Return to

A strange old boot problem I've had for years


Hey guys, so I've had a problem with my pc that's haunted me for a couple years now. I've rarely had to face it simply by keeping my pc on all the time but I'm finally in the mood to smite this annoyance.
First off I'll post the main parts of my build
Core i5 4690K, Gigabyte Z97M-D3H, 16gb Patriot Viper memory, MSI Gaming GTX 970, EVGA SuperNOVA 750w Gold

So, here's how it goes. Whenever I boot my pc, regardless of it's a soft boot or hard boot, there's a few things that can happen. Each scenario seems to have about an equal chance of happening, maybe less so for a good boot and usually do not get successive boots of scenario 1 and 2.
1. There's a chance that the pc doesn't boot at all. The pc fans spin up and the power button LED gets lit up, but nothing else really happens. A quick tap of the power button immediately shuts off the pc.
2. The pc boots completely normally.
3. The pc boots up as normal but the boot is laggy to varying degrees. Now, i don't know what the cause of the lag is, but the way I've been measuring if the boot is a bad boot is by checking dpc latency with "DPC Latency Checker". I know that program introduces latency itself, but a bad boot will typically have astronomically tall bars, so it's a quick and surefire way to know. The issues while I have this kind of bad boot are typically a large amount of audio popping whenever there is audio playing, and artifacting when video plays. Also, if I were to boot up a video game, the frame rate tanks. Sometimes I get 20 fps, other times 0.2. I have no idea if these numbers are on a game-by-game basis or boot-by-boot but it's painful to test so I haven't. Curiously though, when I launched Left 4 Dead 2 on a bad boot, the issues of the boot disappeared so long as the game was running. The other day I had a boot that appeared to be a good boot, but while playing Rainbow Six Siege the game occasionally dropped frames from my usual ~90 to ~65. the frame times were nasty, so I suppose it's not always immediately nasty if I had a good boot or not.

Now, I've swapped my gpu from my usual 970 to my old 6950, and these issues completely went away. Weird huh? I concluded it was probably an issue with my gpu, so i got it RMA'd and the replacement card had the same problem. I got a refund on that card and bought another 970 second hand, this time from a different manufacturer (from gainward to msi) and I STILL have the same problems. This leads me to believe it's a driver issue, but it can't be because this issue has persisted through multiple driver reinstalls and I can't seem to find anyone else with my kind of issue (maybe I'm googling the wrong terms).
I'm left to think that the issue is either with the motherboard, with the psu, or with my windows install.
Any ideas? thanks.


did you use DDU to uninstall your driver cleanly?


From memory, yes. Originally my pc had even worse problems in the form of constant nvidia driver crashes, starting from every few hours eventually to every few seconds couple months later, I had to RMA the graphics card to get that fixed. During this time I had reinstalled my drivers tens of times out of desperation, and doing it in safe mode was my only option. I'm pretty sure I used DDU then but maybe there was one time where I forgot to? My replacement card still very occasionally had driver crashing but none at all from recent memory, it all just sort of went away.
I suppose i should give it another shot?
I just doubt it's a graphics driver issue, because it seems like whenever I get the non-boot scenario 1, the pc never loads up at all. Even if I leave it for an hour as soon as I tap that power button it switches off immediately so it's not like the pc is booted up with no signal to the monitor. I think it has more to do with the immediate boot-up elements.


at this point i would back up the data and reinstall the os. hardware is probably fine and a hard reinstall should cover most of your bases. im assuming windows? in which case you can sync your Microsoft account to the windows license and be sure not to loose it, and if linux then not much to worry about is there.


Unfortunately it's windows. Yeah, it's been a long time coming but I probably should reinstall hey? Maybe it's some anomalous bug that came from the windows 8.0 to 8.1 to 10 upgrades. I haven't synced the license to my account, but afaik it should still reinstall just fine right? Don't microsoft's servers sync licenses to hardware-profiles?


they use micro code to lock the license to the bios of the mobo. however now you can sync your license to your microsoft account at least with 10. im unsure about 8.1 but i think it works.


I totally forgot, this already is a relatively fresh install of windows. I remember reinstalling windows previously to fix a different issue regarding completely messed up registries. This is an issue that has persisted through reinstalls. I might retry a nvidia driver install.


In that case it's probably a hardware problem. Perhaps the PSU has trouble delivering the power to the rest of the PC, perhaps some of the cables are dodgy, perhaps the motherboard is dodgy, perhaps the GPU is dodgy or perhaps it's as simple as checking if all the components are properly seated in their respective slots. It could be any number of things.


Is your CPU overclocked at all?


Sounds like a short in the system.

I have had the same issues, especially the stuttering one, whenever connecting PCs in power loops.

wall -> surge protector -> power cord -> tower -> dvi cable -> monitor -> hdmi cable -> 2nd tower -> 2nd power cord -> surge protector -> wall

So there is a big loop where power is flowing, and that consistently caused massive stuttering in one of the systems. Solved that by breaking the loop, by not connecting two towers into the same monitor at the same time.

The not-turning on randomly issue was also a short from placing a tower on a surface that conducts even small amounts of electricity, like a tower placed on a mattress, or the ground.

wall->surge protector->tower->surface[] ->ground

The metal housing and rubber standoffs were not enough to prevent intermittent shorts on either surfaces reliably.

Solved the "not-turning on" issue by not turning on computers while they were on a conductive surface (bed), and a plank of wood separating the tower from the ground.

So... what could be causing a short in your setup? Is the computer fully isolated from other metal/wet/floor type surfaces? Any power loops?


Interesting. I'm running a fairly standard setup, but there is a power board leading to another power board. They are fairly old power boards too so perhaps I should get a new 6/8 socket power board. Overall there's only 5 appliances attached to the powerboards, 2 of which rarely getting used (printer and desk lamp).
As for shorts, it's a Fractal R5 sitting in complete isolation on top of a wooden floor.


No, my cpu isn't overclocked and never has been.

UPDATE: I've reinstalled nvidia drivers, seemingly no change. I've also swapped from my old powerboard -> powerboard setup to a new single powerboard and it seems to have made an improvement. Restarts of windows so far have not given me a non-boot scenario, and are much less often giving me bad boots.
As for cold boots, I still seem to be getting the same sort of problems as previously but I have not tested cold boots very much yet. WIll do so tomorrow probably.
Previously my setup was wall > powerboard for monitors > powerboard > system > monitors
So there might have actually been a nasty power loop that I'd never noticed going on there.

Another thing that may or may not be worth noting is that I have a dual monitor setup, and every so often a boot will cause my secondary monitor to have a messed up background. Foreground applications are totally fine though.


Since you mentioned that swapping the graphics card seemed to remedy the issue I would sugguest you check the firmware of your GPU that you are wanting to use and make sure it has EUFI support. I've ran into a few cards that were released in the grey area between bios and eufi adoption. For instance the Fury and Fury-X bios at launch had no EUFI support so you couldn't enable secure boot with CSM (compatibility mode) disabled. I have an MSI GTX 970 that has issue with several boards with or without CSM. It worth a look at techpowerup for your cards bios options. If you have a board that hates your card then try a different bios on the GPU and see what changes. You may have to use an older bios to get the desired behavior or try a newer bios if available.


Sorry for the hiatus, been lazy.
Anyways, I’ve flashed my gpu and almost bricked it doing it. Going to a newer bios did nothing, and going to an old one prevented the pc from posting, it’s fixed now though. After the first flash the pc couldn’t turn itself off normally and was followed with another bad boot. Whether this has anything to do with the flash I don’t know, might just be coincidental.
Using latencymon I can see that there’s two drivers responsible for this, dxgkrnl.sys and nvlddmkm.sys - DirectX and Nvidia. I’ve actually found another thread seemingly relating to my issue
and it seems like possible issues are related to power. One thing I’ve always had to do in nvidia’s control panel is to set ‘prefer maximum performance’, so it may be a power management issue with the gpu itself.
Any more thoughts?


sounds to me like you just have dodgy hardware somewhere… most likely mobo.


AdoredTV did a review of the DPC latency stuff and it’s worth a look. Hope you find the issue. One other thing I’ve recently ran into was needing to ditch the pigtail cables that come on most power supplies. I ended up running a cable from the PSU to the graphics card for each of the PCIE power connectors to get stability. I didn’t even have the cards overclocked but while gaming I’d hear a pop sound through audio and then the screen would go black. Windows would recover sometimes and sometimes it’d just hang for good. Once I replaced the power cables stability was restored as well. Anyway check AdoredTV’s breakdown on DCP latency here.