Search Unity

  1. Megacity Metro Demo now available. Download now.
    Dismiss Notice
  2. Unity support for visionOS is now available. Learn more in our blog post.
    Dismiss Notice

Question [Not Solved 13/06]Using GPU not working

Discussion in 'ML-Agents' started by mateolopezareal, Jun 13, 2020.

  1. mateolopezareal

    mateolopezareal

    Joined:
    Jun 4, 2020
    Posts:
    54
    I am using an RTX 2070, and after installing cuda and cudnn I can use my GPU to train my agents. Here you can see how its all well connected
    Code (CSharp):
    1. 2020-06-13 12:32:52.875673: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties:
    2. pciBusID: 0000:01:00.0 name: GeForce RTX 2070 computeCapability: 7.5
    3. coreClock: 1.455GHz coreCount: 36 deviceMemorySize: 8.00GiB deviceMemoryBandwidth: 327.88GiB/s
    4. 2020-06-13 12:32:52.875747: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cudart64_101.dll
    5. 2020-06-13 12:32:52.876022: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cublas64_10.dll
    6. 2020-06-13 12:32:52.876242: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cufft64_10.dll
    7. 2020-06-13 12:32:52.876522: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library curand64_10.dll
    8. 2020-06-13 12:32:52.876760: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cusolver64_10.dll
    9. 2020-06-13 12:32:52.877054: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cusparse64_10.dll
    10. 2020-06-13 12:32:52.877323: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cudnn64_7.dll
    11. 2020-06-13 12:32:52.877827: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0
    12. 2020-06-13 12:32:52.877896: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102] Device interconnect StreamExecutor with strength 1 edge matrix:
    13. 2020-06-13 12:32:52.878090: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1108]      0
    14. 2020-06-13 12:32:52.878330: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1121] 0:   N
    15. 2020-06-13 12:32:52.878852: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1247] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 6646 MB memory) -> physical GPU (device: 0, name: GeForce RTX 2070, pci bus id: 0000:01:00.0, compute capability: 7.5)
    16. 2020-06-13 12:32:54.370941: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cublas64_10.dll
    However, as yiu can see in these images
    Captura.PNG Captura2.PNG
    , the GPU is not been used. Why is this? Is it better to train with CPU (i have a i7-10gen)?
     
  2. MrWetsnow

    MrWetsnow

    Joined:
    Jan 5, 2020
    Posts:
    60
    For training you are better off using the CPU here, it's faster.
     
    vincentgao88 likes this.