Why would I suddenly have to increase memory voltage

Soldato
Joined
19 Jan 2010
Posts
4,806
Hi all.

I've been running 32gb (2x, 2x 8gb sticks) in my pc for a couple of years now and a couple of months ago I started getting random crashes, mostly when the pc is left idle.

This steadily got worse and worse until crashes and freezes would happen a couple of minutes after booting up.

I've replaced the PSU and done all the required testing such as reinstalling windows, installing windows on 2 different SSDs, updating bios, reset bios and cmos, ran memtests, removed overclocking, unplugged everything and ran as a bare system, try each ram stick individually in different slots, reseated CPU, repasted GPU and CPU and still I would get crashes.

In the end I just threw a curve ball in and enabled xmp1 in the bios and strangely the pc ran for much longer before crashing/freezing. I then decided to up the voltage on the ram in xmp1 to 1.3900v and the pc has been running for 10 hours without crashing.

Does anyone have any idea why this would be?
Is my memory on the way out?
Is my mobo faulty?
Is the memory controller in the CPU on the way out?

Specs are:

Cpu Intel i9 9900k
Ram 32gb gskill trident z f4 3200c16 8g tzr 2 packs of 2x 8gb.
Mobo Asus hero vi z390
PSU Asus 850w Thor
Cooler i115 Corsair platinum
GPU asus strix 1080ti oc

Thanks for reading.
 
Soldato
OP
Joined
19 Jan 2010
Posts
4,806
One thing I haven't considered is doing the cooler up too tight. I've read that this can cause instabilities with ram. Maybe a very slight loosen might make all the difference
 
Soldato
OP
Joined
19 Jan 2010
Posts
4,806
I should also make it clear that I have also left the bios at defaults without xmp selected at all so the ram runs at a slower speed and I would still get crashes. This seems really strange to me.

The pc has been running for over 20 hours now without crashes when ram set to 3200 and voltage at 1.3900v.
 
Permabanned
Joined
11 Jan 2019
Posts
3,214
Location
bedlam
Memory control could be degrading

could be a sodding windows update. I remember when I have x99 after a windows update I couldn’t game at xml speeds for about 3 months
 
Soldato
OP
Joined
19 Jan 2010
Posts
4,806
Memory control could be degrading

could be a sodding windows update. I remember when I have x99 after a windows update I couldn’t game at xml speeds for about 3 months
Thats the thing. I cant run it at standard speeds at standard voltage and I can seem to game for hours without an issue. The crashes happen when it's left idle or if I'm on the desktop
 
Permabanned
Joined
11 Jan 2019
Posts
3,214
Location
bedlam
Thats the thing. I cant run it at standard speeds at standard voltage and I can seem to game for hours without an issue. The crashes happen when it's left idle or if I'm on the desktop

defiantly ram? What error codes do you get is it a grad crash all the time if do you get blue screens.

have a look in the error logs and see what you can find. If it just at idle I can’t see it been ram most ram problems pop up with there loaded up with data
 
Soldato
OP
Joined
19 Jan 2010
Posts
4,806
defiantly ram? What error codes do you get is it a grad crash all the time if do you get blue screens.

have a look in the error logs and see what you can find. If it just at idle I can’t see it been ram most ram problems pop up with there loaded up with data

They are all windows hardware errors (whea). I've had 6 so far today
 
Permabanned
Joined
11 Jan 2019
Posts
3,214
Location
bedlam
They are all windows hardware errors (whea). I've had 6 so far today
We need a little more info than that. Post up some error reports and or error codes

it could be the cpu is degrading cpu the ram the motherboard it’s hard to point at one component without the parts to switch it out and test. The easiest way would be buy a 8gb ram stick and test.

Error logs will help but they dont tell you the problem. The error could point at ram but it could be the traces on the motherboard or the memory controller
 
Soldato
OP
Joined
19 Jan 2010
Posts
4,806
We need a little more info than that. Post up some error reports and or error codes

it could be the cpu is degrading cpu the ram the motherboard it’s hard to point at one component without the parts to switch it out and test. The easiest way would be buy a 8gb ram stick and test.

Error logs will help but they dont tell you the problem. The error could point at ram but it could be the traces on the motherboard or the memory controller
I will add more info on the error when I can. At the moment I've stripped the system right down and downloaded memtest86. After I created the usb the system froze and I got the same hardware error bsod.

Memtest86 is currently running and I'm now hoping for an error as I'm only running the one stick of ram.

If it doesn't error then it has to be then CPU or mobo right?
 
Soldato
OP
Joined
19 Jan 2010
Posts
4,806
Double post
NekUCpt
 
Last edited:
Permabanned
Joined
11 Jan 2019
Posts
3,214
Location
bedlam
Memtest86 is currently running and I'm now hoping for an error as I'm only running the one stick of ram.

cpu errors can throw up errors in memtest.

Run a full test if you get errors run the same stick on the other channel and retest.
When testing remove all overclocks on cpu and ram, run spd not xmp
 
Soldato
OP
Joined
19 Jan 2010
Posts
4,806
cpu errors can throw up errors in memtest.

Run a full test if you get errors run the same stick on the other channel and retest.
When testing remove all overclocks on cpu and ram, run spd not xmp
Yea I've removed anything like that. I've reset the bios and cmos before testing. I'm a bit lost with this one as I've been working on it almost 2 weeks now
 
Permabanned
Joined
11 Jan 2019
Posts
3,214
Location
bedlam
Yea I've removed anything like that. I've reset the bios and cmos before testing. I'm a bit lost with this one as I've been working on it almost 2 weeks now

fingers crossed look like ram. Run the test with the stick in A1 or B1 then again in A2 or B2 if you get errors there is less chance it’s the board or control on the cpu.

but the only thing that will give you a 100% is replacement of hardware, buy a ram set or RMA the one you have


Edit: at any point did you have a silly cpu overclock like very high voltage’s or did you ever push the ram hard even just for a day?
 
Soldato
OP
Joined
19 Jan 2010
Posts
4,806
fingers crossed look like ram. Run the test with the stick in A1 or B1 then again in A2 or B2 if you get errors there is less chance it’s the board or control on the cpu.

but the only thing that will give you a 100% is replacement of hardware, buy a ram set or RMA the one you have


Edit: at any point did you have a silly cpu overclock like very high voltage’s or did you ever push the ram hard even just for a day?
I've overclocked the CPU before but not for long using the ai overclock setting on the board. The ram has ran at xmp1 1.3500v 3200 from day one.

3 passes and zero errors. I've put the ram in slot A2 as that's what the mobo manual advises to do with a single stick
 
Soldato
OP
Joined
19 Jan 2010
Posts
4,806
The pc was left on all night. 8 passes on the config that crashes/freezes and zero errors.

Any ideas how I can test the memory controller on the CPU?
 
Last edited:
Permabanned
Joined
11 Jan 2019
Posts
3,214
Location
bedlam
Did you test just one stick? Test the other for a few hours.

there is a ram testing program that’s a lot faster than memtest but right now I can’t remember the name al have a look on my pc lattter
 
Back
Top Bottom