PC boot loop and then eventually boots up

[Arclite]

Hi guys,


A long post coming up so please bear with me.


I built my current system back in 2014 (specs in sig) and had no problems until the November of 2016. Before that I had overclocked my cpu to 4.6ghz, along with ram and gpu and had it on pretty much 10-15 hours a day that included sessions of gaming and occasional benchmarks to check for any abnormalities in temperatures or overheating issues. From memory, i cannot recall my cpu under 100% synthetic load exceeding 80 degrees C and the pc never locked up in any games i ever played.


All this changed in November 2016. Whenever i started the pc from a cold boot, the fans would spin and the pc will boot loop 3-4 times (no post to bios) before booting up and then it would function normally. I did not pay much attention and thought that it may be the motherboard doing its thing (like some are known to do).


Then in December 2016 the boot loop time went from 2-3 minutes to 20-30 minutes and then to 2-3 hours and then over 10 hours near the end of the year. At each point though the pc would eventually start up and have no issues whatsoever. The overclock was still working fine, temps ok, no crashes during gaming or benchmarks and I just could not get my head around this problem except that I dreaded restarting it because I knew the cycle will start again.


Now here I am 4 months in and the problem is still there. Boot loops every 2-3 seconds of powering up, does not post and it could go on for some minutes or hours and then it starts working as normal again.


The steps I have taken to resolve the issue:


1.Cleaned everything as much as I could.

2. Restored factory defaults on everything so no overclocking on any component since December 2016.

3. Tested each component on its own by removing everything else.

4.Memory tested multiple times through memtest and it comes out fine.

5. Breadboarded the setup to see if the problem is somewhere in the case.

6. Re-connected all the wires by removing psu.

7. Removed GPU and ran intel HD onboard graphics.

8. Removed hard drive one by one.


I am sure I am missing one or two more things I have done but you get the gist.


Last week, I decided to have it looked at by a “professional” at a shop and after 3 days and £25 later I was told that my cpu is dead and/or ¾ cores are not working. I was just lost for words at that point especially after telling them beforehand that all benchmarks I have carried out that include multicore stress checked out. Also, they said they took my CPU out and tested on another motherboard but I find it hard to believe since the pc looked fairly untouched.


I am bit strapped for cash currently so cannot really afford another “professional repair” (most are cons anyway) and spend on parts when I do not know where the problem is. Sadly, I do not know of anybody who would let me check some of my parts like cpu or psu.


I do not think it’s a software issue since the boot loop happens like within 2-3 seconds. I doubt hard drive even gets started at that point but I could be wrong.


What do you guys think? Anybody else ever had a similar problem.

 

[Plec]

The symptoms could well be CPU related given the symptoms - as it reads like an unstable clock but I appreciate you've not had it clocked since December. If I had to make an initial guess I would say your clock became unstable and over the months it's damaged the CPU - it may have been desperately trying to let you know in November that it wasn't happy with the repeated boot cycle.

It's unfortunate that you don't trust the 'professionals' (I can totally understand your paranoia - but there are some genuine chaps out there) - because if they have tested your CPU in another system and recreated the same problem then you have your diagnosis. Did they explain what happened when they put the CPU in another motherboard?

You've tested everything else to the nth degree - although the PSU is the next most obvious culprit if you've ruled out the memory.

Have you tried the paperclip test with the PSU - the full Monty test though - checking the rails with a volt meter?

Also, I'm guessing you've done this and just forgot to list it - are you running the latest BIOS?

 

[Arclite]

Thanks for your input.

They told me that they took the cpu out and put it in another 1150 socket motherboard and it boot looped the same way. I guess I may be a bit sceptical about them doing that but I suppose i can take their word for it. I tried the jumper clip setting with the psu to jump start it without pressing the on/off pc button and it did the same. The pc repair person also used a volt meter to check the psu and said it was ok.

The bios was updated to the latest version as soon as it was released.

If you think that this may be a damaged CPU, what else can i do to confirm that? I have tested using the routine benchmarks and I do not see issues with the cores at all. Perhaps I missing something, something that would confirm that it is indeed a damaged processor. I am sure you can understand my hesitance in accepting the notion of a damaged cpu because when the pc runs, it runs without any issues.

 

[Arclite]

Another thing i forgot to mention is that boot loop ranges from anywhere between less than 5 minutes to a few hours sometimes before the pc boots up. It is so inconsistent that I cannot even predict a pattern. Right now the pc is on since last night for about 15 hours. Did some gaming (i miss it a lot) and currently its just idling. Temperature during gaming were 40-50C, idling between 28-34 on all cores.

 

[Plec]

I can totally understand your hesitance - but you're running out of things to test and I've been on threads like this before. It's taken an age and a few grey hairs to get to the point where you are - when it comes to identifying a faulty CPU.

You seemed to have taken every logical step, and I suspect even if I think of another step for you to try the odds are you've tried it, you clearly know what you're doing - and you've even had your suspicions validated by a third party (although, again, I appreciate your scepticism).

The harsh reality is if you weren't strapped for cash I think you would have made your educated guess and diagnosed your problem by now - but I've been there myself - it's the most expensive component and you'll be without your computer, even if it's the right decision...

If you trusted the shop's tests you wouldn't be here - so what does your gut tell you after all your tests? It's nearly always right after this amount of testing...

The only way you can confirm it, for sure, is by sticking it into another rig or testing another CPU in your present rig - but I realise none of these are options.

Other than leaving your rig on 24/7 and rebooting every now and then (what happens with a reboot?) - I can't think of anything else that you haven't already done - other than a different PSU (I know they tested it - but PSUs can be bitchy).

 

[Arclite]

I guess my gut feeling tells me that its either a shoddy power cable somewhere that is causing a short or the psu is failing or at least I want to believe that this is the case. I cannot test the cpu on another motherboard so i guess the only logical step for me to do is to just get as much use out of it as i possibly can before either the pc completely fails or a new symptom appears or when I have the financial means to probably upgrade the entire rig. Until then, however, its going to be a difficult period for me - not only because i will never know when the pc will boot after a restart but also due to not knowing where the actual problem is.

Rebooting puts the pc back into a boot loop so any sort of auto-restart due to software update overnight is annoying to say the least.

 

[Plec]

One last thing to try before you resign youself to an expensive slow death... (is there nobody you could beg/borrow/steal a PSU from - it doesn't have to be all powerful you could run a skeleton setup?)

You've probably tried this but - try slackening the memory timings and manually upping the voltages for the memory and Vcore at stock. It's a long shot but if the CPU is on the way out it could appreciate the extra juice at stock - I would give it a significant bump too. The memory bump is just 'belt and braces'...

I would wait for an update/restart though as any restart seems an expensive experiment - and if it doesn't work you can wait until the next reboot to put them back.

 

[Arclite]

Sadly no, I have just moved to a new place and i do not know anybody locally and most of my friends and family are very basic users of pc and do not have the necessary components.

The memory was running at fairly tight timings and high voltages when the system was overclocked last year. When i restored the defaults, the memory voltage and timings were very slack i feel but I will give it a go again the next time the pc is restarts and I am actually physically present to boot in to bios which I am not because the pc never starts at any particular time point. I just leave it and hope that when i get back home or have free time that its up and running. No harm in trying though.

 

[Plec]

It's the Vcore bump, combined with the slackening of the memory and it's slight voltage bump, that i'm more interested in.

There's a chance that the clock on your chip has degraded it's integrity - so much that it can't run stably even on default voltages at default clock.

For example as my CPU has got older I've had to throw more volts at it and muck around with the BIOS to keep things stable at 4.5GHz - when really I should just drop a couple of MHz like a normal person.

Your chip, however, may be feeling the strain even at stock now - and require a manual kick up the pants to initially wake it up - which is why I suggest a significant bump to start off with. If by some miracle it works - slacken off until it buckles...

 

[Arclite]

Still have not rebooted the pc since 2 days now. Was working fine when i was gaming last night. Perplexing to say the least. Lets what happens next...

 

[Plec]

Unfortunately, there's been a large Windows 10 update (assuming you're using that) - so you may find it reboots today/tomorrow :/

At least you could try the Vcore experiment...