Memtest - Bad RAM stick

[jamieboo]

Hello folks

A system I built a couple of years ago (12700K, Asus Tuf z690 board, 128gb Corsair 3600 DDR4 (4x32GB. Matched kit. On QVL.)) has generally been performing pretty well, but with the occasional bsod - which were infrequent enough that I could try to ignore.
Lately my work has got a bit more intensive, using a larger portion of my RAM, and the bsods have increased.
I turned off XMP and ran memtest, one stick at a time, all were fine until I came to the final stick which had thousands of errors within a couple of minutes.
Bugger.
Oh well, hopefully easy to replace.
But what are the chances that originally the ram was fine, and the mobo slot the stick was in is bad and somehow fried that stick?
I could have course test that slot, but if what I'm testing for turns out to be true, I risk frying another stick.

I look to you much more savvy folk for guidance!

Thanks

 

[Quartz]

But what are the chances that originally the ram was fine, and the mobo slot the stick was in is bad and somehow fried that stick?

Small, but not zero. Do check the power being supplied to the RAM.

 

[Tetras]

But what are the chances that originally the ram was fine, and the mobo slot the stick was in is bad and somehow fried that stick?

A lot more likely it was bad originally and went more bad, and/or has slowly corrupted your OS, hence worsening symptoms.

 

[jamieboo]

Small, but not zero. Do check the power being supplied to the RAM.

Thanks Quartz
Excuse the rudimentary question, but what is the best way to check the power being supplied to the RAM in that slot?

 

[Quartz]

Thanks Quartz
Excuse the rudimentary question, but what is the best way to check the power being supplied to the RAM in that slot?

Check the BIOS.

 

[jamieboo]

Check the BIOS.

Ok, well basically 1.20v with XMP off and 1.36v with XMP on, which I gather is fine.
But is there any way to check if something crazy is malfunctionally putting more voltage through to a specific slot?

 

[Rroff]

It is possible the BIOS settings aren't correct for the RAM - some RAM sticks will have more of a margin of tolerance than others so possible slight manual tweaking will sort it, or the RAM could be bad. Unfortunately when it comes to RAM timings and voltages, XMP, etc. the standard and implementations is very hit and miss.

 

[jamieboo]

It is possible the BIOS settings aren't correct for the RAM - some RAM sticks will have more of a margin of tolerance than others so possible slight manual tweaking will sort it, or the RAM could be bad. Unfortunately when it comes to RAM timings and voltages, XMP, etc. the standard and implementations is very hit and miss.

I guess so. But here's 3 sticks passing flawlessly for hours, then one stick spewing thousands of errors in the first couple of minutes.
Of course there are a million variables, but the memtest result does seem to fairly definitively point to a single dodgy stick. I'm just concerned if I RMA and get a pristine new kit and put them in my machine, if one of the slots was faulty, it might immediately scramble whatever stick is inserted there and I'd be faced with the same problem all over again.
This is why I'm trying to find out if there's any way to diagnose if a single slot could be faulty to the extent that it ruins any sticks inserted there.
As I said before, I could simply try another stick in the slot and see if it does get ruined - but I'd rather find another way.
Is this even a thing though?
I mean all the voltages and timings seem fine according to BIOS. And they're general right? There are no individual slot by slot settings that I'm missing?
Is it possible for a single slot to scramble any stick inserted there?
If so, how can I diagnose?

Thanks

 

[Tetras]

Is this even a thing though?

If the board had some kind of short, or other form of damage, I would expect the stick to just be dead, to be honest. So far as I know, the power delivery is not per slot, like..., there's not one mosfet per slot, so I'd be surprised if there's something which is degrading only one memory stick. That said, there are boards that kill memory sticks, so it is not impossible. There can also be a significant difference in temperatures in some cases, like the stick closest to the CPU might be hotter.

Since this is the first stick you've had problems with and the PC was never 100% stable, I would assume the most likely explanation (that you have a bad stick) rather than the least likely, that the slot itself is bad.

 

[jamieboo]

If the board had some kind of short, or other form of damage, I would expect the stick to just be dead, to be honest. So far as I know, the power delivery is not per slot, like..., there's not one mosfet per slot, so I'd be surprised if there's something which is degrading only one memory stick. That said, there are boards that kill memory sticks, so it is not impossible. There can also be a significant difference in temperatures in some cases, like the stick closest to the CPU might be hotter.

Since this is the first stick you've had problems with and the PC was never 100% stable, I would assume the most likely explanation (that you have a bad stick) rather than the least likely, that the slot itself is bad.

Thanks mate! Your razor is better than Occam's!