Question Thousands of errors in Memtest86

steffeeh

Reputable
Feb 12, 2016
265
1
4,815
I've built a PC for a friend and I'm currently borrowing it for doing some testing to make sure the hardware is working as it should.
This includes running Memtest86 with the memories at stock, so I haven't fiddled anything with the memory, they're at 2133MHz JEDEC standard. I made sure the DIMMs were properly seated the best I could.

The RAM is a pair of G.Skill Trident Z RGB LED DDR4 3200MHz 2x8GB (F4-3200C16D-16GTZR), paired together with an AMD Ryzen 5 2600, on an ASUS ROG Strix B450-F Gaming motherboard with BIOS version 2301.

I first let a session run during the night, and when I checked it in the morning there had been 1785 errors, all in test #7 at pass #3. The session ran for a total of 7 hours, and finished 8 passes.
I then just launched another session while I was at work etc, and in this session there had been 1790 errors, att in test #10. I couldn't see which pass for some reason, but I would suspect somewhat early in the session as well. This session ran for a total of 11 hours with 12 full passes.

Some notes:
- I have updated the BIOS on the mobo and then loaded optimized defaults.
- The CPU ran pretty toasty, mostly at 85C, but one time I shortly saw it at 92C (no errors occured during the 92C temperature). I've seen that CPU's can get quite warm with Memtest86, however this is really hot. I think I may need to reinstall the cooler.
- The PC currently runs on an SSD that we took from a laptop, and we haven't reinstalled Windows yet as we didn't get the time. This means the drivers are all mismatched, but it doesn't feel like it should affect Memtest86 as it runs outside Windows.
- The RAM has RGB LEDs, so some of the power from the DIMM slots goes to the LEDs.
Could any of this affect the stability results in Memtest86?

Do we need to RMA the RAM sticks?
 

DMAN999

Honorable
Ambassador
^ +1 to what Mandark said.
I have that exact same RAM (running at 3400 MHz) and mine ran without errors on Memtest86+ and TPU's Memtest64 multiple times while I was OC'ing them.
And until you get that CPU temp down I would NOT be stressing the system at all.
 

steffeeh

Reputable
Feb 12, 2016
265
1
4,815
Thank you for your replies!

I have the DIMMs installed in the slots recommended by the manual, which is slot 2 and 4. I tested each DIMM separately in slot 2 but found no issues, though I only ran 4 passes this time (however the errors seemed to occur before completing 4 passes when I discovered them).
Could it be that the RAM is actually fine, but slot 4 on the board is bad?
I also swapped place with the DIMMs so the DIMM in slot 2 was in slot 4 and vice versa, but got errors again.

The DIMMs weren't hot when I touched them, and the CPU temperature in BIOS is at around 35C, so at least it's not 85C when idle, but I still agree that 85-90C is too hot. I'll test the temperatures in Windows with RealTemp and try some moderate load and see how the temps behave.

EDIT: @DMAN999 I saw now that the cooler you have in your build list in your sig is the same cooler we used. How was your experience with it? It was very difficult to install in my experience, and I feel I may have installed it wrong somehow, causing too high temperatures (if I indeed get too high temps in Windows). How are your temps with it? Sure your 2600 is OC's but it would still be a helpful comparison to see if our temps are way too high.
 
Last edited:

DMAN999

Honorable
Ambassador
My CPU idles at anywhere from 26-30 C depending on room temperature (67 to 72F). It is usually 28 or 29C.
And it maxes out around 55-56C after gaming (Far Cry 5/AC Odyssey/or similar games set on Ultra) for 3 to 4 hours.
I do use the Ryzen Balanced Power Plan in Windows so at idle my cores drop down to 1.55 GHZ and only raise as needed under load.
Also IIRC the max I saw when stress testing my various OC attempts was around 65C.

As far as installation went, I installed it with the MB out of the case and it installed very easily.
I also used Arctic Silver 5 Thermal Paste instead of the MX-4 paste that came with the cooler because I already had a tube.
But that shouldn't make any difference as they are both decent pastes.
I applied a line to each heat pipe as recommended by Arctic.

You very well may want to remove it and re-install it just to make sire it is making proper contact with the CPU and that the paste is spread evenly and is not too thick or too thin.

And yes it is definitely possible that you have a bad RAM Slot and if so I'd definitely RMA that MB ASAP.
Since it's an ROG I believe they will send you a replacement before you send yours in too.
 
Last edited:

steffeeh

Reputable
Feb 12, 2016
265
1
4,815
Now I'm just confused.

I went on to check the actual temperatures for some different loads within Windows just get a final view on the cooling and such. I fired up Aida64 and ran a CPU stress test, and to my surprise the temperatures were only 54C max in both Aida64 and HWMonitor.
So I then fired up Prime95 small FFTs to see how it behaved when I pushed it harder, and to my surprise it was even lower at 52C.
But I let it continue to run since I know the first 15 min of Prime95 small FFTs is not the hottest so I wanted the test to really kick in, but after 10 min or so the screen goes black and all fans suddenly run at max speed and I can't shut down the computer. I try different ways of shutting it down, but eventually I'm forced to turn off the PSU (not even the killswitch or reset button works).

What is going on here? Are my temperature measurements all broken and I hit TJ max?

I still haven't reinstalled Windows as I first wanted to put the SSD back into the laptop and see if I can link the Windows license etc so it works in the new PC (which I can't do until tomorrow), so the drivers are still mismatched, but this is just weird. Not even a bluescreen, just a shutdown and max fan speeds.

EDIT: Forgot to mention that I tried loading optimized defaults in BIOS again, this time first setting a custom setting beforehand, to really force reseting the BIOS, and settings that I haven't changed showed up, settings that haven't showed up before when just selecting load optimized defaults - so they got reset as well, and then I ran Memtest86 with both DIMMs for 9 hours, and no errors were found. I'll do it again tonight just to check again.

EDIT #2: After some googling I see it's not too uncommon that Memtest86 reads abnormally high CPU temps, especially with Ryzen 2.
 
Last edited: