Question BSOD during windows 10 install

[MercuryUK]

Hi all,

I'm at my wits end with this so decided to turn to you guys to see if anyone can help.

Did a new build about a year ago, main specs are:

• MSI Tomahawk B450 Max
• Ryzen 5 3600
• MSI RTX2060 Ventus XS 6GB
• 16GB (2x8) RAM - Corsair Vengeance DDR4 3200MHz
• Samsung 970 EVO Plus 500 GB NVMe SSD

Been running fine for about a year, couple of days ago it started getting BSOD seamingly randomly, and differnt types each time.
It got quickly worse, where it wouldn't stay up long enough for me to start looking into it.

Decided a reinstall of windows was the way to go, and that's where the real trouble started.

I've tried to install using USB media about 30 times, it will freeze, BSOD or instant reboot each time during the install, normally during the 'Getting files ready for installation' step, sometime about 10% sometimes above 80%.

Often has different reasons on the BSOD; page and irql more common than most I think, but it often just stops without an error

Steps I've taken to see if I can figure out the problem:

• Run memtest86, 4 passes, took over 3 hours, found no errors
• Tried with only one memory module (tried both, different slots)
• Tried a different SSD
• Tried switching out the graphics card
• Only other hardware attached is a Wifi PCI card, tried taking that out
• Tried creating a new USB, then installed windows on another machine using it, worked fine.

Anyone got any idea that will help diagnose it?

Many thanks in advance

 

[Grobe]

Running Memtest should last longer than 3 hours to ensure there is no bad ram stick?

Have you overclocked the cpu ?

Shoot in the dark : Try disabling ram xmp profile.

 

[MercuryUK]

Yeah, I thought it was supposed to run longer (never needed to run it before). But I was watching it a lot of the time, seemed to be working as it should, first pass was more than an hour, but finished the forth in 3 hours 19.

Nope, not overclocked in any way, never has been.

XMP is already disabled, never did work well. The bios recognized the ram as 2200Mhz so I have to manually set to 3200. Ran like that since day 1 without problems.
Since I've been having this issue I've tried with oprimised defaults (which includes ram at 2200), doesn't fix the problem.

Thanks for the suggestions though.

 

[Grobe]

Have you checked that the RAM modules in fact is listed on the motherboard list of supported ram types?

 

[MercuryUK]

Have you checked that the RAM modules in fact is listed on the motherboard list of supported ram types?

Yeah, the RAM is supported, and listed as such. I've run this rig for a year before this started happening.

 

[Grobe]

Please have Memtest to run overnight or so, to properly rule out RAM as the cause of failure.

The things left are Motherboard, PSU and CPU. If every components are of similar age, I'd say that the motherboard is more likely to be the issue.

Something you may check is if the thermal paste of the CPU is properly laid - that is, you may try to re-apply the thermal paste just in case the cooler somehow has jerked itself loose over time. Shouldn't happens, but still.

 

[MercuryUK]

Update:

Thanks for the suggestions

• Tried multiple BIOS versions; the one that ran fine with for a year, the latest, the oldest, and v32 (which has reported as the most stable for some people)
• I've tried with optimized defaults (RAM is at 2144) and setting RAM manually to it's rated 3200 (as it was configured for the last year)
• I've tried a Linux install, and a Linux boot USB. both freeze up. The bood USB did get to desktop, but then froze
• Can't run memtest86 for more than 4 passes without paying for pro

New conclusions:

• I think the bios version is a red herring; although there are reports of this motherboard and CPU having problems with more recent bios, all seem to report that old bios are fine. Also, this is a very common cpu/mobo combo so loads run it with no issues (I did for a year)
• Looks like it is hardware, as Linux has issues as well as the windows install media.

One additional odd thing to note; most of the time the SSD is recognized as UEFI, but occasionally it isn't. This can been seen in bios, and also during install (if it is UEFI windows creates 4 partitions, if not then it creates 2). It's now only recognizing it as bios not uefi. So a tried the other SSD again. It recognizes as UEFI, but as before I still have the same problems and can't install

Anyone got any more ideas? I'm thinking that this is just a doorstop now. Once I have the money I'm going back to Intel and never looking at AMD again.

 

[Grobe]

Can't run memtest86 for more than 4 passes without paying for pro

You probably got the "fake" one (never trust the first search result on google, they've probably payed google to get promoted in web search). Use duckduckgo search engine instead. DO NOT PAY FOR THIS !

This is the proper web site - Memtest86+ official home page:

Memtest86+ | The Open-Source Memory Testing Tool

Memtest86+ is an advanced, free, open-source, stand-alone memory tester for 32- and 64-bit computers (UEFI & BIOS supported)

www.memtest.org

Alternatively, you may already have Memtest included in your Linux ISO image, and is possible to select under boot.
Or you can download the Ultimate Boot CD, where Memtest is included.

 

[Impala640]

Update:

Thanks for the suggestions

• Tried multiple BIOS versions; the one that ran fine with for a year, the latest, the oldest, and v32 (which has reported as the most stable for some people)
• I've tried with optimized defaults (RAM is at 2144) and setting RAM manually to it's rated 3200 (as it was configured for the last year)
• I've tried a Linux install, and a Linux boot USB. both freeze up. The bood USB did get to desktop, but then froze
• Can't run memtest86 for more than 4 passes without paying for pro

New conclusions:

• I think the bios version is a red herring; although there are reports of this motherboard and CPU having problems with more recent bios, all seem to report that old bios are fine. Also, this is a very common cpu/mobo combo so loads run it with no issues (I did for a year)
• Looks like it is hardware, as Linux has issues as well as the windows install media.

One additional odd thing to note; most of the time the SSD is recognized as UEFI, but occasionally it isn't. This can been seen in bios, and also during install (if it is UEFI windows creates 4 partitions, if not then it creates 2). It's now only recognizing it as bios not uefi. So a tried the other SSD again. It recognizes as UEFI, but as before I still have the same problems and can't install

Anyone got any more ideas? I'm thinking that this is just a doorstop now. Once I have the money I'm going back to Intel and never looking at AMD again.

It sounds like either a PSU failure or maybe a mobo failure. What kind of PSU do you have cause my PSU was the first thing to go on my rig and from the things your saying I believe it is a PSU issue.

 

[MercuryUK]

Alternatively, you may already have Memtest included in your Linux ISO image, and is possible to select under boot.
Or you can download the Ultimate Boot CD, where Memtest is included.

Thanks. Took me a little while to work out how to do this, so for anyone else that comes across this:

• Memtest86+ doesn't work with uefi so in your boot order put 'legacy' USB drive first.

Ubuntu then had the memory test in the menu, so running if now and will leave it for a long time.
It's a long shot though, as the errors happen quickly every time for me, 4 passes probably would have found it if it was the memory. Not ruling it out, but I think the memory is ok

 

[MercuryUK]

It sounds like either a PSU failure or maybe a mobo failure. What kind of PSU do you have cause my PSU was the first thing to go on my rig and from the things your saying I believe it is a PSU issue.

It occured to me that it could be the power supply. It's an EVGA 650 GQ, purchased at the same time as my build; about a year ago.

Any way to validate this other than getting a new one?

With all the other tests done, think this is down to PSU, CPU or Mobo, but no idea how I find out which one?

 

[Grobe]

Any way to validate this other than getting a new one?

Yes and no, depends on what equipment you have at home. If you have a oscilloscope and can use it, that is a nice tool to test if the voltage are straight.

If not, then you must test another PSU in order to rule out a bad psu.

Also, you may try to take a look inside the psu to see if you can spot a faulty capacitor (looks like it's about to boil over). But don't open, even if it's not plugged in the wall, it may still hold high voltage inside.

 

[Calvin7]

Memtest86+ has not been updated since 2013 and is not recommended on anything but old legacy BIOS motherboards.

Check the warranty on your power supply. You may have a claim.

Try borrowing another PSU to test with.

 

[helper800]

I would rule out PSU before you conclude it is either the CPU or MOBO. In my experience, it is very rarely a faulty CPU. Motherboard failure rates seem a bit lower than PSUs.

 

[Grobe]

Memtest86+ has not been updated since 2013 and is not recommended on anything but old legacy BIOS motherboards.

Do you have a source (other than passmark inc.) for this?

Or better : a reliable alternative to Memtest86+ that is free to use (not freemium variants) ?

 

[MercuryUK]

Update:

I've now tested with another power supply, same issues occur so doesn't look like it's PSU
Memtest86+ run overnight, no errors, so does't look like it's the memory

Any ideas for away to identify if it's the cpu or mobo?

 

[Impala640]

Update:

I've now tested with another power supply, same issues occur so doesn't look like it's PSU
Memtest86+ run overnight, no errors, so does't look like it's the memory

Any ideas for away to identify if it's the cpu or mobo?

If your really gangster you could buy a b450 tomohawk max. Test it then return it from either a local shop or a big company