Anaconda showing this error , can't train model properly

Question

2021-02-03 19:46:53.571084: W tensorflow/stream_executor/gpu/redzone_allocator.cc:314] Internal: Invoking GPU asm compilation is supported on Cuda non-Windows platforms only
Relying on driver to perform ptx compilation.
Modify $PATH to customize ptxas location.
This message will be only logged once.

I am getting the above error when I am trying to train my deep nueral network. It doestn't stop training but the first iteration is not working fine. See below:

Epoch 1/10
   1/1875 [..............................] - ETA: 0s - loss: 2.3001 - accuracy: 0.1250
WARNING:tensorflow:Callbacks method `on_train_batch_end` is slow compared to the batch time (batch time: 0.0000s vs `on_train_batch_end` time: 0.0116s). Check your callbacks.

Help me to fix this error.

I am using tensorflow verson 2.3.0, cuda tool kit version 10.1.243, cudnn version 7.6.5 and my pc has NVDIA GEFORCE RTX 2070 with Super Max-Q design.

Some solutions say to downgrade tensorflow version, but for some projects I need 2.3.0, so please any one solve this error.

Answer 1

I met the same problem, my TensorFlow version is 2.4.1. with Python 3.9, and cuda 10.1. Luckily, I found a solution by installing the following package,

conda install -c conda-forge cudatoolkit-dev

The original link is here.