0

My display link dock/ multi monitor setup works poorly with the Nvidia drivers, and I'd like to maximize the resources available on my GPU while running so I want to run CUDA programs while letting the Intel chips handle graphics rendering. I've had luck doing this sort of thing in Windows, but it seems that the nVidia driver is "not running" when Intel is the active gpu. When I change back to Nvidia I get no errors.

I tensorflow/stream_executor/dso_loader.cc:128] successfully opened CUDA library libcublas.so locally
I tensorflow/stream_executor/dso_loader.cc:119] Couldn't open CUDA library libcudnn.so. LD_LIBRARY_PATH: :/usr/local/cuda/lib64:/usr/local/cuda/extras/CUPTI/lib64
I tensorflow/stream_executor/cuda/cuda_dnn.cc:3459] Unable to load cuDNN DSO
I tensorflow/stream_executor/dso_loader.cc:128] successfully opened CUDA library libcufft.so locally
I tensorflow/stream_executor/dso_loader.cc:119] Couldn't open CUDA library libcuda.so.1. LD_LIBRARY_PATH: :/usr/local/cuda/lib64:/usr/local/cuda/extras/CUPTI/lib64
I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:165] hostname: IT-42380-BIADM
I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:189] libcuda reported version is: Not found: was unable to find libcuda.so DSO loaded into this program
I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:193] kernel reported version is: Permission denied: could not open driver version path for reading: /proc/driver/nvidia/version
I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1092] LD_LIBRARY_PATH: :/usr/local/cuda/lib64:/usr/local/cuda/extras/CUPTI/lib64
I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1093] failed to find libcuda.so on this system: Failed precondition: could not dlopen DSO: libcuda.so.1; dlerror: libnvidia-fatbinaryloader.so.367.57: cannot open shared object file: No such file or directory
I tensorflow/stream_executor/dso_loader.cc:128] successfully opened CUDA library libcurand.so locally
E tensorflow/stream_executor/cuda/cuda_driver.cc:509] failed call to cuInit: CUDA_ERROR_NO_DEVICE
I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:145] kernel driver does not appear to be running on this host (IT-42380-BIADM): /proc/driver/nvidia/version does not exist
3
  • 1
    "Power saving mode" = Intel. Nvidia isn't being used and CUDA depends on Nvidia. CUDA doesn't run on Intel.
    – user589808
    Feb 3, 2017 at 20:53
  • 1
    Sensible assessment. But I'm not looking to render graphics with NVidia, just CUDA. I've run CUDA programs (Theano) like this in Windows before. It may not be possible with Linux drivers though. :( It's almost like the nvidia chip isn't mounted as a device when the intel chip is active. Feb 23, 2017 at 15:55
  • I have similar aspirations on my laptop.. seems there's no answer yet askubuntu.com/questions/952426/…
    – Erik
    Sep 4, 2017 at 13:08

0

You must log in to answer this question.

Browse other questions tagged .