I'm new for GPU related model training. I have Tesla C2075 with 6GB GPU and using keras CuDNNLSTM for faster training. I have installed cuda-9 with cudnn=7.0.5, tensorflow-gpu==1.12.0 and using ubuntu 16.04. For Tesla C2075 GPU model is compatible with cuda-9? I have checked https://developer.nvidia.com/cuda-gpus link in this they have mentioned tesla C2075 is compute compatible to 2.0. what is compute compatible?
And while running my model tensorflow log,
tensorflow/core/common_runtime/gpu/gpu_device.cc:1482] Ignoring visible gpu device (device: 0, name: Tesla C2075, pci bus id: 0000:03:00.0, compute capability: 2.0) with Cuda compute capability 2.0. The minimum required Cuda capability is 3.5.
And I'm also getting error while model.fit(...),
InvalidArgumentError (see above for traceback): No OpKernel was registered to support Op 'CudnnRNN' with these attrs. Registered devices: [CPU,XLA_CPU,XLA_GPU], Registered kernels:
device='GPU'; T in [DT_DOUBLE]
device='GPU'; T in [DT_FLOAT]
device='GPU'; T in [DT_HALF]
[[node bidirectional_1/CudnnRNN (defined at /usr/local/lib/python3.5/dist-packages/tensorflow/contrib/cudnn_rnn/python/ops/cudnn_rnn_ops.py:922) = CudnnRNN[T=DT_FLOAT, direction="unidirectional", dropout=0, input_mode="linear_input", is_training=true, rnn_mode="lstm", seed=87654321, seed2=0](bidirectional_1/transpose, bidirectional_1/ExpandDims_1, bidirectional_1/ExpandDims_2, bidirectional_1/concat)]]
Thanks
The CUDA compute capability relates somehow to the architecture and hardware capabilities of the GPU, there is quite an extensive list in wikipedia.
The tensoflow webpage suggests that you need a GPU with CC bigger than 3.5 (older versions seemed to accept 3.0, but never lower).
Unfortunately this is a hardware limitation, the only way of changing your compute capability is using a different GPU. Simply said: you can not run Tensorflow in that GPU.