I am currently experiencing issues with the OpenCL stack on my NVIDIA Quadro RTX 4000 GPU under Linux RHEL9. The issue is specifically related to system sleep/wake cycles.
System Details:
OS: RHEL 9
Kernel Version: 5.14.0-284.18.1.el9_2.x86_64
Architecture: x86_64
GPU: NVIDIA Quadro RTX 4000
NVIDIA Driver Version: 535.54.03
OpenCL Platform Version: OpenCL 3.0 CUDA 12.2.79
Issue:
When the system is first booted, the OpenCL stack works as expected and the clinfo command returns correct results. However, after a system sleep/wake cycle, the OpenCL stack appears to be non-functional with clinfo returning “Number of platforms 0”.
The issue is temporarily resolved by a full system reboot, but recurs after every sleep/wake cycle. This issue persists despite the reinstallation of NVIDIA drivers and attempts to refresh the OpenCL stack on wake.
Additional Observations:
Other software, such as Discord, also seems to crash after a system wake.
The GPU appears to function normally for non-OpenCL tasks post-wake.
Any assistance or insights you can provide into this issue would be greatly appreciated.
Yum Package install history Nvidia
Begin time : Wed 14 Jun 2023 04:17:27 PM
Begin rpmdb : 50e18f50b24b5cdb4f3924e1080d75c658928c0fd3b6df53ffcedd3b63b3310a
End time : Wed 14 Jun 2023 04:17:39 PM (12 seconds)
End rpmdb : e6d190e17dacdc174abaf542f5a342cc6e5a14d1516f15c17a567c3517e2c2ff
User : edna
Return-Code : Success
Releasever : 9
Command Line : install nvidia-driver nvidia-settings
Comment :
Packages Altered:
Install dnf-plugin-nvidia-2.0-1.el9.noarch @cuda-rhel9-x86_64
Install kmod-nvidia-530.30.02-5.14.0-284.11.1-3:530.30.02-3.el9_2.x86_64 @cuda-rhel9-x86_64
Install nvidia-driver-3:530.30.02-1.el9.x86_64 @cuda-rhel9-x86_64
Install nvidia-driver-libs-3:530.30.02-1.el9.x86_64 @cuda-rhel9-x86_64
Install nvidia-kmod-common-3:530.30.02-1.el9.noarch @cuda-rhel9-x86_64
Install nvidia-libXNVCtrl-3:530.30.02-1.el9.x86_64 @cuda-rhel9-x86_64
Install nvidia-settings-3:530.30.02-1.el9.x86_64 @cuda-rhel9-x86_64
Install egl-wayland-1.1.9-2.el9.x86_64 @appstream
Install libvdpau-1.5-1.el9.x86_64 @appstream
and cuda
Begin time : Wed 14 Jun 2023 04:17:54 PM
Begin rpmdb : e6d190e17dacdc174abaf542f5a342cc6e5a14d1516f15c17a567c3517e2c2ff
End time : Wed 14 Jun 2023 04:17:58 PM (4 seconds)
End rpmdb : 4401fc932d692379928259ef6cb1446ef110b3541a7cb53cef1a8e3f967ff8a6
User : edna
Return-Code : Success
Releasever : 9
Command Line : install cuda-driver
Comment :
Packages Altered:
Install cuda-drivers-530.30.02-1.x86_64 @cuda-rhel9-x86_64
Install nvidia-driver-NVML-3:530.30.02-1.el9.x86_64 @cuda-rhel9-x86_64
Install nvidia-driver-NvFBCOpenGL-3:530.30.02-1.el9.x86_64 @cuda-rhel9-x86_64
Install nvidia-driver-cuda-3:530.30.02-1.el9.x86_64 @cuda-rhel9-x86_64
Install nvidia-driver-cuda-libs-3:530.30.02-1.el9.x86_64 @cuda-rhel9-x86_64
Install nvidia-driver-devel-3:530.30.02-1.el9.x86_64 @cuda-rhel9-x86_64
Install nvidia-libXNVCtrl-devel-3:530.30.02-1.el9.x86_64 @cuda-rhel9-x86_64
Install nvidia-modprobe-3:530.30.02-1.el9.x86_64 @cuda-rhel9-x86_64
Install nvidia-persistenced-3:530.30.02-1.el9.x86_64 @cuda-rhel9-x86_64
Install nvidia-xconfig-3:530.30.02-1.el9.x86_64 @cuda-rhel9-x86_64
Install xorg-x11-proto-devel-2021.4-2.el9.noarch @appstream
Install opencl-filesystem-1.0-15.el9.noarch @appstream
Install libXau-devel-1.0.9-8.el9.x86_64 @appstream
Install libxcb-devel-1.13.1-9.el9.x86_64 @appstream
Install ocl-icd-2.2.13-4.el9.x86_64 @appstream
Install libX11-devel-1.7.0-7.el9.x86_64 @appstream