amdgpu Error...failed, Not enough memory for command submission!

[rnagel]

Anybody else running t740 ThinPro 7.2, SP 10.3 with;

Model:                                 AMD Ryzen™ Embedded V1756B with Radeon™ Vega 8 Graphics

Line:                                      V-Series V1000 with Radeon™ Vega Graphics

and have these errors in dmesg log?
[drm:amdgpu_cs_ioctl [amdgpu]] *ERROR* amdgpu_vm_validate_pt_bos() failed.
[drm:amdgpu_cs_ioctl [amdgpu]] *ERROR* Not enough memory for command submission!

After about two weeks of 24x7 uptime, physical monitors connected to t740 will start experiencing slow movement of windows or have trouble writing over others with background windows losing parts of the window where the foreground window is moving over the top. Eventually, the t740 will stop displaying anything altogether and all you are left with is mouse pointer which can still move around the screens, but the t740 is unresponsive to keyboard key presses and power button no longer shuts down t740 unless you press and hold for 3-5 seconds.  Can still ssh to the t740 and execute hptc-restartX which disconnects ZCentral Remote Boost session from the host and can reconnect with remote host not missing a beat.

I've set some t740s to Force their graphics to use 2048MB in BIOS and some on auto, but neither seem to make a difference, have seen same amdgpu errors on both.

Is there a way to update video drivers separately? And any ideas to monitor GPU memory?

Looking for answer(s) other than "have you updated to latest everything to see if that fixes it?"

Considering environment and rigid configuration management, I cannot just go out and run a fix or update on one or two.

Thanks!