[SOLVED] MSI Crosshair 15 B12UGZ keeps restarting when using both DIMMs (runs fine on 1 only)

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.

otringal

Distinguished
Feb 13, 2008
64
0
18,540
Hey, there! Ok here it goes... So, my brand-new laptop has a big problem: ever since I got it, it kept resetting randomly (anywhere between 10 min to 2 hours). Just so that you know, I'm not gonna ask why this happens, as I've already figured it out. BUT I really need your help with the next steps from here on out.

So, the reason for the resets is the fact that both my memory slots are occupied with 32GB of ADATA DDR4 modules running at 3200MHz, with 22-22-22-52 latency. That means I have a matched kit of 2x 32GB = 64GB in total. And the laptop says it's supporting 64GB of RAM, so I thought ok, since the resets smelled like bad RAM to me, I started testing the memory with both MemTest86, as well as Windows 11's own Memory Diagnostic Tool (Run -> mdsched.exe). And guess what, when both slots are populated, MemTest86 restarts around Test 13 (Hammer test) near the end, so it takes up quite some time (2-3h). Even better, MDSCHED (when pressing F1 for Options and selecting "Extended" mode) restarts super early on, around the 3-5% mark after no more than a few minutes. Again, both MemTest86 and MdSched fail with both DIMMs are present. If I am to remove any one of the modules from its slot, tests are passing great with no issues, no resets, no matter how I swap the DIMMs, they work fine individually, no matter what slot they're sitting in.

Therefore, I thought to myself, ok, let's do all types of BIOS upgrading/downgrading and see if anything solves it. No success. Then I started messing with my MSI BIOS settings in Advanced Mode (Left Alt + Right Ctrl + Right Shift + F2 for those who might find this useful) and with the now-unlocked THOUSANDS of options (yes, it's overkill), I played around with the clock limits, the 1DPC/2DPC settings, 2R performance optimizations, forcing 1R on both DIMMs, messing with the gear ratios (G1 and G2 work, G4 won't even post) etc., all that I could possibly think of. And still nothing. It's so strange, since it's clear that my ADATA modules are not faulty and also the motherboard memory slots themselves are not faulty, it's something about the memory controller possibly not being able to cope with both DIMMs running simultaneously for some reason.

Then another weird thing that popped up in the BIOS' memory settings is this following status page where I have some (possibly SPD) information which gives us (color coding is mine, you'll see below why):
  • MC0 CH0 DIMM0 - Populated & Enabled, 32768 MB DDR4), 2 Ranks, Unknown manufacturer
  • MC0 CH0 DIMM1 - Not Populated / Disabled
  • MC1 CH0 DIMM0 - Populated & Enabled, 32768 MB DDR4), 2 Ranks, Unknown manufacturer
  • MC1 CH0 DIMM1 - Not Populated / Disabled
If you pay close attention to the color coding that I've used above (in order to make things easier to spot) you'll see that apparently my laptop only has 1x memory channel (CH0), I am using the same DIMM0 on this only channel, but there are 2x different memory controllers (MC0 and MC1), all while the manufacturer is not even properly detected (it's ADATA, not some cheap unbranded RAM from AliExpress). This doesn't make any sense to me at all, it's physically incorrect. I mean, I only have 2 DIMMs, not 4 like on some desktops where you can risk populating the slots incorrectly for dual-channel mode, so I should have DIMM0 and DIMM1 populated, they should be running in dual-channel mode and there should be only 1x memory controller. Am I missing something here?

Any ideas, guys? Please help me understand if there's any way to fix this apart from buying new RAM and just hope for the best. Because for reasons I don't wanna get into right now (like this post already being huge), this is really not a good option for me at the moment. Also, I won't be able to get my hands on another set of 2x RAM modules to test the motherboard's brand compatibility for at least 2-3 more weeks and I really need to do a lot of work in the meantime. So, is there any modern-day expert out there who might be able help me out on this? Cause I used to think of myself as one like back 15+ years ago, but I'm not sure I'm aware of all the current trends and options when it comes to configuring the latest hardware.

Thank you sooo much!!!
 
Last edited:
Solution
Ok... I finally managed to fix this problem. And ironically, using ADATA memory - the same brand which didn't work initially. Long story short: I changed the modules with some others, but the difference is that the ones which did not work have the package number AD4S320032G22-BGN, while the ones that did work are AD4S320032G22-SGN. A quick lookup on ADATA's website reveals that the only difference is that B means Bulk, why S means Single Tray, along with the other options (R - Retail and DT - Double Tray), which I haven't tried. Apart from this general package number, all my Bulk SO-DIMMs have their individual serial numbers (the small labels stuck on the chip dies themselves) starting with...
No, I agree. I just wanted you to be aware so that in the future you might determine that perhaps MSI isn't the right choice for you or at least go into it being aware of their questionable practices.

That's just the tip of the iceberg really but I'll leave it at that because you get the idea.
 
Ok... I finally managed to fix this problem. And ironically, using ADATA memory - the same brand which didn't work initially. Long story short: I changed the modules with some others, but the difference is that the ones which did not work have the package number AD4S320032G22-BGN, while the ones that did work are AD4S320032G22-SGN. A quick lookup on ADATA's website reveals that the only difference is that B means Bulk, why S means Single Tray, along with the other options (R - Retail and DT - Double Tray), which I haven't tried. Apart from this general package number, all my Bulk SO-DIMMs have their individual serial numbers (the small labels stuck on the chip dies themselves) starting with 2Mxxxxxxxxxx, while the Single Tray ones start with 1Mxxxxxxxxxx. And I've tested a whole batch of modules.

Therefore, I'll have to come back to my earlier questions about packaging serial numbers: if the only thing that would differ would be the way in which they get packaged, and therefore offer some cost saving solution for Bulk for example, when compared to the others, then:
  1. why would they go through the extra trouble of labelling them differently (both the package and the chips themselves) and also separate them in the catalog by different categories?
  2. why would the Bulk ones exhibit problems, while the Single Tray ones be just fine (I didn't even need Double Tray which is the equivalent of a dual-channel kit)?
There's clearly a difference somewhere in the manufacturing process, they cannot be simply manufactured on the same assembly line, with the same specs, at the same location, from the same materials, one after another in identical batches and then get randomly selected for being a B, R, S or DT version. I remind you that the Bulk ones are not faulty SO-DIMMs, they work perfectly well by themselves, but my MSI rejects them if they're both running at the same time. And I tested 7 modules in different pairs/combinations, the results were always the same.
 
Last edited:
Solution

Latest posts