TWO sets of faulty DRAM - or something else?

[riskyfrisky]

Hello, I'll try and keep this short.

For the past few weeks I've been noticing issues with my PC, I posted up on these forums but got no response at the time. I was getting the odd error at boot up, which I THINK was down to a faulty SATA cable. Cable replaced, not had the issue again, one problem solved.

On top of this I've been getting a Windows explorer error when shutting the PC down. Never when using the machine, never when it's idle, or even when gaming. The error contained a bunch of numbers and 'the memory could not be read' message. I did a fresh install of Windows 10, thinking it was just one of those things, but the errors came back not long after, again only when shutting down though.

I updated the BIOS but that made no difference so finally thought I would check the RAM. Testing both sticks together (2x8gb) came up with nothing, so then I thought i'd try each stick individually. Within about 5 minutes of testing each stick Memtest came back with loads of errors.

I raised a RMA ticket with Corsair (the sticks are Vengeance LPX 3000mhz) and sent the RAM off to them. This morning DPD arrived with a new RAM kit with the exact same specs. I popped one into the PC and just loaded up Memtest and it came back with over 600 errors!

Can Memtest be wrong? Surely I cannot be so unlucky as to receive two dodgy sets of RAM? Is it possible that something is wrong with other hardware in the PC?

I am getting desperate here; I would appreciate any responses!

Thanks

 

[8 Pack]

Run the memory at 2133 defaults first and test it.

 

[riskyfrisky]

Run the memory at 2133 defaults first and test it.

Thanks for the response.

Doing it now. Which is the best way to run Memtest - the UEFI way or the traditional blue screen way? Does it matter?

 

[riskyfrisky]

Memtest has done one pass and found no errors. I used the non-UEFI interface with the memory @2133 as suggested.

What should I do now do I leave it running the test?

 

[Alexrose1uk]

I've often left it running overnight. Sucks to potentially get back replacements which can't do the expected rated speeds. Check they're being supplied enough voltage perhaps incase they're being undervolted. A slight RAM vbump may help

 

[riskyfrisky]

I've often left it running overnight. Sucks to potentially get back replacements which can't do the expected rated speeds. Check they're being supplied enough voltage perhaps incase they're being undervolted. A slight RAM vbump may help

Hey, thanks for the tip.

The sticks are rated for 1.35v, in the BIOS the voltage is 1.3530v so I think it's spot on voltage wise.

I'm currently testing the same stick using the UEFI interface of Memtest (i'm paranoid they will give me different results!) running @ 2133mhz and it's been over an hour with no errors currently. Running the stick @ 3000mhz Memtest would come back with errors within 5 minutes ervery time.

How do I work out whether this is a RAM or motherboard issue?

I've responded to the Corsair ticket but they haven't responded yet.

 

[Alexrose1uk]

Only real way would be to try the RAM in another machine/motherboard with another CPU (IMC is on the CPU), which you may not be able to do.

You could try bumping the RAM up to say 1.375V and see if that makes any difference on the stability/number of errors at 3000Mhz. If it does, don't mention you've tested this way but it would give you clarity on if maybe the memory could do it at higher voltage, but can't do it at default voltage; and therefore if the memory is duff or not.

Take pictures of the memtest results at stock voltage and keep them as evidence

A number of years ago there was a period where a load of factory overclocked RAM (during the DDR2 days IIRC) couldn't do the speeds at the nominal voltages, and actually required a bit of a bump voltage bump...loads of motherboard manufacturers then started including a slight bump at default voltages despite what the output figures were saying, to avoid being blamed for such things. Not directly relevant, but it does highlight that this sort of thing can and does happen, and memory manufacturers don't like to admit if they're being cheeky, or not QCing as tightly as they should do for OC'd bins.

 

[riskyfrisky]

Ok, do Corsair reps lurk on these forums?

Whatever, i'm past caring tbh.

I bumped DRAM voltage up to 1.3794v and ran the stick at the rated 3000mhz speed and, sure enough, Memtest crashes within a few minutes of starting. I certainly wouldn't want to increase voltage past there.

The ONLY way this RAM works stable is when I run it at 2133mhz as suggested by 8 Pack.

All I can think of is to take the RAM to a local repair shop and get them to test it in a different machine and see what happens.

Any further help would be appreciated!

 

[riskyfrisky]

I had a quick look at the motherboard manual and the recommended memory configurations for the DIMM slots is actually slots A2 & B2; I've been using A1 & B1.

I put a stick in slot A2, set the XMP profile, so voltage is back to stock 1.3530v, 3000mhz and set Memtest to work once again and it has completed a pass without fail.

Does this point to a faulty DIMM slot?

 

[kitfit1]

I had a quick look at the motherboard manual and the recommended memory configurations for the DIMM slots is actually slots A2 & B2; I've been using A1 & B1.

I put a stick in slot A2, set the XMP profile, so voltage is back to stock 1.3530v, 3000mhz and set Memtest to work once again and it has completed a pass without fail.

Does this point to a faulty DIMM slot?

No, it dosn't point to a faulty DIMM slot. It points to the user installing ram in the wrong DIMM slots
As a pair they should only be run A2 and B2, A1 and A2 are there for another pair if you wanted to run 4x8Gb of ram.

 

[riskyfrisky]

No, it dosn't point to a faulty DIMM slot. It points to the user installing ram in the wrong DIMM slots
As a pair they should only be run A2 and B2, A1 and A2 are there for another pair if you wanted to run 4x8Gb of ram.

How dare you!

What a tool I am honestly.

It's been running perfect for the past ~14 months or so; does that explain all the issues I've been having then?! I'll put them in the right slots, get a fresh install of Windows 10 on and see how it goes I guess. Flaming contraptions!

 

[Alexrose1uk]

Fingers crossed it is just a configuration issue then

 

[riskyfrisky]

Fingers crossed it is just a configuration issue then

Yeah hopefully it's just down to my ineptitude!

The other issue I've got is I don't have an optical drive in the PC, and the media creation tool hasn't been working for me properly for whatever reason, so getting a bootable Windows 10 memory stick is a hassle right now. I tried it earlier and it fails on a laptop so I need a different approach I think.

 

[Habakkuk]

Not all errors are errors - It's possible the ram is fine & it's the motherboard not reading the sticks properly.

I popped one into the PC and just loaded up Memtest and it came back with over 600 errors!

 

[freddie64]

The other issue I've got is I don't have an optical drive in the PC, and the media creation tool hasn't been working for me properly for whatever reason, so getting a bootable Windows 10 memory stick is a hassle right now. I tried it earlier and it fails on a laptop so I need a different approach I think.

Give Rufus a go. I used it as the Media Creation Tool never worked for me as well.

https://rufus.akeo.ie/

If you need a W10.iso

https://pureinfotech.com/download-windows-10-iso-without-media-creation-tool/

 

[wookiee87]

It's a long shot but with the new windows 1803 creators update you may need to flash your motherboard bios to the latest version, this in turn will improve overall performance but will inevitably help ram speeds and stability, it worked 14 months ago, but a lot of updates have dropped in that time, which set of ram were you running in the past before, it would be worth trying this again just to test.
i had terrible trouble with a new build (amd) getting ram to work and after updating bios and windows it would still crash which every setting i tried 2133 - 3466mhz eventually sent it back and rebuilt my intel system

if you've tried all the above steps my apologies for not paying attention, if not have a go and report back

 

[riskyfrisky]

Just a quick update...

Everything seems to be fine so far; a fresh install of Win 10 64 bit using Rufus to get the image onto USB stick, latest motherboard update, sticks in the right slots, and I've not encountered the issue since. Time will tell though.

I have noticed a critical 'Kernel-power' log in event viewer at least 3 times (the last being about 8 days ago). I always shut down properly so I'll keep an eye on that but I haven't noticed any power-related issues at all.

Cheers.

 

[riskyfrisky]

Error has returned

Had the PC on this morning and when it was shutting down an error popped up as before, this is the first time it's happened since I RMA the originals and reinstalled.

I don't understand why it only does it on shut down, it is absolutely flawless otherwise.

Thoughts please folks?

 

[wookiee87]

if you've rma'd the old modules for new ones and re installed windows there are only a few options left as to what could be at error, the two main i'm thinking of is either the cpu or the motherboard its self.

regarding the cpu it could be down to the imc built in if you've overclocked the cpu and uppded voltages on vcore, and vccio and vccsa going too hight on the last two will case all sorts of problems as they are related to memory overclocking, i've just looked and a max of 1.25 on both the vccio and vccsa are advised any higher will case serious problems, other motherboard manufactures may allow higher but depends on which brand you have.

in regards to the motherboard depending on how old it is, the memory issues you have could be down to loose solder points on the reverse side of the board where the ram modules are, or it could be down to dust in the slots, it wouldnt hurt to use compressed air to clean all the slots and the ram to make sure everything's spik and spam, before reinstalling the modules.

i know it will be a pain but if the above fails try and install your ram and cpu into a freinds pc and see if the symtons continue, if they do it may very well be the cpu that is the problem, just make sure your donor pc is working before making the switch for testing, this will also work as you can use your freinds cpu and ram to test your motherboard to see if that is working as it should.

hope this helps

 

[riskyfrisky]

Thanks for the response wookiee.

I’m running the cpu at stock so i’m hoping there is nothing wrong there.

Unfortunately I don’t have a donor pc or know anyone who does either and with this kind of intermittent problem it makes it even trickier to diagnose quickly.

I could try and rma the motherboard but it leaves me without a pc so I may end up buying new parts at this rate.