Beware of over / under clocking


There is no question that over and under clocking could cause system instability that could lead either to loose of data or system hangs, so why write about this? Because system stability, as I found hard way, is also OS specific.

Common way to decrease system power usage is undervolting components so that way, even thought they do operate within same spec, they do use less power to do so. Less power equals to less heat being generated, so this also helps to decrease component temperature a bit.

Personally I do use, or should I say used to use, undervolt by setting negative offset for CPU and after initial setup and testing in Windows OS everything was working fine. I have Debian on dual boot and at that time initial testing there also passed. At the time was happy with setup as everything was working fine.

Fast forward few months and BIOS upgrade, GPU replacement, and on Windows OS there was no problem whatsoever. Everything from compiling, video editing, gaming to rendering was working fine and system was stable.

The problems started when I did booted up to Debian after few months. All of sudden system exhibited random hangs. There was no pattern to it, it could work for 5 minutes and hung or 1 hour and then hung. Workload also didn’t mattered. Lightweight workload like web browsing using Firefox could cause hang or during compile. The confusing part was that in Windows doing same workload worked fine.

Of course after few months people tend to forget stuff and obviously I also forgot about undervolt being applied on system. Spending few days! to troubleshoot issue finally did find solution … remove undervlot. Just disabling it resolved every instability that I did experienced on Debian.

The strangest part is that Windows didn’t exhibit those issues. One could think that this means that that OS is better, but in my book that is quiet opposite – personally I always expect system to crash / hang when hardware related errors occur (and this apply also to software, no graceful error handling, just dumping backtrace and error’ing out, and obviously fixing issue). Of course I can’t tell if Windows did encounter those issue but I did experienced some issue with my nvme ssd’s having some fs errors so it could be that those errors did happened.

Bottom line if system is stable on one OS it doesn’t mean that is will be stable on other OS. And always remember about applied over/under clocking on system and disable those as first step during troubleshooting.

, , , ,

Comments are closed.