Conflicting Memtest86 reports

[KiwiInBlighty]

I bought some RAM to upgrade to 32Gb (Team Group Vulcan Z T-Force 32GB (2x16GB) DDR4 PC4-25600C16 3200MHz Dual Channel Kit - Grey). I had looked it up and as far as I could see, my motherboard could handle it, and it is the same generation and speed as my old RAM (just 32GB instead of 16GB). My PC has crashed several times since installing it and as soon as I swap back to my old RAM, everything is fine. I have version 7 of Memtest86 installed on my boot loader so I gave it a test and it showed multiple failures. However, when I then installed the latest version of Memtest86 on a USB drive and ran it, the test was fine, no errors at all.

Does anyone know why I'd get conflicting test results?

 

[Tetras]

However, when I then installed the latest version of Memtest86 on a USB drive and ran it, the test was fine, no errors at all.

How long did you run the tests for?

How many errors did you get?

If you only get a few errors, that it is very different to thousands, since the first suggests a config issue (e.g. overoptimistic settings), whereas the second is almost certainly faulty memory.

 

[KiwiInBlighty]

I ran it for 4 passes (roughly 8 hours) and I got 50ish failures. So if it is a failure, I should expect a lot more?

 

[Tetras]

So if it is a failure, I should expect a lot more?

Yes, I would definitely expect more errors from faulty memory in a 8 hour run. Is this with XMP enabled, or disabled? What CPU do you have?

 

[JollyJamma]

I ran it for 4 passes (roughly 8 hours) and I got 50ish failures. So if it is a failure, I should expect a lot more?

That's a very big failure since memtest isn't actually that demanding when compared to a tool like Testmem5 using the ANTA777 extreme preset.

You'd need to compare the types of NAND modules on both RAM kits as they can change over time and you won't be able to run two different kits (without tons of tweaking and or fiddling) if there is even a tiny difference.

Do another test with Testmem5 with the anta 777 extreme preset and let us know if you have any issues. What are your system specs?

 

[KiwiInBlighty]

Yes, I would definitely expect more errors from faulty memory in a 8 hour run. Is this with XMP enabled, or disabled? What CPU do you have?

It looks like I do have XMP enabled. I might try with it disabled and see if that makes a difference. My CPU is a Ryzen 5 5600X 6-core.

 

[KiwiInBlighty]

That's a very big failure since memtest isn't actually that demanding when compared to a tool like Testmem5 using the ANTA777 extreme preset.

You'd need to compare the types of NAND modules on both RAM kits as they can change over time and you won't be able to run two different kits (without tons of tweaking and or fiddling) if there is even a tiny difference.

Do another test with Testmem5 with the anta 777 extreme preset and let us know if you have any issues. What are your system specs?

Cool, I'll redo the tests with Testmem5.
I'm running a Ryzen 5 5600X 6-core CPU, Radeon RX 6650 XT GPU, 16Gb DDR4 RAM at 3200 Mhz, and a Gigabyte B550M S2H motherboard.

 

[JollyJamma]

Cool, I'll redo the tests with Testmem5.
I'm running a Ryzen 5 5600X 6-core CPU, Radeon RX 6650 XT GPU, 16Gb DDR4 RAM at 3200 Mhz, and a Gigabyte B550M S2H motherboard.

Try setting the SOC voltage to 1.1v and your ram voltage to 1.35v

 

[KiwiInBlighty]

Turning off the XMP has worked. So far no crashes. I've also tried it on, but with a slightly lower voltage, and that is fine. Thank you both for your help.

 

[McCarthy]

Turning off the XMP has worked. So far no crashes. I've also tried it on, but with a slightly lower voltage, and that is fine. Thank you both for your help.

I know it sounds a little basic, but did you clear CMOS before installing the new memory?