle the CPU-s creating the illusion of a fully performing system