Overclock.net banner

1 - 13 of 13 Posts

·
Registered
Joined
·
32 Posts
Discussion Starter · #1 ·
So, quick overview:

ASUS Crosshair V Formula-Z motherboard
AMD FX-8350 4.0GHz Processor
Corsair H80i Closed Loop cooler

I followed a guide to bring the 8350 to 4.8GHz, written for my specific hardware setup. (Mobo and cooler, too; I got lucky.) I couldn't actually push it to 4.8, so I dropped it to 4.7 (all multiplier,) backed the voltage off until I found the minimum voltage increase at which it was stable through an hour of P95 benching, and called it good.

Temps topped off at around 44c at heavy load. The thing was nimble, and solid as a rock. I kept it this way for about three months.

Over the course of a week and a half or so, I had three "hard" hangs. No BSOD, no reboot, no crashdump. Sometimes the GPU got the memo and gave up, blanking my screen, sometimes it just still-imaged. No ctrl-alt-del, only thing for it was a hard reset. It came back with no memory of what happened to it. I think once of the three times, I got the Windows "hey why'd you shut me off like that!?" message.

So, I backed it down to stock settings in the bios. No more hangs. I've been happy with it like this for a bit, but I'm sitting here thinking of the potential I'm missing. Does this story echo the experience of any seasoned overclockers out there? Should I shoot for the 4.7GHz again, maybe bump the voltage one more notch, or go for a more modest 4.6? I guess my concern is that after happy overclocking for such a long period, that aberrant behavior is a sign of something in my hardware gasping its' dying breath. But, with the temps I was seeing, that doesn't make sense to me.

Someone please educate this noob.
redface.gif
 

·
Not New to OCN
Joined
·
1,750 Posts
Wear is normal over a long period of time. I used to have my 2500k at 1.255V for my daily clock of 4.5GHz but now after 3 years I set it to around 1.27V to keep it happy as I have had some hangs like you mentioned. Upping the voltage a notch at a time until it stopped has worked for me.
 

·
Registered
Joined
·
32 Posts
Discussion Starter · #3 ·
I have to say that is exactly the opposite of what I thought I would hear. I'll go ahead and try it, but can you explain how a chip "degrades" so that is necessary? I'm really curious.
 

·
Not New to OCN
Joined
·
1,750 Posts
Here is an article I skimmed through that seems to explain it. LINK
 

·
Registered
Joined
·
91 Posts
Well, before you jump to that conclusion, have you performed some simple cooling system checks. My system did do that 6 months in, and discovered that the chipset heat sink (sorry, I'm old school and don't call the main 970 chip (the RX970) a north bridge given that the actual physical north bridge is built into the CPU) which was thermal taped came loose and was no longer transferring the heat away from the chip. I took it off, cleaned it up, lapped it down, applied new thermal compound (as opposed to thermal tape) and reattached it down to the mobo and haven't had a problem since.
 

·
Premium Member
Joined
·
1,671 Posts
Quote:
Originally Posted by nukedathlonman View Post

Well, before you jump to that conclusion, have you performed some simple cooling system checks. My system did do that 6 months in, and discovered that the chipset heat sink (sorry, I'm old school and don't call the main 970 chip (the RX970) a north bridge given that the actual physical north bridge is built into the CPU) which was thermal taped came loose and was no longer transferring the heat away from the chip. I took it off, cleaned it up, lapped it down, applied new thermal compound (as opposed to thermal tape) and reattached it down to the mobo and haven't had a problem since.
This. Before you give up, check your motherboard, specially if it has those horrible plastic push pin things holding down a heatsink on something. A hot northbridge can cause issues too. Same with VRMs, etc.

I bought a "junk" x58 motherboard. The northbridge was overheating. So I took the plastic pin things off, went to the hardware store, bought some rubber o-rings, bolts, and nuts and bolted the heatsink to the north bridge. I put the o-ring between the motherboard and the screw head to make sure I didn't short anything out.

Those sort of things happen. Have you been watching temps lately before it crashes? Sometimes radiators get filled with dust too. Could be a few things, not sure I want to jump on the chip degrading right away when there's other possibilities.
 

·
Registered
Joined
·
32 Posts
Discussion Starter · #7 ·
I'm pretty sure cooling isn't the issue, at least for the CPU. I had CoreTemp reporting CPU temps to my Logitch G13, so I was able to keep an eye on the temps while gaming or doing anything else, and my temps never shot far above 45c.

That said, I have to admit I'm only so-so when it comes to IDing important things on a mobo, unless I have to plug something into it. I'm taking my rig apart this weekend for cleaning and to change out some fans and graphics cards, so I'll be checking out all the heatsinks on the motherboard to see if I can spot any loose ones.

Thanks for the heads up.
 

·
Premium Member
Joined
·
1,671 Posts
Try something like HWMonitor that'll tell you your VRM temps and northbridge temps and stuff. I was able to isolate the problem on that x58 motherboard because I checked every temp sensor on my motherboard and the northbridge had no problem idling in the 90s.

It's a good idea to keep an infrared thermometer handy too. You just point it at things and it'll tell you the temp of what it is where the laser is. I had one but it broke. But I used it for everything, even cooking.
 

·
Registered
Joined
·
32 Posts
Discussion Starter · #9 ·
Quote:
Originally Posted by sdlvx View Post

Try something like HWMonitor that'll tell you your VRM temps and northbridge temps and stuff. I was able to isolate the problem on that x58 motherboard because I checked every temp sensor on my motherboard and the northbridge had no problem idling in the 90s.
Noted! I'll do that. What are acceptable temperatures for the VRM and northbridge, generally speaking?
 

·
Premium Member
Joined
·
1,671 Posts
http://www.overclock.net/t/1161772/what-are-your-990fx-northbridge-temps

Here's a good thread discussing it. TL;DR: it should be staying in the 40s or 50s. It's safe to higher temps but if it's not staying in that range stock there's a cooling problem.

VRMs can get very hot, like 100c+. I would imagine they could start causing problems at 80c+.

Also, if you want to check for degraded parts, underclock things. Underclock ram, run it at 1333mhz and see if it stabilizes. Lower the multipliers on CPUNB, NB, CPU, keep voltages the same, and see if it stops crashing. That's how you can check if it's degraded.

When I ruined my P4 Prescott on a suicide run, I could actually get it stable. I had to underclock it to get it stable to like 2ghz from 3.2ghz stock but it worked (but painfully slow, the chip sucked at 3.9ghz).

The fastest way to check would be to underclock everything by a good amount. If it fixes the issue, set half of the things you underclocked to stock and half underclocked. If it still keeps crashing, the thing that's causing it was set to stock. If it's fixed, the thing that's causing it is still underclocked.

Keep dividing things in half like that until you isolate the problem. That way will be much faster than just randomly picking things to underclock and seeing if that's the problem.
 

·
Joined
·
2,449 Posts
I'd check all of your temperatures before doing anything drastic.

I've always had trouble with parts of my AMD boards getting too hot. I think every one I owned required some sort of cooling mod. One needed heatinks on the VRMs and a fan on the northbridge, another needed screws on all the heatsinks plus some extra airflow around the southbridge.

With my Phenom II x6 1090T, I needed to keep it under 50C or else stability while overclocked became a gamble. I could easily boot with a CPU-NB of 3.5GHz. It was stable stress testing individual cores, but I couldn't stress test all 6 cores at the same time.
 

·
Registered
Joined
·
2,309 Posts
Honestly, if a visual inspectiion of the board reveals nothing obvious in terms of cooling problems, I would suggest just a little bump in Cpu voltage. It's normal to need to bump up voltages to keep overclocks stable over a certain amount of time. Besides, I've never seen a machine lock up because of a cooling issue.

Thermal shutdown is what you'll usually see, but usually a lock up can be remedied by upping the CPU voltage just a ltitle. Try putting your settings back as they were, but knock the CPU voltage up a notch and then start stress testing. If all looks well in terms of temps and stability, keep going mate.
 

·
Registered
Joined
·
5,447 Posts
Quote:
so I dropped it to 4.7 (all multiplier,) backed the voltage off until I found the minimum voltage increase at which it was stable through an hour of P95 benching, and called it good.
In this case I would try for a tiny bump in voltage, and running it for a few hours or running IBT (intel burn test) 10 passes at maximum.
 
1 - 13 of 13 Posts
Top