cuStreamWaitValue32 returns CUDA_ERROR_NOT_SUPPORTED

Hi,

We are trying to use cuStreamWaitValue32 function. The return value we get when calling the function is CUDA_ERROR_NOT_SUPPORTED. We are using Quadro M6000 GPU with driver version 430.40 and cuda toolkit 10.1.168.

Please advice,
thank you.

Support is disabled by default.

You need to enable the feature when the GPU driver loads:

https://docs.nvidia.com/cuda/cuda-driver-api/group__CUDA__MEMOP.html#group__CUDA__MEMOP

Thank you for your answer.

According to the documentation the driver needs to be loaded with the following command:
modprobe nvidia NVreg_EnableStreamMemOPs=1

In our system the driver loads at boot time, can you please tell us if the modprobe command is executed from a script that runs n system load? If the answer is yes, than what is the name and the path to that script? We are working with ubuntu 18.04.

Thanks.

The driver may be loaded in a variety of ways “at boot time”. I won’t be able to tell you precisely what is happening on your system.

This article discusses creating a init script to take charge of the loading process:

https://www.resultsovercoffee.com/2011/01/cuda-in-runlevel-3.html