Boot Loop

Hi,

I've reached the end of my knowledge & Googlefu, my computer is stuck in a boot loop randomly (aren't intermittent issues the best) the USB devices aren't getting power either. It did it briefly one night, then I gave it one last start and all was well for the evening. Now it's permanent

In no particular order I've tried:

• Clearing the CMOS
• Changing the CMOS battery
• Trying both sticks of ram in each slot to see if I have a dead stick or slot
• Re-seating the CPU & GPU power cables
• Plugging my keyboard and mouse directly into the PC instead of the hub
• Removing and repeating the CPU
• Re-seating the GPU
• Using my known good power cable from a monitor into the PSU
• Powering on without any cables

I know the monitors are good as I use then with my work laptop and checking the ASUS Q codes (14, 15, C8 & very briefly 3E). The codes all point to CPU &/or RAM, am I missing something here? The CPU is less than a year old and the RAM isn't much more from memory, so a warranty claim is an option... but which one?

Specs:
AMD 5800X3D AIO cooled
32GB Corsair 3600MHz CL16 2X16GB (can get model number if needed)
ASUS ROG Srix X570-E Gaming
EVGA (RIP) 3090 FTW3 Hybrid
Some Western Digital NVME C:/ drive
An old Samsung 980 pro
2 really old Western Digital Sata spinning rust drives
Windows 11 (less than 3 months since fresh install)

Hopefully there's something I overlooked and it's an easy fix If you need anymore info let me know.

2-BIT-GEEK

P.S. I couldn't be bothered to update my signature on my phone and write all that

 

How old is the system board?

 

It's from early 2020, but the BIOS was flashed to the latest version at the time in August 2022. It's been running fine ever since.

 

The red flag is the USB ports losing power. That has me thinking power supply or system board. How old is the PSU?

 

Yeah, it did seem strange. The Q codes and RGB working were a positive sign to me, but maybe not. The mobo and PSU are the same age, just over 4 years old.

 

The problem with intermittent faults is that they come and go... it turned back on an hour ago. I'd still like to get to the bottom of the issue if possible, I'm not very confident tough

 

Well it's back of again, I got about 2 hours out of it including restarting for updates. It cut out and got stuck in the boot loop again in the middle of normal use, just a YouTube video while running ASUS fan tuning.

I'm not intentionally bumping the post, stuff keeps happening this evening

I can't say I've noticed before if this is normal, but when I turn it on I can hear what sounds like a faint electrical short. That may be normal and it only happens when I first power on, not each loop.

In the interest of including anything relevant I vape near the PC, any of you who also do will know it gets in no matter what you do and builds up. I didn't include this earlier because in my experience eliquid isn't conductive and my last system ran for 6 years without issue in the same environment, dust does get stuck to it though.

 

There's different voltage outputs for different components. 12v, 5v, and 3.3v. Also, Windows could be turning the USB ports off to "conserve power". But, the boot loop may also point to shaky RAM modules as well. Bad bits will write erroneous data and kerfuffle the OS. Are you familiar with offline hardware testing? The Ultimate Boot CD is one of my favorite resources. It can run off a bootable USB stick as well.

https://www.majorgeeks.com/files/details/ultimate_boot_cd_(ubcd).html

 

I'm leaning more to messed up RAM at this point, I ran 3D Mark Time Spy benchmark when it came back to life briefly and it ran fine end to end. I feel that if the PSU wasn't working right it would crash putting that kind of load on it.

I've never heard of offline hardware testing, I'm guessing the tool is part of the boot CD? Hopefully the computer will be lucid enough tomorrow evening for me to setup the USB and give it a shot.

I've not seen this message before after failed POST's, I also haven't seen a lot of boot loops recently until now luckily. I doubt there's any useful information that'll help either.

 

Yes, the tools run without booting Windows. The flash drive will create it's own runtime environment using Linux. MemTest would be the one you're looking for. Any red is bad. Have you gone into BIOS and loaded your optimal settings? Have you overclocked your hardware?

UBCD official page:
https://www.ultimatebootcd.com/

 

Having reset the CMOS I've got the default settings for the RAM & CPU, I turned on some boot setting to allow my NVME boot drive to work. So at default speed I ran the Windows Memory Diagnostic to quickly check it before the system started POST looping again, it passed. So I ran Memtest86 and it started looping again before the test could finish


No overclocks, I turned off the RGB on everything just in case. I'm not sure I'll be able to run memtest as the system probably won't be working long enough.

 

Did you run Memtest offline, or inside of Windows? The key to running MemTest with Linux is it rules out hardware, and leads to software.

P.S. Why did you clear CMOS and perform a BIOS reset?

 

I see what went wrong. I was asking to see if you had done anything like a BIOS reset, 'cause that could create problems. I was not instructing you to load optimal settings. Hence the question mark.
So, to be clear. When you loaded the UBCD and ran MemTest, the machine restarted?

 

I had a cpu fail on me years ago and the symptoms were that the system would just suddenly freeze. The monitor was on and still showing the last image displayed. A hard reboot seemed to fix it but only temporarily until the system froze again at some point.

Because I had spare parts for every component, I began by systematically replacing one piece of hardware at a time (including motherboard) until I discovered that it was the cpu, which I did last! Needless to say it took ages to fix the issue.

That being said I am not convinced from the symptoms you describe that your cpu is at fault here but of course that is merely my opinion.

My suggestion to you is as follows: do a barebones set up ie disconnect everything from the motherboard with the exception of the cpu, vga, one stick of ram. Obviously, the monitor needs to be connected as well as kb and mouse.

Boot and run memory tests for that stick of ram for 24 hours and also run Prime 95 again for 24 hours if possible.

If those tests pass, then start adding devices one at a time, starting with a boot drive, and continue testing.

I know you are going to say that this will take ages and will drive you mad but right now you are trying to trouble shoot in the dark and that may end up likely taking you even longer!

 

Well, we could just jump to this. It's in the CPU category.

https://systester.sourceforge.net/about.php

 

Sorry for the late reply, I had a wicked cold the last couple of days (probably COVID).

I ran memtest from its own bootable environment since I couldn't get the UBCD to boot, but memtest86 booted and ran fine... it just didn't complete before the system restarted itself back into the POST loop. I've not reset the BIOS other than when I reset the CMOS.

Not right away, but yes.

A 24 hour test per RAM stick isn't possible, I've not had it working for more than 2 hours before reverting back to the POST loop. I've currently got everything non essential unplugged anyway. My main drive is an NVME drive under my GPU, that's the only one not unplugged (it's a serious pain in the ass to get to).

I feel like throwing a little money at the problem to see what happens. I can't find the receipt for the RAM annoyingly, but I have the CPU's as it's under a year old. If I get the same RAM again the best case scenario is I don't the issue, the worst case I then have 64GB of RAM with the same issue. Would this help to pinpoint the issue at all?

If we 100% rule out RAM what would the next steps be?

Thanks for your time and patience people, I appreciate it

 

All good. It's a little tougher when the machine isn't in front of you. Getting down to faulting the CPU or the system board isn't fun. Technically. Me? I'd be curious to look for computing errors, however. See about running a basic CPU test. You don't want to throw money away, so maybe we give condemning the right part a shot? Once had an old AM2 on a BioStar board (cores unlocked) that ran like a top, but had a single burnt pin/receiver. Out of 941...

 

I presume that you ran memtest individually with only 1 stick of ram and then ran it with the other stick and got errors on both counts? If so, hard to imagine 2 ram modules going bad at the same time.....

Not sure how much 32GB of ram is gonna cost you. As a cheaper alternative, why not buy say an 8GB used ram stick from fleebay? If memtest fails again with another ram stick, then ram cam be removed from the list.

Would then leave mobo, cpu, psu and gpu to diagnose?

Doesn't feel like a psu issue. What psu do you have?

From what you have said, your pc will go into this boot loop even when booted from cold, so I think that can rule out chipsets & vrms overheating.

About the only other thing left that it could be, given the description of the problem, is that the motherboard is for some reason now perhaps shorting out on the standoffs. Heard that this can lead to intermittent boot loop issues.

The only way to remove the possibility of that kind of a short would be to take the mobo out of the case and try.

 

I can officially rule out RAM... I saw a great 'limited time' offer and grabbed the same kit again, it changes nothing unfortunately/fortunately. The same thing happens with the new stuff only, the old stuff and both together.

In the time it's been cooperative I've run a 20 minute loop of the 3D Mark Time Spy extreme stress test successfully, then Prime95 small FFT's for approximately 8 minutes (error free) before it resumed POST looping.

I'm not disregarding the standoffs being an issue, but the motherboard was installed only once and never removed. I might be wrong, but it doesn't seem possible to run fine for years without moving and now being an issue, could they corrode in 4 years to cause a short though?

Boot loop in hindsight isn't the most accurate description. It's not POSTing, it's a POST loop. It stops on ASUS Q code 15 Pre-memory system agent initialization is started with the RAM Q led staying lit. The led suggests the system has error checked the boot drive, VGA and CPU before being unable to check or pass the DRAM. After Q code 15 it cycles, I listed the ones it briefly stops on before getting stuck on 15 and trying again.

It's hard to say as the time is constantly changing, I think it will always boot eventually and the same for system on time, it will always crash after 'an' amount of time regardless of the activity or load. Just this evening I got 8:37 while on Discord and downloading a game update, so not a massive load. Then about 2 minutes later it ran for 23:08 running Spider-Man Miles Morales without DLSS, using Discord and streaming it.

I've even tried turning the system off immediately after it's fully booted to see what happens, I left it 10 minutes between each boot she it worked 5 times in a row without fail.

After all of this I can't imagine it's the PSU, it wouldn't allow a stress test or gaming for random amounts of time surely. It's a Corsair RM850 2020. It must be the motherboard or CPU, hopefully the CPU as that's under a year old and under warranty. Unless the PSU is still on the table, it seems to me like a computer shop with parts to throw at the system could at least tell me which of the two parts is causing the issue, if it's one of them. I could also keep their meddling and repair time down by supplying them with only the mobo and CPU right?

 

Run some of those calculation tests in the CPU section of the UBCD.

What System Stability Tester can do in brief

• Calculate up to 128 million digits of Pi.
• Create up to 32 threads.
• Recognize the CPU model, manufacturer and operational frequency.
• Compare the calculation results of all the threads after each step and issue a warning if something went wrong.
• Run continuously for up to 50 turns and log the results after each step and each turn.
• Record the calculated digits of Pi in a human readable text file.
• Single threaded calculation of Pi for benchmarking, similar to the well known SuperPi benchmark.
• Checksum validation over the result.
• Provides graphical and command line user interfaces.

 

I am far from convinced myself that it is a standoff issue and now even less so after the further description that you have added in so much as you have a clear POST failure. But that possibility, however remote it may now be, still exists.

So still leaves mobo, cpu, gpu and psu as the most likely suspects.

My opinion, again based on your POST failure is that it is not the PSU

The easiest and perhaps cheapest one to deal with next imo would be the cpu and as the mechanic said

Or get yourself another cpu to replace the exiting one. Usually, at the stage that you are at now, I would be so fed up with it all that I would just buy a replacement cpu to test and be done with it. If you decide to go down the replacement rather than testing route for the cpu, I would again suggest buying something used and cheap from fleebay in the form of the lowest spec cpu available that your mobo could run.

From what you said and from having had a cpu fail on me before, I don't think that the cpu is the issue here but until such time as it is ruled out, it remains a candidate.

Good luck

 

I was going to go down the testing route, but after the system wouldn't stay on long enough for me to make the UBCD this time I hopped onto the bay of fleas and picked up an AMD Ryzen 3 2200G for £30! It's on the motherboards QVL and has an integrated GPU as a bonus, so I can use it to rule out the CPU and GPU at the same time The ironic thing is the system stayed fully functional for over 4 hours without issue as soon as paid for the CPU (it's spiteful like that), I didn't think it was going to go off but it did eventually. After an hour on YouTube I started paying a game online with friends, but we know how this ends by now

I also picked up some fiber motherboard insulating o-rings, I remember those used to come with motherboards in the olden days when the PCB's were garish colours haha

So by Friday night I should be able to cross off the CPU, GPU and if not the motherboard itself...at least the shorting theory. Is it worth picking up a PSU tester do we think? I've only ever used expensive ones, I don't know if the cheap plug in units are worth using.

 

Good thinking!

I have some in my spares cupboard, just in case!!

I don't know, I have no idea. Hopefully, someone more knowlegeable can chime in.

I remember testing an old PSU, pre sata, with molex connectors and had a multi-meter plugging the probes into the connectors and I felt somewhat uncomfortable despite wearing rubber gloves!

Good Luck

 

If the CPU is an upgrade, all well and good. I was also thinking about which direction the Event Viewer might point.

 

Still no joy, but I feel very close to the answer now. I've more or less ruled out that the motherboard is shorting on the standoffs. There's one screw not insulated, but as it's under a cover I'm not sure I even put a screw in there. As that screw lives under the GPU and a cover I'll find out tomorrow.

The CPU is unfortunately fine (it's covered by warranty) and I tested the PSU (below). According to the cheap, but well reviewed tester it passed... whatever that's worth


Tomorrow I'll try using the integrated GPU on the APU I got to rule out the GPU, I feel like the chances it's that are very slim. When it's working it runs perfectly and someone's for hours regardless of the load. While the GPU is out of the way I'll double check for Schrödinger's screw, if it exists l just leave it out, I'm sure 8 screws are enough.

As the test that fails is memory related, the CPU & RAM are not at fault and the PSU has tested okay I feel fairly confident at this point it's the motherboard. I'll still check the GPU and screw just to rule them out, but assuming neither of them are an issue that must be the answer right?

It's the opposite of an upgrade. I've got the last and best AM4 CPU, this APU is for netbooks I'm guessing. 4 cores with no hyperthreading, pinned at 100% utilisation during startup and drops to a nice low 65% idle hahaha

I can smell the finish line!

 

You are getting close.

One small wrinkle at this stage is possibly with the PSU in so much as you probably only tested it for a short while and from what you are saying the reboots sometimes don't occur for hours. Having said that, imo, it is still not a PSU issue.

CPU, GPU or mobo are the only three likely suspects....For me the hot favourite at this stage is now the mobo

I guess that you will narrow it down as soon as you replace the cpu.......

Years ago when I built a desktop system, I started buying a spare mobo just in case. Same model, same version so that if my mobo failed which it did once years ago, replacing it and getting up and running again only required little down time and no messing around with windows as it did not recognise the new part....

Good Luck!

 

Part swapping is all well and good, but I still don't see a definitive CPU test posted. Before or after budget swap.

That tool is like a "wiggy" and tests under load, yes? Anyone got a good South or North Bridge test?

 

The OP, according to his last post, has as yet not replaced the existing CPU with the 'new / budget' one which has an integrated GPU.

He did try to run a CPU test on his existing CPU but the system crashed. Therefore the part causing the problem has as yet not been identified.

 

The UBCD failed to load for whatever reason. The USB ports being wonky and losing power could have something to do with that. The machine shut down during a RAM test. It'd be interesting to see what the Event Viewer logs show. There's also the odd matter of the Q code 15...

 

I'll try to answer everything, replying directly on my phone is a pain in the ass

So I did swap the CPU with the cheapo APU and it crashed all the same. I have done stress tests while the system was cooperative, 20 minutes here, an hour there, they run and run without issues. It's really random how long the system works, all I know is it can sometimes run got hours and then it if nowhere is POST looping again. I've run the 3D Mark Time Spy CPU test, Prime95 small FFT's for around an hour. I had Prime95 and Fur mark going for about 40 minutes without issue, I don't think the crashes are caused by load because I've just watched movies and it crashes all the same.

I don't think the PSU tester puts any load on the PSU, it just measures the rails to look for obvious issues. I'd prefer a new PSU, they're a little cheaper than a mobo.

I think the Q code 15 is a motherboard issue, I've tried brand new ram and a new to me CPU. That's, 'Pre-memory system Agent initialization is started.'

I can't be 100% sure, but the crashes aren't necessarily linked to tests. I've had crashes when doing nothing, low load, extreme load and transitioning between all the those. It seems like an error that's always coming like a pothole in the road, but unlike a road you don't necessarily get to it faster by driving faster... if that makes sense.

Just now I had the system working for a little over 3 hours using the cheap CPU and integrated GPU, but it crashed too. I feel like that's confirmation that it's the motherboard right?

 

Well aside from the PSU and it is extremely doubtful that it is that, you have replaced everything else apart from the MOBO......!

 

Amazon to the rescue, I picked up an ASUS Strix B550-F to be delivered on Wednesday. My mobo has been discontinued and the one's that are available are more expensive than when it was new I'll be buying another Windows license I guess lol

I'll report back once I'm 100% sure it's fixed... or sooner or it's not haha

 

Try running your original key first. MS holding users responsible for equipment failures never sat well with yours truly. May be worth the time to activate over the phone if necessary. Or still an option...

 

Well the new mobo is POST looping too!

I've checked both ends of the PSU wires in case I missed something, unfortunately I didn't. I'm going to try the cheapo CPU in this board to make sure it's nether the new board or the existing CPU (you never know). I also stumbled on an old Haswell intel barebones that I know worked when it went into storage, I could give that a go to double check the PSU.

It now looks like it's the PSU or at least power related. The same POST loop on different mobos has to be almost impossible.

Any other theories while I triple check my work?

 

Remarkable.

Well you tried different ram so eliminates that, different cpu with integrated gpu which removes two further elements and now a different mobo. You also made sure the standoffs were not shorting!

What else is left parts wise?!

One question. When you tried the replacement mobo did you reconnect the NVME drive to the mobo's onboard NVME slot?

If so try removing the boot(?) NVME drive and have a test in Linux booting from an external medium eg usb drive

 

Thinking about it, do you have an open USB header on the system board? Like, as in never used? If you do, under Advanced Settings get into Single Port Control. It might be fun...

 

The PSU, I got a Seasonic 1000W ATX 3.0 (at least it'll be useful for years to come) and that's made no difference. I removed the M.2 drive and all the SATA drives from power/data, I removed my powered RGB hub and reinstalled the new RAM. No difference, it's not currently POSTing at all. With the low end APU I could reasonably reliability get a couple of hours each night to use it.

I'm going to reinstall the APU and new RAM tomorrow night to see what happens, that's literally a completely different system. At this point I'm clutching at straws The only thing that I haven't tried replacing is the case or fans.

I can't currently get into the UEFI to do anything, what does it do?

 

Quite incredible!!!

In your shoes at this stage, the only thing that I would do and I can't really think of anything else, is to do a barebones set up, meaning take the mobo out of the case and try booting with just have cpu, gpu and ram. Connect a key board and mouse and monitor....Hmmm I presume that you are using a usb keyboard and mouse. If so try different ones.

I have never heard of a k/b or mouse causing the kind of issues that your are having but.....If your k/b is a fancy one which draws lots of power.....

Do not connect any hard drives and see if the system starts ie do you get post? If you do reboot plug in a Linux usb stick and try and boot to that.....

 

Okay I've gone full paranoid and I'm not trusting anything until it's worked for 2 or more hours. I've installed windows on a SATA SSD and I'm running barebones on the mobo box.

Currently I'm running the new mobo, APU, WiFi, keyboard, mouse and the original PSU. I'll add the GPU and switch CPU before I try any RGB or extra USB devices. Hopefully I can find the issue or I don't and it all just works again, either way I'll be happy. Next report is hopefully a positive one!

 

Well, I was leaning toward a southbridge issue. Just my two cents...