Looking for a software able to verify the GPU (TESLA P100) is properly installed…
Using Infoworks ICM (Innovyze), CUDA 8.0, 2 GPU TESLA installed (recognized by nvsmi).
GPU worked properly until 21/08 (accelerating 20x or 30x calculations).
Since 29/08 run of models can’t use GPU anymore.
No change on ICM ; drivers have verified and reinstalled.
Would like to run a “reference” test to verify GPU is proprerly installed and can work on NVIDIA standard executable …
Thanks in advance.
It is not clear from the description whether you have CUDA installed on your machine or just the driver package. If you have CUDA installed, start with the deviceQuery and bandwidthTest sample apps that ship with CUDA. You may want to post the output here, along with complete output from
nvidia-smi -q. We may or may not be able to discern something from this.
I am not familiar with Infoworks ICM (Innovyze). Maybe this application requires an add-on license for GPU acceleration which expired? Did you acquire the machine with the Tesla GPUs already installed from an integrator ([url]https://www.nvidia.com/en-us/data-center/where-to-buy-tesla/[/url])? You may want to check with their technical support. Have you inquired with the software vendor about possible error scenarios?
What happened between August 21 and August 29? An update to the operating system perhaps? What is the OS being used?