failed to create kernel for CNv8 about teamredminer

[todxx]

Hi PNixx,

Strangely enough, it seems that your vega is running with ROCm drivers. The driver version 2679.0 (HSA1.1,LC) is what ROCm typically reports. The amdgpu-pro 18.30/40 drivers will report a driver version of 2671.3 (PAL,HSAIL) or similar.

Can you run apt list *rocm* and post the output?

[PNixx]

libfcgi-procmanager-maxrequests-perl/bionic,bionic 0.2-1 all
libfcgi-procmanager-perl/bionic,bionic 0.28-1 all
procmail/bionic 3.22-26 amd64
procmail-lib/bionic,bionic 1:2009.1202-4 all
procmeter3/bionic 3.6-1 amd64
rocm-clang-ocl/Ubuntu 16.04 0.3.0-7997136 amd64
rocm-cmake/Ubuntu 16.04 0.2.0-5e74c90 amd64
rocm-dev/Ubuntu 16.04 1.9.211 amd64
rocm-device-libs/Ubuntu 16.04 0.0.1 amd64
rocm-dkms/Ubuntu 16.04 1.9.211 amd64
rocm-libs/Ubuntu 16.04 1.9.211 amd64
rocm-opencl/Ubuntu 16.04,now 1.2.0-2018090737 amd64 [установлен]
rocm-opencl-dev/Ubuntu 16.04,now 1.2.0-2018090737 amd64 [установлен]
rocm-profiler/Ubuntu 16.04 5.4.6878 amd64
rocm-smi/Ubuntu 16.04 1.0.0-72-gec1da05 amd64
rocm-utils/Ubuntu 16.04 1.9.211 amd64
rocm_bandwidth_test/Ubuntu 16.04 1.0.0 amd64
rocm_smi64/Ubuntu 16.04 0.8.0 amd64
rocminfo/Ubuntu 16.04 1.0.0 amd64

[todxx]

Yep, you have rocm installed. What I would suggest doing is first uninstalling the amdgpu-pro drivers you installed, then uninstalling rocm, after that try installing the amdgpu-pro drivers again.

Though be warned that the last time I accidentally installed both rocm and amdgpu-pro I ended up with a really screwed up machine that I gave up on and just did a full reinstall on. Let me know if you need help with uninstalling rocm.

[PNixx]

Works! Thanks.
I run:

amdgpu-pro-uninstall && apt autoremove rocm-opencl rocm-opencl-dev && ./amdgpu-pro-install -y --opencl=pal && reboot

[todxx]

Awesome! Glad I could help :)