1

I have a server which hosts three different GPU platforms: Onboard GPU, Nvidia and AMD GPUs. I have not installed X server, as I do not intend to bring the desktop up. I always use ssh and use the GPUs for computations only. Now, I have been able to install the Nvidia driver and run both OpenCL and Cuda applications on my Nvidia card. But, the AMD driver installation apparently failed. I followed the instructions on installing the latest AMD driver 18.30 but I got this warning message:

WARNING: amdgpu dkms failed for running kernel

I also noticed that the no driver is in use for my AMD card:

$ lspci -nnk | grep -i vga -A3
04:00.0 VGA compatible controller [0300]: ASPEED Technology, Inc. ASPEED Graphics Family [1a03:2000] (rev 41)
    DeviceName: ASPEED Video AST2500
    Subsystem: Super Micro Computer Inc ASPEED Graphics Family [15d9:096d]
    Kernel driver in use: ast
--
3b:00.0 VGA compatible controller [0300]: Advanced Micro Devices, Inc. [AMD/ATI] Device [1002:67df] (rev e7)
    Subsystem: XFX Pine Group Inc. Device [1682:c580]
3b:00.1 Audio device [0403]: Advanced Micro Devices, Inc. [AMD/ATI] Device [1002:aaf0]
    Subsystem: XFX Pine Group Inc. Device [1682:aaf0]
--
5e:00.0 VGA compatible controller [0300]: NVIDIA Corporation Device [10de:1b06] (rev a1)
    Subsystem: Micro-Star International Co., Ltd. [MSI] Device [1462:3609]
    Kernel driver in use: nvidia
    Kernel modules: nvidiafb, nouveau, nvidia_410_drm, nvidia_410
--
86:00.0 VGA compatible controller [0300]: NVIDIA Corporation Device [10de:1b06] (rev a1)
    Subsystem: Micro-Star International Co., Ltd. [MSI] Device [1462:3609]
    Kernel driver in use: nvidia
    Kernel modules: nvidiafb, nouveau, nvidia_410_drm, nvidia_410

I also searched over this issue and it seems that it is not easy to install both AMD and Nvidia cards together on the same machine. Is there anybody here who have done that and can share the experience?

Arya Mz
  • 111
  • 1
  • 1
    The last time I tried, the AMD driver required that OpenCL code be running under an X server running on the same AMD card. Unlike CUDA it was not possible to run the OpenCL code without an X server. Perhaps it is different now, but I have not heard of any changes. – Michael Hampton Oct 19 '18 at 16:12

0 Answers0