Memory shenanigans or?

[Anthos]

Hi all,



I build a pc back in dec with an asus dark hero, 5950x and crucial ballistix 3600C16 (2x16gb dual-rank)

All quite well since, but noticed the past month or two that when I go to my pc in the morning (I tend to leave it on overnight), will kinda be in a mostly non-responsive state. Like you can at times click things (based on what's already cached I suppose) but further clicks and so on will tend to be unresponsive. If it happened recently the music might still be playing but once the cached amount if finished then it will stop. If I ask it to restart it will go the logon screen and never actually go further. I have to manually shut it down. This thing would usually happen like 50% of the times overnight but very rarely in the first 5-7 hours I'd use it before bed. As an aside I never really had any crashes or any problems in any games, light work, heavy work, browsing etc...

Anyway, this weekend thought that it might be time to have a look at what might be going on. I mostly thought about possible software issues but obviously first is easier to make sure the memory is fine.

Started with TestMem5 with a specific config file which I don't remember now its name. Finished with some errors. I was like ok, great. Put everything on bios on stock (including mem) and then:

1. Booted off and single stick. No TestMem errors.
   
   
2. Booted off the other stick. No errors.
   
   
3. Put both of them together. 1 error
   
   
4. Put both of them at XMP/DOCP speeds. Around 5 errors I think.
   
   
5. Then during the night I ran memtest86+ for near 10 hours with everything at DOCP/XMP speeds and both sticks. No errors
   
   
6. Tried again today with one stick at DOCP/XMP (these settings on subsequent tests as well). No TestMem errors.
   
   
7. The same with the other stick. No errors.
   
   
8. Both sticks. No errors.
   
   
9. Ran again with the 2 sticks. 7 errors.
   
   
10. Ran again. 1 error.
    
    
11. Ran again. No errors.
    

Now in theory pretty much anything that ever gives any sort of error is just not right. The only reason I want to make sure that problem lies with the memory is that when I bought my ram it was at the time that Crucial switched the kits of the model I got from Dual Rank to Single Rank. So if I get a replacement and the problem is not the memory then I kinda got downgraded. Obviously it's not the most important thing it's just that it would suck to lose "some" performance when the problem doesn't really lie there.

Anyone got any idea?

P.S Also due some unrelated reasons some months ago I had done some test runs with TestMem and as far as I can remember I wasn't getting any errors except when I was overclocking my ram further.

I also quickly ran TestMem in another windows installation that is mostly barebones once when first started the tests and it also seemed to come up with some errors (I didn't do however any further testing in that OS).

P.S2 Between writing all of this I had ran yesterday the windows memory diagnostic tool in extended mode and it found no errors after being ran for 20 hours (both sticks at DOCP/XMP)

 

[EsaT]

You could try upping memory voltage slightly from that specified 1,35V.
(say 1,36/37V)
Also could try Ryzen DRAM calculators values for SOC voltage.

And to check OS for integrity use System File Checker.
https://support.microsoft.com/en-us...ndows-10-365e0031-36b1-6031-f804-8fd86e0ef4ca

And check Event viewer and reliability history for errors timed after known functioning and before that "freeze".

 

[Anthos]

You could try upping memory voltage slightly from that specified 1,35V.
(say 1,36/37V)
Also could try Ryzen DRAM calculators values for SOC voltage.

And to check OS for integrity use System File Checker.
https://support.microsoft.com/en-us...ndows-10-365e0031-36b1-6031-f804-8fd86e0ef4ca

And check Event viewer and reliability history for errors timed after known functioning and before that "freeze".

Yeah I did think about playing around with the voltage to see if it stabilizes. I checked generally in the past (not exactly at freeze time) to see if there's any WHEA 18/19 errors and I couldn't notice any. For the time being kept the pc continuously on for the past 2 days and didn't get any of those weird freezes yet, I ll try to have a look at the event viewer when it does. Shouldn't be too hard as I think it freezes the clock in the corner as well so it will be easy to know when it happened. Silver lining I guess.

 

[Quartz]

Anyone got any idea?

Have you approached Crucial for advice?

 

[Anthos]

Have you approached Crucial for advice?

Not yet, but once I get some more time to test some more I'd probably will. Although I'd most likely assume that they 'll have me run a few tests, then tell me that the windows memory diagnostic is finding no errors so buzz off or just have me RMA. Rarely have I found the tech supports to actually do something else than that.

Edit: one more thing I am thinking is whether it has to do with the motherboard itself. Maybe the memory controller. Because in testmem I never got an error running just with a single stick. Haven't tested though to see if I get a lockup in windows.

 

[Joxeon]

Have you updated the bios since December as the early versions were a bit hit and miss?.

 

[Tetras]

The very small number of errors and the inconsistency makes me more think of a config issue, if the memory was faulty I'd expect a lot more than that.

 

[Anthos]

Have you updated the bios since December as the early versions were a bit hit and miss?.

Yes, I've been updating the bioses as they were coming along except the last one which I just updated to a few days ago.

The very small number of errors and the inconsistency makes me more think of a config issue, if the memory was faulty I'd expect a lot more than that.

Yeah, it's not really straightforward which is why I am not exactly dashing at RMAing the memory. I ll probably try to reset the bios first completely just in case an old value clinged at some point and it's causing issues (because even when you update or load previous settings, some can still get stuck). I am almost working all through the weekend extensive hours so as I won't be needing the pc I ll just load up a fresh windows installation and keep it running throughout and see if any of the weird locks appear to rule out any software issues and then just take it from there I guess.

 

[adrichardson81]

Are they definitely in the right slots? Is GDM enabled?

 

[Anthos]

So I've loaded a fresh windows installation and did a hard reset on the bios on Friday morning and kept the pc running constantly. Up to this morning when I ended the testing the windows installation ran fine without locking at any point (around 95 hours in total).

And I obviously I thought I had finally figured it out (likely stuck settings from soft resetting bios and not hard resetting).

Loaded my normal windows. Did a TestMem 5 run. Still gives errors. Hell, even more than before probably. Haha, man what the f...

Are they definitely in the right slots? Is GDM enabled?

Yeah they are in the right slots, I don't remember at the momen about GMD, I'll have to double check. EDIT: GMD is enabled

 

[mickyflinn]

If your getting eerors in mem rest you either need to change the settings manually and add more volts then retest, did you try this ?

Did you test them individually

Or send them back.

 

[Anthos]

If your getting eerors in mem rest you either need to change the settings manually and add more volts then retest, did you try this ?

Did you test them individually

Or send them back.

Yeah, I 've moved now to trying to change values manually and see where that leads me.

As I 've said originally, I've tested them indivindually multiple times and I never got a single error that way.

 

[mickyflinn]

Yeah, I 've moved now to trying to change values manually and see where that leads me.

As I 've said originally, I've tested them indivindually multiple times and I never got a single error that way.

Set them manually and add more volts you could also loosen the timmings slightly to cl18 just to see if that helps.

 

[Anthos]

Set them manually and add more volts you could also loosen the timmings slightly to cl18 just to see if that helps.

I used the REC settings from Dram calculator based on the rated speed (3600), SAFE and MANUAL settings aaaand got a blue screen (1.4V [docp is 1.35V]). Kinda unexpected because in the past I did run this memory at 3800 uncoupled from the IF and it was actually stable (I think, I benchmarked it but not sure now if I actually TestMemed it).

I ll probably flash to an older bios as well just in case any of the new ones have something messed up. Nothing else changed in the past few motnhs and a few tests I did early this year I didn't get any errors (then again, I don't get now every single time either so could be just luck).

I 'll try to retest at base speeds (2667) even though I still got some errors when I tested this way a week ago.

EDIT: small amount of errors even on 2667 which by default comes with loose timings. I ll try an older bios and see if there's any difference.

 

[mickyflinn]

I used the REC settings from Dram calculator based on the rated speed (3600), SAFE and MANUAL settings aaaand got a blue screen (1.4V [docp is 1.35V]). Kinda unexpected because in the past I did run this memory at 3800 uncoupled from the IF and it was actually stable (I think, I benchmarked it but not sure now if I actually TestMemed it).

I ll probably flash to an older bios as well just in case any of the new ones have something messed up. Nothing else changed in the past few motnhs and a few tests I did early this year I didn't get any errors (then again, I don't get now every single time either so could be just luck).

I 'll try to retest at base speeds (2667) even though I still got some errors when I tested this way a week ago.

EDIT: small amount of errors even on 2667 which by default comes with loose timings. I ll try an older bios and see if there's any difference.

Do you know anyone where you could test your ram in there pc for errors that would give you more if a dfinitave answer especially if you think its a bios issue .

 

[Anthos]

Tried 6m old bios, base speeds (2667), TestMem errors... sigh..

Do you know anyone where you could test your ram in there pc for errors that would give you more if a dfinitave answer especially if you think its a bios issue .

Unfortunately not. And my previous build is literally a dozen years old of the DDR2 era so that's kinda dead end.

 

[mickyflinn]

Tried 6m old bios, base speeds (2667), TestMem errors... sigh..



Unfortunately not. And my previous build is literally a dozen years old of the DDR2 era so that's kinda dead end.

Looks like your going to have to send them back.

 

[Anthos]

Looks like your going to have to send them back.

Yeah, I ll probably need to do something about it.

Decided to run corecycler because I did quite a lot of times in the past when testing the core curves and seeing if its stable etc and it failed with a rounding error on one of the cores (stopped the test after the first error appeared). Definitely something is not right.

Used a 6m old bios and it didn't made a difference so I guess one more thing ruled out.

EIT: Ran each indivindual stick @ DOCP/XMP speeds on its usual DIMM 3 times (around an hour each) on TestMem and no errors. Ran a full corecycler cycle and no errors. So for whatever is worth it seems these errors appear when both sticks are on.