Interesting that Cubase is working stably even with heavy projects, but watching videos and gaming is not stable.

It would be a really interesting [and usable] test to put that old Radeon inside instead of RX480. RX480 can draw a lot of power. If it works with the old card then it really is a power supply to blame for it.

 

thank you for the answer.
well, isn't it a sign that it could be the GPU too? I mean, using a lower TPU card fixes both the theoric problems: broken GPU (1) and too much power consumption from the PSU (2). So I actually could not say which was the problem.
And, I could not replicate the actual power consumption, with a smaller card. So maybe if I switch PSU and then I'll put the RX 480 back on, I can have the same problem as before (IF and only IF it's a GPU problem).

or am I wrong?

 

I made a power consumption calculation with this tool, here is the result: https://outervision.com/b/ABuHOn

any thoughts?

 

You can check PSU and general stability with a software like
https://www.ocbase.com/

With occt, you can stress the system CPU GPU without any other software running.
If it crash, you know it is an hardware problem.
If thermals go crazy, you know where it is and who drain power.

"Power calculation" is not really a problem : it is STABILITY related
A lot of PSU just fluctuate too much. When there are "voltage drops", components simply crash.

It can be RAM, CPU, GPU ... everything can crash due to voltage drop.
I know it from experience : 20 years computer hardware troubleshooting
And it happened to me personally too.

I suspect it is your GPU getting too hot AND/OR dropping your PSU
Use OCCT and check thermals/voltages

 

Thank you for this! I'll run it this evening for 1 hour. Do I need to select something inside of it, or should I leave everything at default? Maybe something related to GPU or PSU, don't know. Thank you!

 

Ok, I'm unable to run this test for more than 20 seconds.

The first 2 times I opened it and I got 2 BSOD with SYSTEM_SERVICE_EXCEPTION and dxgkrnl e ntfs.sys.
The 3rd time I opened OCCT, pressed play to start the test and in something like 10 seconds, it found more than 350 errors.
Then it crashed.
Then BSOD with DRIVER_IRQL_NOT_LESS_OR_EQUAL and ucx01000.sys

btw, I'm running this tests without any USB connected, with a wireless Logitech set (not gaming kb+mouse I usually use), only system nvme ssd to run the OS and nothing at all. CPU, RAM, GPU and kb+mouse. I disconnected all the fans too, except for the CPU fan, of course. running without the glass panel side, not having temperature so high. between 30 ~ 50 ° for both CPU and GPU. yes the GPU arrives at 50° in idle, sometimes.

what should I do to read the OCCT errors? thank you
here is a screenshot of OCCT after 7 seconds and 10 errors
https://imgur.com/a/xDkNn3p

 

Last edited: Dec 27, 2020

hi guys, happy new year!!

I resetted BIOS with factory settings, now I don't have all the BSODs anymore.

just one a couple of hours ago, with a driver error I've never seen before:

do you have any thought? thank you !!

 

up

 

Do the memory test to be sure.
https://www.memtest86.com/download.htm

and check HDD S.M.A.R.T, with CrystalDiskInfo
http://crystalmark.info/en/download/

 

thank you, already did some time ago. no issues.

always load lots of the RAM on Cubase, never had any issues on it.

 

If the PC you built has onboard video...

Prior to doing the GPU check ...

0. NTFS.sys is the file system driver for Windows NT, this usually suspects to file table issues, partition issues. So test that SSD, or HDD out ( many free tools available)
Right click your SSD or HDD in "Computer" properties, then Tools, click Check then scan this drive. Windows will look for any file errors.

##GPU part##

First check the tower for any loose cabling or plugs, check the SSD plugs, make sure the ram is seated properly

1.Test the system by removing your GPU and using the onboard video adapter

Boot into Windows 10, yes it will default to the MS display adapter since no driver is installed.
Run the PC for about 10 minutes while you run some applications ( they may be slower or give your some errors due to using the built in MS video display driver.

The reason why is the GPU uses a lot of power, so we to eliminate any power issues (Power supply not big enough) and also Driver issues for that GPU vs that Windows 10 installation.

Check the system out and see if the Blue Screens go away ( or go down at all)


2. Use task manager/ startup and see what is starting up ( uncheck everything that you believe may be running that could cause issues, uncheck any nvidia services for example)

3. Is to test the SDD for read speeds to make sure it is working correctly

Last step it to run each software you use individually on the Onboard video card ( if it is available)

If things improve put back the GPU and redo these tests with it ON ( disable the onboard video adapter)

 

Last edited: Jan 14, 2021

thank you for the answer. sadly my CPU does not have onboard GPU. I have to try with another similar (by power consumption) GPU, I'll try the other advices however. thanks!

 

Well I had an issue similar to yours last year with a FX 6300, I know it is an old socket with nothing in common with yours, but what solved my problems after 3 months bashing my head against a wall and switching GPU's and RAM sticks was reseating the CPU, when you do this make sure you switch the thermal paste and check for bent pins in the motherboard socket. Might be a shot in dark but maybe it will solve your problems.


PS: Also check your XMP settings against what speeds your memory calls for!