New build - stability issue

BoomAM

BoomAM

BoomAM

Soldato
Joined
5 Jul 2003
Posts
16,206
Location
Atlanta, USA
Afternoon everyone,
Its been a while since I've posted here, so i'm not sure if this is the right forum for this, so feel free to point out and i'll re-post and report this thread for removal. :-)


I'm having a strange issue where a random time into games, the screen goes black and the monitor thinks there's no signal.
The system is otherwise on, fans spinning, lights inside on, etc; but is completely unresponsive.
I have to flip the power on the PSU to get it back to life as apparently the 4-8 second power switch/reset button do diddly squat.
I've run stress tests on CPU (Prime95), GPU (Furmark), both (AIDA64) at the same time, and it passes them with flying colors.


One thing that I've noticed, and it may be a red-herring, is turning the nVidia control panel into debug mode which resets the clock speeds from whatever overclock is in place, allowed one of the games (DX:MD) to run for a much longer period without the issue happening (then i closed the game and went bed, so i dont know if it was a total fix or not).


I've read about similar issues, but not mid-game, but it may be related?
https://reddit.com/r/nvidia/comments/8p7t6g/geforce_hotfix_driver_version_39818/
http://nvidia.com/object/nv-uefi-update-x64.html

Other tests/diagnosis done:
  • I changed cable to a DisplayPort cable, no difference after 30mins of Deus Ex: Mankind Divided.
    I turned off DX12 mode (DX:MD) and reverted to DX11 mode, no difference.
  • Checked event viewer, no logged entries however.

System itself is less than a week old, and temperatures are all very low for the system, latest drivers, updates and firmware's are also installed:
  • Ryzen 7 2700 w/ Corsair H100i -- max temp seen, 56*c
  • 16Gb G.Skill 3200
  • Asus ROG Strix X470-F Gaming -- max temp seen, 47*c
  • Asus Strix 1080Ti - max temp seen, 62*c
  • Corsair RM750W PSU
Only other thing i can think of trying, is changing the power cables to the GPU from being a single PCI-E cable with Y-end to 2x 8-pin, to dual PCI-E cables with Y-end, but only using the primary PCI-E on each cable.
However with this PSU seemingly only having 1x 12V rail, im not sure it would help.

Anyone seen this before?

Thanks in advance all!
 
To add more to this:
I took my 1080ti back to the store to get it replaced, ran furmark and it crashed in the same way after 5mins or so - so it was worse!

I updated to the latest hotfix driver version, 398.18, and retested, furmark ran for a solid 15mins, max temperature i saw was 72*c.
I then gave DX:MD a try for about an hour, rock solid. Max temp i saw was 67*c.

Driver issue perhaps?
 
...and its reared its ugly head again.
Even though i can play hours of DX:MD, in the last 2 days I've had two occurrences of it happen, this time when just doing random things in Windows itself, like browsing the net.
Fresh out of ideas at this point...
 
orbitalwalsh said:
Set ram speed down to 3000hz
Click to expand...
May i ask, what is the reasoning behind this recommendation?
Have you had this issue before?

D.Roberts said:
Hi

1 - Right click desktop and go to Nvidia control panel
2 - now go to adjust desktop color and settings
3 - you will now see a drop down box named content type reported to display (its set to auto)
4 - Change that to I think its the second optoin in the drop down list I cant remember the option name sorry but try messing with that.

As said my issue was not 100% like yours but worth a go I suppose at this stage.
Click to expand...
There's no drop down named 'content type' or anything set as auto.
On this screen its just 'other application control colour settings' or 'use nvidia settings'.
I had a look around the other settings in there, nothing similar...

Cheers.
 
Last edited:
D.Roberts said:
Hmmm how odd I have a 1080 none Ti my Bro has the Ti and he has the drop down box I'm sure he does shall check with him tomorrow but I'm sure I did same tweak to his pc as his had same issue as mine.

What driver version u useing mate?.
Click to expand...
The 398.18 hotfix driver, seemed better/less likely to happen than the 'latest' 398.11 driver.
& yes, if you could it would be very much appreciated. :)

Its starting to sounds more and more like a driver issue at this point...however orbitalwalsh's comment on RAM speed has me intrigued too.
 
Last edited:
orbitalwalsh said:
Just helps eliminate possible issues . Ryzen and ram speed aren't the greatest , specially when XMP is an intel thing to.

Best mothod is just to set everything to stock on bios and see if your still having issues above - then means it's down to drivers/GPU and not your core system
Click to expand...
Its using the D.O.C.P profile out of the box. Not overclocked.
I did have issues getting D.O.C.P initially working, but it cleared up when i changed the slots.
Are you suggesting removing D.O.C.P so it reverts to 2133 (the out of box detection)? Or reducing the speed to the AMD max ram speed spec of 2933?

Smffy said:
1. BIOS update board
2. Clean install windows to eliminate ‘driver issue’
3. It’s Ryzen so most likely due to RAM if above fails, or manual overclocking need adjusting (v, ratio etc)
Click to expand...
1. Already latest.
2. System is less than 2 weeks old, no clean install needed. Other than the nVidia hotfix, everything is 1 driver version ever installed.
3. No OC in place, using D.O.C.P.

Thanks for the input and suggestions here everyone btw! :)
 
orbitalwalsh said:
should be able to selec DOCP and then change speed from 3200 to 3000 manually , Located a little lower down in the menu with the title MEMORY FREQUENCY - should be a drop down box and change from 3200 to 3000
Click to expand...
I've seen that option, will drop it to 2933 (the Ryzen 'official' max) as one of my tests.

also, what is the AI Overclock Tuner set to? some boads have the ability to enable DOCP/XMP as well as auto overclock the CPU to best suit- normally leave this off/opt out/standard
Click to expand...
Its set to default. No OC is in place on anything.
 
I saw that RAM calculator, but i dont see what its achieving?
I'm inputting the settings of the RAM, to tell me the settings of the RAM for optimizations and overclocking?
Surely defeating the idea of running at stock?

Or am i missing something?
 
Smffy said:
2. Yes but if you feel drivers are borked reinstall and there’s your answer.
Click to expand...
A feeling isnt an answer though ;-)

This is the answer to your problem potentially, select XMP and it will load the correct RAM speed and voltage all going well. You are the free to overclocking the CPU independently leaving BLK at 100. Make sure you reset the BIOS to defaults first to get rid of any changes.
Click to expand...
ok, im trying to use the Ryzen DRAM Calculator right now to work out some safe settings to use, but i have no idea how to use this thing as neither Thaiphoon nor CPU-Z give me all the values i need. Anyone got any usage tips?

I also have 8x instances of HCI MemTest on the go, 2Gb chunks on each to check for errors there too.

##EDIT##
This video helped:
https://www.youtube.com/watch?v=xWJD5P0x9Ow
Currently letting the HCI MemTests hit 100%+ before rebooting and trying the settings from Ryzen Calculator.
Only thing i cant work out is if i have Hynix AFR or MFR RAM...
 
Last edited:
Well the initial HCI MemTest test's passed, but i managed to find a thread on reddit with a guy with the same RAM, so i've duplicated his timings and retesting...seeing as Ryzen RAM Calculator even with an import still misses key bits of info that i need to get every setting. :-(
 
Well that failed about a minute into the HCI test. :-p
So i've reverted back to DOCP/3200, gonna run a longer test this time, get to 500% testing, at that point i can effectively take the memory out the equation i think...
 
ok, so a few hours later, HCI MemTest at 500%+ on each of the 8 instances, no errors.
That's with the RAM at DOCP/3200 mode. Considering most web searches seen to suggest anything from 200-400% is stable, and only a few people being a bit #### and running to 1000%, the 500% i hit without issue to me says the memory being at fault is a very very slim chance right now. Anyone in stark disagreement with that assessment?

Once the test completed, i installed the just released 398.36 drivers (clean install option) and rebooted. Lets see how it goes...still interested in the setting that d.roberts mentioned too as a possible avenue.

Also, interesting, first few posts....thoughts?
Reading the whole 9 pages seems to suggest a motherboard bios issue with multiple monitoring apps and/or apps accessing the smbus at the same time causing it. With a suggestion that another Asus motherboard got a BIOS update that fixed it.
 
Last edited:
Smffy said:
Honestly take off DCOP as per my post and use over clocking mode XMP which was an option on my Asus Ryzen...
Click to expand...
I'm pretty sure that the HCI MemTest (not to be confused with MemTest86), proves my memory stability beyond some freak occurance.
It's allegedly the most stressful and thorough of all the memory testing apps available.

So as it passes with DOCP on, with pc3200 set, my next step would be to drop to 2933 if the latest Nvidia driver doesn't help.

unless the BIOS has changed and there is no longer an XMP in the overclocking tuner.
Click to expand...
There isn't.
Auto/manual/docp

if you still think drivers what’s the harm in installing fresh windows to test/prove.
Click to expand...
I'm not against the idea, but other options arnt exhausted yet, and as the system is barely a few weeks old with the only driver updates being from Nvidia which improved, but not entirely fixed, the issue thus far, a reinstall isn't on the table right now.

The new Nvidia driver and the reported smbus/monitoring app issues are logical routes to take, and easily tested out too.
Those and that Nvidia control panel option that has been mentioned.
Willing to listen to other ideas too, but an OS install on such a fresh system is jumping the gun in my books.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom