Drivers & Software

[johnhardyiv]

I just got a PowerColor Red Devil Radeon RX 5600XT 6GB.

Using 20.2.1 drivers.

Tried to run userbenchmark.exe from userbenchmark.com, today's version, 2.9.5.0

When I hit the graphic performance section I saw the planes, the cubes, but on whatever the next test is the entire system shut down (?) with no error or warning.  Had to cold power it back up, and it left a 140GB temp file in c:\userbenchmark

I had set the drivers to 'standard' not gaming or egaming.

--

I've had additional graphic corruption on Links 2003 and Visual Pinball using this combo which I posted earlier about.  Did not have these issues w/ my prior AMD card, an 8GB R390.

Corsair Obsidian Series 750D Airflow Edition, Full Tower ATX Case CC-9011078-WW
Corsair Hydro Series H115i Extreme Performance Liquid CPU Cooler Cooling CW-9060027-WW
Intel Boxed Core I7-6700K 4.00 GHz 8M Processor Cache 4 LGA 1151 BX80662I76700K
Corsair Vengeance LPX 16GB (2x8GB) DDR4 DRAM 3200MHz (PC4-25600) C16 Memory Kit - Black (CMK16GX4M2B3200C16)
Corsair Vengeance LPX 16GB (2x8GB) DDR4 DRAM 3200MHz (PC4-25600) C16 Memory Kit - Black (CMK16GX4M2B3200C16)
SAMSUNG SD590 Series S27D590P Black High glossy 27" 5ms (GTG) HDMI Widescreen LED Backlight LCD Monitor PLS Panel 300 cd/m2 DCR Mega Infinity (1000:1)
SAMSUNG SD590 Series S27D590P Black High glossy 27" 5ms (GTG) HDMI Widescreen LED Backlight LCD Monitor PLS Panel 300 cd/m2 DCR Mega Infinity (1000:1)
ASUS SABERTOOTH Z170 MARK 1 TUF LGA1151 DDR4 DisplayPort HDMI M.2 USB 3.1 Z170 Standard ATX Motherboards
Microsoft Windows 10 Pro 64 Bit Retail from MSFT
Corsair AX Series, AX860, 860 Watt (860W), Fully Modular Power Supply, 80+ Platinum Certified
Samsung 970 EVO 2TB - NVMe PCIe M.2 2280 SSD (MZ-V7E2T0BW)
Samsung 850 EVO 2 TB 2.5-Inch SATA III Internal SSD (MZ-75E2T0B/AM)
PowerColor Red Devil Radeon RX 5600 XT 6GB

[Update 3/4/2020] Installed a new Corsair H115i XT cooler last night, had the temps drop 30 degrees under load and no crashes in userbenchmark.  Thanks to all in the thread for the help!

[mstfbsrn980]

Have you tried turning off the Windows 10 automatic game mode feature? You can disable it with the following regedit content.

Windows Registry Editor Version 5.00

[HKEY_CURRENT_USER\Software\Microsoft\GameBar]
"AllowAutoGameMode"=dword:00000000

[johnhardyiv]

Yes, game mode was disabled.  With this card and driver it's 100% repro unfortunately. I can't seem to find the userbenchmark log to see what's happening.

[mstfbsrn980]

Have you made changes with WattMan settings? Have you tried an old driver?

[johnhardyiv]

I don't have wattman installed.  I just have a few older games I deal with, many of which are having trouble w/ this card/driver combo, laugh.

I tried w/ 20.1.3 and same thing happened.  Were there earlier radeon packages that supported the 5600XT? Thanks -

[mstfbsrn980]

5600 XT new production. For this reason, I think you do not have many driver options. Unfortunately, the cause of graphic disruptions is the driver. For now, it is difficult to find solutions to graphic disruptions. But your system should not be shutting down. The shutdown issue may be caused by the PSU. When the PSU is overloaded, it may be protecting itself. CPU temperature may also be increasing. The motherboard may be protecting the system. In other words, you may have a shutdown problem due to similar problems.

What are your PSU?
Can your system successfully perform the OCCT CPU test for 10 minutes?
Can your system successfully perform the OCCT PSU test for 10 minutes?
What are the maximum CPU and GPU temperatures in OCCT PSU test?

[johnhardyiv]

Ran OCCT 5.5.3 'Power' test for 10 mins. successfully.  It's the combined CPU:OCCT and GPU:3D test.

CPU temps were max 100 degrees at 4.5Ghz, GPU temp max 74 degrees. Resolution was 1024x768

Interestingly, if I grabbed the 3D test circle thing by its title bar and tried to move its window left or right, the system warm boots like above.

My hardware is listed above, the PSU is a Corsair 860.

Corsair AX Series, AX860, 860 Watt (860W), Fully Modular Power Supply, 80+ Platinum Certified 

[mstfbsrn980]

You have no GPU problems. I think you have a CPU problem. It is an unacceptable temperature of 100 degrees. I think your CPU should go up to 4.5 comfortably but its temperature is not normal and the motherboard may be protecting the system. You should try thermal paste replacement and check fans for your CPU.

Also I think you don't have a PSU problem. Check the CPU core voltage and increase it if necessary for OC....

[johnhardyiv]

FWIW, I went back to default stock settings on the 6700K (4Ghz) just to check, and I still get the crash/warm boot when moving the 3D test window.  Guess I'll hang out and wait for 20.3.1.   Have a few more weeks to return the card to Amazon. Thanks -

[mstfbsrn980]

There may be a problem with your CPU cooling system. Open your PC case and check if the fan is spinning. Perform thermal paste replacement for the CPU. It is not normal for the processor to rise to 100 degrees. If the result does not change, you would consider returning the card. Because if CPU cooling is problematic, different graphics card will not solve the problem.

[johnhardyiv]

The only delta in the system is the new 5600XT.  Interior fans are all spinning as expected.

The prior R390 was working properly even w/ the 20.2.1 drivers. System idles at 34 degrees and a 4.5Ghz OC.

Cooling is via CORSAIR Hydro Series H115i AIO Liquid CPU Cooler.  If that pump was malfunctioning would the system even stay up for 2 minutes?  It's curious.  I do appreciate your follow ups.

[mstfbsrn980]

You have explained in detail everything that makes you think that the problem is caused by the GPU. But driver errors do not cause a restart problem or shutdown. Driver error usually causes freezing or blue screen error. You are using a different hardware that uses same way (so PCIe) to which the graphics card is connected. You have a M2 disk. The presence of this disk can cause a similar problem. So...

Have you tried updating Intel chipset drivers?

Have you installed Intel PCIe controller driver?

Have you tried the system without M2 disk?

Have you tried new Windows installation with an USB stick?

In short, I still think that a new graphics card will not be corrupt. Hope you can fix the problem...

[kingfish]

Here are the previous drivers for your card > https://www.amd.com/en/support/previous-drivers/graphics/amd-radeon-5600-series/amd-radeon-rx-5600-s... 

[johnhardyiv]

Thanks, this issue was present w/ 20.1.4.  

[johnhardyiv]

I did a clean install of Windows 10 from USB and applied a new batch of thermal paste.

The userbenchmark.exe at least no longer crashes out.  Unfortunately the temps are still too high when running OCCT up in the 90s.  Could be the cooler itself.  I may look at replacing it with a newer model as it's going on 4+ years now and I don't have the Corsair USB Link to attach and check its functioning.   Thanks -

[darkio]

Make sure you are running the pump at full speed, should be about 3k rpm. I saw a 20degree drop in temp when running the 115i pump at 3k rpm instead of 2k as it is per default. 

[johnhardyiv]

Thanks, how can I determine or control that w/out the Corsair USB Link cable?  This is an ASUS Sabertooth z170 Mk 1 motherboard, the AIO is hooked to the w_pump header.  Thanks -

[darkio]

Are you missing that cable or? I would think you can adjust the speed in the bios if you are missing that cable, but cant say that for sure.

Btw, the latest drivers ( 20.2.2 Optional)for the 5600XT seems to be alot more stable on my RD 5600XT so i would suggest updating the drivers too.

[johnhardyiv]

Hi, yes I was missing the USB Link cable so I couldn't crank up the pump.  Corsair thinks the impeller/pump may have gone south too.  I installed a new h115i XT yesterday and temps dropped 30 degrees, laugh.  So I'll see about resolving this thread.

[johnhardyiv]

Just another note, I picked up an XFX Fatboy RX590 yesterday to check my other issues and the Fatboy crashes out in the same manner (during the start of the 3D tests though) unless I dial its clock back to like 1000Mhz from stock in the GPU tuning section.  Seems like these cards had this problem too and ran hot.