[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [HTCondor-users] GPUs not detected in 9.0.6 version



On Thu, 2021-09-30 at 12:39:41 +0200, Carles Acosta wrote:
> Hello all,
> 
> On one side, Stuart and Steffen's tests do not reproduce my issue. On the
> other side, there is no segfault for our machines using cuda 11 and GeForce
> RTX 2080 Ti and Testla V100. So, I've updated Cuda from 10.1 to 11 and,
> voilà, there is no segfault anymore using the condor_gpu_discovery. In
> conclusion, the condor_gpu_discovery segfault for version 9.0.6 seems to be
> related to the CUDA version < 11.

Hi Carles,

indeed:
| NVIDIA-SMI 460.32.03    Driver Version: 460.32.03    CUDA Version: 11.2     |

As it turns out, none of my GPU machines are still at CUDA 10.x...


- S