Computer won't boot unless severely underclocked

[Skeezix the Cat 3,089]

So I recently got the Thermal Take Tower 900 case because it's big and awesome, and transferred my system over.  At first I got no video out unless I switched the card to a different slot, but that problem is gone now so I think there was just some dust in it.  Anyway every time I start the system, it posts perfectly normally, the bios works great, etc. but when Windows starts loading with the black screen and the rotating circle of dots appears, it freezes right at the moment it would usually disappear and go to the sign in screen.  Sometimes it would hang completely, other times it would restart.  This is all at 100% stock, reset multiple times BIOS settings.  3.3GHz on the cpu, 2133 on the ram.  Even when it would try to load startup repair, it would still freeze or reset the same way.  I tried booting from a windows 10 usb and the exact same thing happens so it's not the installation itself.  The ONLY way I can get it to boot, and this works both for my installed windows and the bootable usb, is to severely underclock both my cpu and ram to 1.2GHz.  Even that hasn't always worked.  The way it behaves, it acts like stock speeds or even remotely close to stock speeds are a super overzealous overclock when that is clearly not the case.  I have reseated my gpu, all power connections, removed every optional device even down to one stick of ram which I tried in different slots and even checked for bent cpu socket pins.  Nada.  No temp problems either.  28c on the cpu, 34 on the gpu right now.  Has anyone heard of anything like this?

 

edit: gonna try the intel overclocking utility and bump my cpu up while I'm already in windows and see what happens

Edited February 24, 2018 by Skeezix the Cat

[tarikja 918]

Are all the power connectors really plugged in?

[Skeezix the Cat 3,089]

Yes.  Checked that because I have had problems before from not getting the 24 pin connector all the way down.

[Sceny 3,138]

Do you have a modular psu? Check the cables are plugged in correctly.

[Skeezix the Cat 3,089]

7 minutes ago, Sceny said:

Do you have a modular psu? Check the cables are plugged in correctly.

Not modular, although the wire sleeve on one of the cables is frayed.  I doubt it's the PSU, but I'm gonna find out soon enough.  Current update is I can boot consistently with my ram at 1866mhz, and I try raising my cpu clocks but no matter what I set it to in the bios it reports 1.2GHz in task manager and speccy.  Intel's extreme tuning utility doesn't let me raise the core multiplier in windows so I couldn't do it there.

 

I made the mistake of browsing new parts and since the i7-8700k is a hexacore and turbos to 4.7GHz stock, it would be a pretty sweet upgrade.  It's also on sale.  Then I saw the Gigabyte Aorus motherboard is only $160-200 for the Z370 chipset and after many more attempts to get my system to work at even close to stock speeds, I went ahead and ordered them.  Now my motherboard and graphics card will match.  I'm pissed that shit breaks, but shit breaking is the ultimate excuse to upgrade, amirite?  I'll still try suggestions of things I haven't already done or checked though because if it is some incredibly small or stupid thing and not actually a damaged component, then I could make a few bucks off my current board and cpu by selling it to a friend.

[ShikuTeshi 1,299]

Try removing your GPU and running of integrated graphics without underclocking. I imagine that would reduce the power threshold and if it boots then your PSU has gone bad.

Something like this had happened to my cousins PC when he games. It just started to shutdown in the middle of gaming due to not being able to supply enough power, and wouldn't boot sometimes until you fiddled with the PSU on/off switch and tried again.

 

An alternative to doing this if you don't wanna mess with moving hardware around is just to underclock very slightly until you can boot, then run simultaneous benchmarks on your CPU and GPU at the same time to stress the PSU.

Edited February 25, 2018 by ShikuTeshi

[Skeezix the Cat 3,089]

It's the X99 platform, so no iGPU.  Besides, I was able to stress test my 1080ti at 2GHz for a good 30 minutes while also "stressing" the cpu and had no crashes.  As far as trying to lower it until I can boot instead of rising it from the bottom, it doesn't even seem to want to change from 1.2GHz now.  The last time I got it to actually change, I tried 1.5 and it did not start successfully.  I've noticed it keeps resetting the cpu voltage to about 0.775.  This would explain the severe instability, but raising it to even 1.35 doesn't make a bit of difference.  It's almost like my bios is trolling me and only actually applying certain settings when it feels like it.  I may try flashing a newer version if there is one

[Solaris 3,581]

I would use a multimeter on your power connectors and verify that the pinouts are getting the correct voltage. Sounds like a PSU issue to me.

 

Did you happen to install any of the Spectre/Meltdown patches? It's been reported that those have caused a lot of issues with instability and reboots.

[Skeezix the Cat 3,089]

2 hours ago, Solaris said:

I would use a multimeter on your power connectors and verify that the pinouts are getting the correct voltage. Sounds like a PSU issue to me.

 

Did you happen to install any of the Spectre/Meltdown patches? It's been reported that those have caused a lot of issues with instability and reboots.

It does seem like a power issue and I'll try that when I have some free time to mess with it.  And no, I didn't install any patches related to that unless they were automatically installed without my saying so.  Even if they did cause my windows install to bork, I wouldn't think they would prevent a separate bootable USB drive to bork too.  I even tried booting from the USB with no other drives connected and it behaved the same way.  So there has to be something screwed up with my power delivery or some other hardware fault but I honestly can't see it being the psu because I ran both AIDA64 (on all components except drives because SSD) and Prime95 stress/stability tests for hours.  Not both at once, but separately and had no hiccups anywhere.  With a 1080ti running full bore at 2GHz, if I had a PSU fault then I think it would have crashed.  I feel like I might have somehow shorted or electrostatically shocked part of the VRM on the board, so the CPU is incredibly starved for power which would explain why the RAM has to be lowered too.  Haswell-e has a terrible memory controller and I have never been able to run it very high.