Failed on demo_server.py

[ronykalfarisi]

Please help, My GPU is RTX 2070 8GB and I'm running using TF 1.12 with CUDA 9.0 and CuDNN 7.5. When I run the demo_server.py, I got the following error

Loading checkpoint: ./tacotron-20180906/model.ckpt
Serving on port 9000
127.0.0.1 - - [23/Aug/2019 10:56:53] "GET / HTTP/1.1" 200 1718
127.0.0.1 - - [23/Aug/2019 10:56:53] "GET /favicon.ico HTTP/1.1" 404 0
2019-08-23 10:57:05.092767: E tensorflow/stream_executor/cuda/cuda_fft.cc:261] failed to make cuFFT batched plan:5
2019-08-23 10:57:05.092788: F tensorflow/stream_executor/cuda/cuda_fft.cc:444] failed to initialize batched cufft plan with customized allocator: Failed to make cuFFT batched plan.
Aborted (core dumped)

[CharlesSL]

I got this problem too First, I'd tried different versions of CUDA, TF...using anaconda, but it doesn't work at all. Then, ocationally I tried training on a Win10 mechine instead of Ubantu, and I didn't change the other things, this problem just simply disappeared. I haven't found the reason but I hope this could help you :)

[ronykalfarisi]

Hi @sl911948113, Thanks for the help. You're right. This problem is caused by the GPU RTX architecture. After browsing the internet for a while, I found that the solution is to use CUDA 10.0/CuDNN 7.5 and TF-1.13.