Question Really odd problem with 3900x system.

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,434
136
OK, I usually am the master of these things, but here is a new one to me. So I have a 3900x, a X370 Taichi motherboard 4 x 8 gig ram, an AIO, and a 2060 video card, with an 850 watt PSU. For a year, all was fine, then the last few weeks, after about 2 days, it just shuts down. I unplug the PSU for a few hours, plug it back in, and then power up. All is fine for a few more days. I changed out the PSU to a brand new gold EVGA 850, no change, so thats not it. The temps are fine after boot, even@100% load.. But about 2-3 days later it will shut down again.

Motherboard getting overloaded with a 3900x ? Old VRM's that are a little weak ?
 
  • Like
Reactions: Drazick

TheGiant

Senior member
Jun 12, 2017
748
353
106
OK, I usually am the master of these things, but here is a new one to me. So I have a 3900x, a X370 Taichi motherboard 4 x 8 gig ram, an AIO, and a 2060 video card, with an 850 watt PSU. For a year, all was fine, then the last few weeks, after about 2 days, it just shuts down. I unplug the PSU for a few hours, plug it back in, and then power up. All is fine for a few more days. I changed out the PSU to a brand new gold EVGA 850, no change, so thats not it. The temps are fine after boot, even@100% load.. But about 2-3 days later it will shut down again.

Motherboard getting overloaded with a 3900x ? Old VRM's that are a little weak ?
I had this problem once and only one move worked
Completely dissasembly, out of case everything
Run a while out of the case if it happens again
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,434
136
I had this problem once and only one move worked
Completely dissasembly, out of case everything
Run a while out of the case if it happens again
Remember, without touching anything or powering off, it ran for a year, no issues, no reboots.
 
  • Like
Reactions: Drazick

TheGiant

Senior member
Jun 12, 2017
748
353
106
That is freakin odd.... But of course the problem is odd. I don;t really have space to do that, too many computers in the house.... 2 dining room tables filled.
yeah special cases require special treatment
I know it sounds like dissasembly, put out of the house and howl at the moon at midnight but it worked
it was an old core2 duo system
and I remember that I cleaned the dust and cleaned the case in bathroom with shower
witchery :)
EDIT:
if you try running it out of the case, prepare prime95/whatever high load, start it and hear and smell
my experience is high freq sound=problem, smell can point you to damaged parts of the board...
if nothing try to run the system with only one RAM module
 
Last edited:

VirtualLarry

No Lifer
Aug 25, 2001
56,226
9,990
126
Yeah, I've seen cases before, of PCs that "just needed a little TLC". Take them apart, dust them off, re-paste, re-install the RAM, maybe a little "DeOxIt" on the contacts / sockets, etc., and BLAMO, back to full-strength working.

Of course, this may not apply if the VRMs somehow degraded. I had some ASRock B350M Pro4 boards. Really early-model AM4 boards, really cheap too. I stuck some Ryzen 1600 6C/12T 14nm CPUs in there, and tried to OC to 4Ghz, didn't quite get there, but it really put a strain on the VRMs. They had a weak SoC VRM section, too, that wasn't used for much with the Ryzen Zen1 CPUs, but when the APUs came along, the graphics used the SoC voltage plane, and they didn't have enough current capacity, so running APUs like a 2200G, at least in my tired and worn boards, didn't work so well. (I thought that I could re-use them, and "put them out to pasture" as APU-using browser-boxes. Didn't work out too well. They were able to work with some Athlon 200GE chips, though, those were low-power enough.)

Anyways, give it some TLC, maybe check the VRM temps afterwards, maybe take the VRM heatsinks off and re-paste THOSE, maybe that's what the real problem is, or the southbridge heatsink.
 

Iron Woode

Elite Member
Super Moderator
Oct 10, 1999
30,854
12,353
136
OK, I usually am the master of these things, but here is a new one to me. So I have a 3900x, a X370 Taichi motherboard 4 x 8 gig ram, an AIO, and a 2060 video card, with an 850 watt PSU. For a year, all was fine, then the last few weeks, after about 2 days, it just shuts down. I unplug the PSU for a few hours, plug it back in, and then power up. All is fine for a few more days. I changed out the PSU to a brand new gold EVGA 850, no change, so thats not it. The temps are fine after boot, even@100% load.. But about 2-3 days later it will shut down again.

Motherboard getting overloaded with a 3900x ? Old VRM's that are a little weak ?
It is possible that the board could simply be dying.

Did you test the ram to make sure there is no issue there?
 

Ajay

Lifer
Jan 8, 2001
15,332
7,792
136
Yeah, stress test the memory, GPU, and CPU - one at a time. That's how I'd start. Run HWInfo at the same time, so you can check temps and power levels.
What's the motherboard, BTW?
 

VirtualLarry

No Lifer
Aug 25, 2001
56,226
9,990
126
I have a "spare" (LOL) ASRock X370 TaiChi here, not sure if they made an X370 TaiChi Ultimate, or if that was just the X470 version, but mine has 10 SATA6G ports on it. Whichever version that is. Still BNIB, from Newegg, maybe 8-12 months ago, I think.

LMK, if you need a replacement, make me an offer. (Preferably a somewhat generous one, considering how hard AM4 boards are to get these days. Plus, I still need to pay my rent for June.)

Edit: I also have an MSI B450 Tomahawk MAX, and an Asus B450-F ROG STRIX Gaming ATX mobo available in my FS thread, if one of those would interest you more. But I think that the TaiChi probably has better VRMs. Although this incident may make you re-think that observation.

Edit: I bought it nearly a year ago, to re-build my unRAID server, and also bought a PCI-E controller card with 8 SATA ports off of Amazon renewed, in preparation for a full re-do, but ... honestly, I'll probably never get around to doing so. (I have a number of discrete NAS units already set up.)

Edit: I looked up my Newegg invoice. I bought it "Open Box", and it came in the retail package. So, possibly, unused, and original purchaser had second thoughts about the board? Anyways, it LOOKS new, I haven't opened the box yet. Would offer with a 30-day non-DOA guarantee if you wanted it, Mark.
 
Last edited:

Fardringle

Diamond Member
Oct 23, 2000
9,184
753
126
Are there any errors/crash reports in WhoCrashed or BlueScreenView?

Maybe try a different power outlet, or a UPS, just to be sure it's not an incoming power problem from the wall outlet.

Also, if you decide to take the system apart for a cleaning and reassembly, check the motherboard for scorch marks around any of the components, and make sure none of the capacitors are bulging or cracked.
 

Magic Carpet

Diamond Member
Oct 2, 2011
3,477
231
106
It does sound like a mobo problem to me, I’d RMA it if possible. Shut downs happen due to fail safe built-in , something‘s definitely amiss.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,434
136
Yeah, stress test the memory, GPU, and CPU - one at a time. That's how I'd start. Run HWInfo at the same time, so you can check temps and power levels.
What's the motherboard, BTW?
Its running linux, not much in the way of debug software. I shut down the 2060, and its back running 23 threads of Rosetta. We will see how long that lasts.
 
  • Like
Reactions: Drazick

EXCellR8

Diamond Member
Sep 1, 2010
3,979
839
136
No X370 Taichi "Ultimate" version that I'm aware of. Might be time to bust out the multi meter near the VRM's if you care that much... like you stated, degradation perhaps?

do you have another memory kit to try?
 

Hans Gruber

Platinum Member
Dec 23, 2006
2,092
1,065
136
The latest windows updates have been crashing a lot of systems recently, including mine. Phantom crashes. I cannot speak for systems running Linux right now. My 3600 even crashed when I ran it stock everything for a few days.
 
  • Like
Reactions: VirtualLarry

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
@Markfw

I would start stress testing the CPU and looking for odd vdroop problems, high VRM temps, or anything else that would indicate problems with power delivery. Try to peg the CPU at the highest package power you can get (around 142-146W). Don't assume something like Prime95 will automatically get you there.
 

DisEnchantment

Golden Member
Mar 3, 2017
1,590
5,722
136
I don;t really have space to do that, too many computers in the house.... 2 dining room tables filled.
What about bathroo... emmm. :blush:

I have similar problem. The PSU acted weird. When I connect the SATA SSD to the end connector it shuts down. Connect it to the middle it works. That too on a Seasonic Gold+ PSU
I just switched PSU and everything is fine again.
 

mopardude87

Diamond Member
Oct 22, 2018
3,348
1,575
96
That is freakin odd.... But of course the problem is odd. I don;t really have space to do that, too many computers in the house.... 2 dining room tables filled.

You think that is odd? I had a old Gigabyte not post for ages, i thought it was bricked. I found one tiny itty bitty piece of dust in a ram slot, i cleared it as it looked almost like a flake of iron almost. When i mean small, i had the mobo up to my face and that is when i could see it. Overall it looked very clean, just that little deal caused all that damn trouble. Once removed,the thing posted worked fine for years, till the build got sold.
 
Last edited:
  • Haha
Reactions: richierich1212

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
My money is on RAM or VRMs. My advice is to do what you mostly advice people with RAM issues to do; set the RAMs to DDR4 2133MHz and see if that makes a difference. Zen 2 is quite new and I'm not sure many people are running them with all slots full for a whole year at full blast so the IMC itself could also be an issue. Plus, you're working with four different modules of RAM so the probability pf any of those flaking out is also quite high.
 

moinmoin

Diamond Member
Jun 1, 2017
4,933
7,619
136
Its running linux, not much in the way of debug software.
There should be logs that may or may not be helpful. What distribution are you using? If it's something using systemd you can get the log through journalctl, e.g. get a list of boots journalctl --list-boots and pick the one you want to look at like the previous boot journalctl --boot=-1, maybe just list errors so the list is more concise journalctl --boot=-1 -p err. If your Linux happens to not use systemd chances are that you can find a text file at /var/log/kern.log.
 

mopardude87

Diamond Member
Oct 22, 2018
3,348
1,575
96
My money is on RAM or VRMs. My advice is to do what you mostly advice people with RAM issues to do; set the RAMs to DDR4 2133MHz and see if that makes a difference. Zen 2 is quite new and I'm not sure many people are running them with all slots full for a whole year at full blast so the IMC itself could also be an issue. Plus, you're working with four different modules of RAM so the probability pf any of those flaking out is also quite high.

This, i had a bunch of weird no post issues cause my motherboard in the signature simply did not support speeds of 3600mhz. I had to drop it to 3200mhz and since then zero issues. I got all 4 ram slots populated. My old 16gb kit ran at 3600mhz without issue though.

Not sure if this will help Mark though, but yeah its still worth a mention in case anyone else may read this.
 

reqq

Member
Feb 26, 2020
31
35
61
The latest windows updates have been crashing a lot of systems recently, including mine. Phantom crashes. I cannot speak for systems running Linux right now. My 3600 even crashed when I ran it stock everything for a few days.

blue screen crash or just reboots without blue screen?
 

myocardia

Diamond Member
Jun 21, 2003
9,291
30
91
Mark, since this system is not air-cooled, I would add an 80mm or larger fan, directly over the VRMs. Just because the VRMs didn't need more cooling when they were younger/newer, doesn't mean they won't benefit from it now. It's fairly easy to just wiretie a fan down, and is also a dirt-cheap fix, assuming that is the problem. If you want to make sure that is the problem, before spending the money for a new fan, just leave the side panel off, and aim any household fan in the general direction of the VRMs. The problems you are describing are the kind that happen from a slow buildup of heat, IMO.

edit: Also, don't forget that using an AIO also doesn't cool the RAM any. Consider strapping a fan over your DIMMs as well, although I'd use a 120 or 140mm fan over the RAM. Neither of these fans need to be high airflow models, BTW.
 
Last edited:
  • Like
Reactions: Magic Carpet

Hans Gruber

Platinum Member
Dec 23, 2006
2,092
1,065
136
blue screen crash or just reboots without blue screen?
Non blue screen crashes. The computer hangs/freezes and usually doesn't even restart itself. I found an article that said the new Windows update was a serious problem. It's basically an all new windows without a clean install and significant beta testing. I am sure they will figure it out eventually.