Question Developing Windows or Memory Problem?

[Mantrid-Drone]

Ten days ago I had a blue screen on this desktop Win7 Pro SP1 64bit PC - first time. The machine has been up and running for two and a half years. Only non-factory new parts were an additional 2 x4GB Corsair XMS DDR3 1600Mhz modules which version matched exactly the two others I had fitted. The whole PC feels a bit slower but neither of the system monitoring tools I use show anything amiss in the 4 x 4GB RAM modules.

The BS occurred when just doing my normal weekly program updates, nothing unusual. When I have a problem on any PC I scheduled a boot time CHKDSK scan and run System File Checker. Never had any reported trouble on this PC when using either but this time I had a message about SFC not being able to fix a problem. The CBS log seemed to indicate some files had been repaired but a few others could not be as the source backup ones were also corrupt.

My 'solution' was to boot into Safe Mode, run a full anti-virus/anti-malware scan (both negative) and then use an earlier restore point. That all seemed to go OK.

However when I ran SFC again I received the same sort of message about not being able to fix some files (different from the previous log) and when I scheduled another boot time CHKDSK that, for the first time I'd seen, fixed a whole load of problems including resetting the whole USN Journal.

At boot I'm often now getting CHKDSK running automatically because of a possible disc consistency problem. I've had the PC BS twice more and freeze twice too ie. nothing working, even Task Manager, forcing a power off reset. Other problems include tab crashing in my browser (never happened before) and even in Safe Mode its not immune to crashing and rebooting.

In doing so it again runs CHKDSK automatically to fix an issue or to check disc consistency. No new problems found.

Yesterday I ran Windows Memory Diagnostic just out of frustration and almost immediately it said there was a hardware issue detected. The scan went on and finished but the report didn't appeared on reboot and is nowhere to be found. I ran it again and same thing.

It should be in the Event Viewer but whilst it shows the Windows Memory Diagnostic test as being scheduled the results are missing.

Does this mean I have a RAM module problem and that might explain everything else? Or is it a more general system file issue.

Whatever is wrong I do not know where the problem is and how fix it. Any suggestions appreciated.

 

[mindless1]

I'd suspect a ram problem that has now caused file corruption and restore a backup prior to the memory being added.

If you have a spare SSD or HDD you could use that as a testbed for continuing to use the new memory config, perhaps with relaxed bios timings and a minor voltage bump.

The BS occurred when just doing my normal weekly program updates

Not sure what you mean by that but if you are having the programs update themselves, I wouldn't do that on an old OS unless having a specific problem that the update addresses. Win7 may not have been sufficiently tested.

What is the message or stop code on the bluescreen? Is it always the same?

Recheck all cables in the system in case something was disturbed while installing the memory, for example SATA cable. It is not outside the realm of possibility that brute force caused the motherboard (solder joint) to crack while installing memory, or poor board design or a weak case backplate but I would explore all other suspects first.

 

[Steltek]

You can manually run a session of the Windows memory diagnostic, or download a diagnostics ISO to create a bootable flash drive and run MEMTEST86+. if it was my system I'd pull all the memory and test the modules one at a time.

I believe there was a 3rd party utility (maybe SFCfix or something like that) which could repair corrupt Win7 system files that SFC could not. Or, once you fix any hardware problems, do a repair install to fix your Windows installation.

 

[Shmee]

Sounds like unstable RAM, given an error was found in Windows memory diagnostics. This isn't even that thourough compared to memtest86 and others.

 

[Mantrid-Drone]

I should make it clear the additional second-hand RAM modules were added about six months after I first had the PC up and running in 2019. So they've been working fine for at least two years. I've not been inside the case since then.

On the update programs matter: I update manually by choice almost everything on both my PCs. The only things I've set to update automatically are my anti-virus definitions, no choice really in those cases but I'm not a fan of having to allow that either.

I've considered testing each RAM module but although there is some sort of system file corruption that has gone on the trouble caused by whatever this is appears intermittent.

One thing I was going to ask is could having a backup folder from the other PC which contain a disc image of both another Win7 Pro and WinXP installation have been involved?

I ask because I hadn't done a SFC on that machine since I'd decided to use its 1TB storage HDD for that purpose. The reason I'm supicious is that the first time it blue screened when I used System Restore and rebooted CHKDSK ran automatically with that message about disc consistency. But it was specifically the "M" drive ie. the 1TB storage HDD that was being checked and indexing errors found.

That struck me as odd because I was wondering about the use of the System Restore point having been involved. I don't have System Restore set up to include that HDD.

 

[Shmee]

This still sounds like a hardware issue, but I would also check any and all drives to make sure they are healthy. I would use some utility that can view the SMART data, such as crystaldiskinfo.

 

[Steltek]

It is also possible that a memory issue has just recently developed. Do the memory testing with a bootable ISO burned to a flash drive so your OS file system isn't an issue.

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

If you are using spinner drives, download a diagnostic ISO from the drive manufacturer (most of them provide one) and use it to run a full test on the drives.

Finally, if your system is set up to write crash dumps when it crashes, install software like the free version of Resplendence Software's Whocrashed or Nirsoft's BlueScreenView (note: if you get a malware warning on Nirsoft's software, you can ignore it as it is a legitimate software package -- the toolkit Nirsoft uses to build some of their utilities sometimes generates false malware detections). These utilities can often point you towards a hardware or software source for crashes.

If your system isn't presently set up to write kernel dumps when it bluescreens, here is how to set it to do so.

 

[Mantrid-Drone]

Thanks all here for the advice so far.

I've installed and used use the WSSC suite for years which includes NirSoft stuff. AV warnings about some of their programs is a constant annoyance. They hate ProductKey in particular.

I didn't think to check WSSC had any useful tools in relation to these problems so thanks for reminding me about the BlueScreenView.

I'm going to try some of the suggestions here and will report back any useful results.

EDIT1

Just used BlueScreenView on the latest BS (just a few minutes ago) and the primary thing marked as the problem is a Windows OS driver notoskrl.exe at address notoskrl.exe+93ea0.

BlueScreenView


WhoCrashed


Online information suggests, if you dig down through all the suggestions about driver corruption and subsequent failed fixes on that assumption is that the actual problem, is it most likely a RAM module or modules that have developed a fault.

The problem is now identifying which are the additional 2 x 4GB used ones. I think they're the most probable culprits as they are the only non-new parts in the PC. I guess they will will be in slots 2 and 4.

Still a financial pain but I'm wondering if I should buy a completely new 4 x 4GB or 2 x 8GB pack and replace them all.
just to be safe.

I assume DDR3 modules are still available.

EDIT2

Burned Memtest86 v5.31b to CD and ran that at boot and after a bit of time it reported reported thousands of errors.

The GUI for that program, at least from disc, is not exactly intuitive but after 2 hours I realised it was just going to go on testing after 6 "Passes" I managed to end it and the errors reported started at around 9500MB.

I'm guess one of the modules chips has died. I'd assume that at 9500MB it is the third module of the 4 x 4GB. I think that means it is the first of the secondhand RAM I used. But it is isn't certain because Gigabyte number their RAM slots: 1, 3, 2, 4 to 'help' with the dual channel memory configuration.

So my two, originally new modules, are in 1 and 2 which are actually the first and third slots. But I'm hoping that in dual channel mode those are treated as one or at least as adjacent RAM volumes. That would mean the bad RAM is most likely the one in slot 3, the first of the used modules.

Being secondhand from Ebay from a private seller I doubt I'll be able to persuade Corsair of any obligation to replace it even with their lifetime guarantee.

 

[quikah]

You can run memtest with just 1 stick of ram in slot 1, repeat until you find the one with the errors. As soon as errors appear you can stop it.

You might also have developed a bad RAM slot, so go through each module individually before buying new stuff.

I RMA'd a DIMM to corsair a LONG time ago, it was pretty easy, just told them I was seeing errors with memtest. I don't recall if they requested a receipt or anything, doesn't hurt to try.

 

[Mantrid-Drone]

Thanks; yes, that individual testing is the next thing on my list to do.

I found a rather good YT video about using Memtest86 and it covered module testing and checking the slot itself too. I'm certainly not going to shell out for new RAM before doing both.

 

[Mantrid-Drone]

Just a quick update which may help any others coming here in the future with similar PC problem symptoms.

Used Memtest86 (current version) from disc and as I suspected my from 'new' RAM modules were both good and the MB's four RAM slots all OK too. Quickly identified the erroring 4GB Corsair DDR3 XMS3 module of the two (matched) other ones I'd fitted - tens of thousands of errors reported.

The other module is fine and still works as it should with the two other good ones after three full Memtest runs. But, of course, my RAM is now down to 12GB (3 x 4GB).

Interesting, a bit, is that with just the 'bad' RAM module fitted the PC still booted. I couldn't be bothered to test it any further than that but it shows the module was still functioning, so it was not clearly not entirely dead.

I do not think there is any way I'm going to find a single or double pack of perfectly matching DDR3 XMS3 4GB 1600MHz now.

So what to do? Should I replace all the RAM with 4 x 4 new 4GB ones (any make) or buy, at roughly the same price, 2 x 8GB (any make) or see if I can find 2 x 8GB 1600MHz Corsair XMS3 modules and add them to the existing best 2 x 4GB Corsair XMS3 1600MHz ones to give 24GB total?

 

[quikah]

Might as well submit a support request to Corsair and see if they will replace it. Worth a shot...

Just attach a screenshot of the memtest errors and don't mention you bought from 3rd party.

Is this a matching stick?
Corsair Memory CMX4GX3M1A1600C9 Corsair XMS3 4GB (1x4GB) DDR3 1600 MHz (PC3 12800) Desktop Memory 1.65V at Amazon.com

 

[Mantrid-Drone]

They won't, unfortunately its a "Limited Lifetime Guarantee" which excludes any not bought from a retailer. The pair of modules which included the one that is now faulty were effectively secondhand, bought from an Ebay private seller even though sold in the original packaging and as unused.

I have no receipt or proof of purchase and as I want to be honest I cannot in good conscience try it on with Corsair. In any case it is doubtful they would be able to replace that particular module or pair of modules with an exact match now.

The advice from my MB manufacturer (Gigabyte) who responded commendably quickly to my inquiry about this matter is that mixed latency modules, even of the same type from the same manufacturer are not approved. They recommend replacing the full set with matched modules as best practice for system stability

That is what I thought would be the case as many other MB manufacturers say likewise.

Good find that Amazon.com link. Spec and ID looks to be pretty much identical except my ones are shown as ".....2B....." not ......"1A....." but that might well be because it is a single stick and both sets I used were 2 x 4GB ie. separately matched pairs.

I'm in the UK and and with the shipping and likely customs charges it is simply not worth the hassle or cost.

The advice I've been given elsewhere is actually to avoid Corsair entirely and go for any brand that uses Samsung RAM. So I need to be looking for 4 x 4GB low profile modules or 2 x 8GB ones from somebody else.

 

[Shmee]

What CPU and motherboard do you have? If you were to replace the RAM entirely, I would go with Gskill probably. If your motherboard has proper controls, you could also try reducing speeds to see if that will stabilize the system.

Depending on your motherboard and CPU, a DDR3 kit could be recommended. DDR3 has been in use a long time, and different generations of platforms are speced for different voltages and speeds.

 

[Mantrid-Drone]

Gigabyte G-Z77 DS3H with an i5-3553 CPU.

As far as I know the Corsair DDR3 240-pin DIMM 1600MHz 1.65v rating only applies if you're using a XMP (Extreme Memory Profile) and I'm not. The voltage shown in the BIOS and system monitoring programs is 1.5v.for all RAM modules. I've not messed around with anything despite the capability of the Z77 ie. no over-clocking

There's no system instability AFAIK - the particular Corsair 4GB XMS3 RAM module just partially failed, probably one chip went bad for some reason.

I've been looking at alternative 16GB 1600MHz RAM solutions and yet to check Gskill's DDR3 offerings.

BTW just to conclude: I've tested the PC with 'just' 12GB of RAM (the three remaining good RAM modules) and it is very clear now that the only thing wrong with the PC was that bad one.

I've run everything I could think of since identifying and removing it and the random BSDs have gone even under heavy loads, SFC reports no issues, CHKDSK and Window Memory Diagnostic no problems either. It was all caused by the bad RAM even down to the browser tab crashing.