System Freezing, Looking for Advice

[t3hwUn]

Hello all, having some odd system issues.

I have a MSI Z87-GD65 Mobo with an i5-4670K. I had the system stable at 1.265 vCore and a multi of 47. The only other changed option was RAM voltage and an XMP setting. When playing BF4 I have began experience hard locks without a BSOD. When I restart the system the display is disfunctional.

What has worked is removing the GPU (GTX 670 2GB) and connecting to the onboard HDMI. System will then boot and I can reconnect to the HDMI on the 670 and the system seems fine from that point. Very odd issue.

I am thinking its the PSU not giving correct voltage. I have reinstalled Windows and still get the same issue. I have upped the vCore up to around 1.295 and will attempt again tonight.

What I don't understand is I can run AIDA 64 for 24 hours on 1.265 (SAME SETTINGS) and am rock solid stable.

I'm just trying to eliminate the cause and I'm thinking it could be the PSU, however in my experience the system wouldn't resume functioning as normal because of a bad PSU. Is it possible that my PSU is working but under BF4's load its not sending voltage properly but will come back on after my troubleshooting steps?

I've bought a new mobo to elimiate that as a possibility and its still occuring. GPU works fine and stress tested it to make sure. I'm just a little lost as to the actual cause. Obviously removing the OC would be a good step, but then why would it be stable for 24 hours under synthetic testing?

Any input would be greatly appreciated, thinking of trying a new and higher watted PSU. Suggestions?

 

[ATMINSIDE]

Have you tried cutting off the onboard GPU in the BIOS?

 

[t3hwUn]

I haven't yet, but once I plug into the onboard HDMI and back into the GPU it works fine... you still think thats worth a shot?

 

[ATMINSIDE]

Might as well. If that doesn't work, uninstall your GPU driver from Device Manager, download the newest version for your OS, and install the new driver.

 

[t3hwUn]

With the fresh install of Windows I included a fresh GPU driver of latest version. Again, the system is booting fine, but after dropping out of this system lock and going to onboard HDMI and then back, the system works fine normally again.

 

[ATMINSIDE]

Also, can you make a signature like mine with your full system specs? It'll help us help you.

 

[Mandrake4565]

+1 to the signature with the system specs, here's how.

Is the Gpu overclocked as well?

 

[t3hwUn]

Updated signature. GPU is overclocked but it seems to run fine with the clock. The one time I did get a BSOD with this crash that relates to no display from either VGA or HDMI on the 670 but from onboard HDMI was a 101 vcore related BSOD.

Yes, the GPU is OC'ed but its working fine after posting again...

 

[ATMINSIDE]

Drop the system back to stock, see if it runs fine.

 

[t3hwUn]

I understand thats the next troubleshooting step, but even if it does, then what would the culprit have been. The CPU is stable under synthetic load after these issues for the span of 24 hours!

I could maybe see the GPU, but again I want to get the system clocked again so I'm just wondering what would be the next step after that point. Make sense?

 

[t3hwUn]

***EDIT***

Sorry I was still at work and kind of treating this thread like an XMP client

Let me throw out more of what I know.


I've replaced the mobo with the exact same model. Same issue. Mobo has multi bios. Boots fine on alt BIOS just fine. On old model Mobo (which I am deliberating RMAing still) the flashed BIOS from backup wouldn't hold on the corrupted BIOS. Would POST and boot once, but after restarting or turning off no post.

Didn't realize unplugging and replugging GPU and going from external HDMI to on-board HDMI with no VGA (second monitor) gave me a display. Able to boot, turn off, restart and reconnect second monitor and external displays no problems, same settings all around.

What I don't know:

1) Can an underperforming PSU cause BSODs related to vcore (101) by not delivering enough or consistent enough power?
2) Can a CPU phase itself into higher voltage requirements? (Most I've pushed through chip is 1.41)

 

[ATMINSIDE]

A synthetic benchmark doesn't guarantee stability in all situations. BF4 is a demanding game, it might be that your system simply won't run it on quite as high of an overclock.

 

[t3hwUn]

I understand. I guess I'm just confused because the system was stable at that voltage for a good month. 24/7 in AIDA, multiple runs of over 12 hours and one run at 24 hours. BTW I was using 1 gig per thread in those tests.

Perhaps I just got lucky and didn't see the issue until a month later, just making me question what is actually stable.

I've up the vCore to 1.29 for now and I'm going to give this a shot. Had a crash on AIDA last night at 1.265 and a hang on the MOBO at pre-memory North Board Initialization, which gives me some more insight.

So either its the RAM, GPU or the CPU itself and its settings.

Thanks again for your help I'll be sure to keep this updated if I run into anymore trouble.

 

[t3hwUn]

Just wanted to touch up the thread. It seems its related to my RAM clock. BF4 doesn't like 2133 for some reason, lowered to 1600 and its working flawlessly. I will look into trying some tight 1866 timings as that seems to be the sweet spot from my research. Thanks everyone.