The tensorflow cudnn version is not compatible with tensorrt cudnn version

Autonomous Machines Jetson & Embedded Systems Jetson TX2
#1

When I run the example tf_to_trt.py,the error is as below:

2017-10-19 10:02:31.683051: E tensorflow/stream_executor/cuda/cuda_dnn.cc:352] Loaded runtime CuDNN library: 7001 (compatibility version 7000) but source was compiled with 5110 (compatibility version 5100). If using a binary install, upgrade your CuDNN library to match. If building from sources, make sure the library loaded at runtime matches a compatible version specified during compile configuration.
2017-10-19 10:02:31.683772: F tensorflow/core/kernels/conv_ops.cc:671] Check failed: stream->parent()->GetConvolveAlgorithms(&algorithms)
Aborted (core dumped)

the cudnn version of tensorrt is cudnn7.0, but tensorflow does not support cudnn7.0 now. How to solve it?

#2

Hi,

You can solve this by installing a CPU-only TensorFlow package.
For importing a TF model, a CPU-based module should be enough.

Thanks.

#3

you will also see an error similar to “can’t find cuda-8.0” when you “pip install tensorflow-gpu” (gpu version) vs. “pip install tensorflow” (cpu version).

there’s not a lot on these issues, so i wanted to document them here.

#4