Call to nvmlDeviceGetComputeRunningProcesses fails with NVML_ERROR_NOT_SUPPORTED

While calling “nvmlDeviceGetComputeRunningProcesses” it returns with NVML_ERROR_NOT_SUPPORTED.

This occurs on Windows 10 after upgrading my graphic driver from version 441.22 to 461.40. My application is compiled with CUDA 10.2.

Other NVML calls (including NVML initialization) work as expected.

Also, according to the NVML documentation, the function mentioned before does not seem to be deprecated (see here).

Is there a compatibility issue with the new driver?
Is the NVML library binary compatible with new drivers (as mentioned here)?