Lession 1 - The first dog/cat finetune/fit code, how long does it take on the AWS P2.xlarge to train?

I tried to start lesson 1 and setup an AWS P2 to run.
Got this error on In[4]:
ERROR (theano.sandbox.cuda): nvcc compiler not found on $PATH. Check your nvcc installation and try again.
Using Theano backend.

And the training time ETA is at 9267s at the moment.
Is that normal? How should I fix this error?
I am using the standard 128GB image from the instruction.

For other people who may have same problem.
I used this stackoverflow answer:

In current image I cloned from this course, it is cuda-8.0
I added the following to ~/.theanorc:

[cuda]
root = /usr/local/cuda-8.0

Seem the error is fixed and now the training of epoch=1 of total xxx/23000, it is shown ETA at about 700s -800s for the full length.