search

ry / tensorflow-vgg16

conversation of caffe vgg16 model to tensorflow
672 stars 285 forks source link

aborted (core dumped) #6

Open daxiongshu opened 8 years ago

daxiongshu commented 8 years ago

Thank you very much for the code. I got the following error message. Any idea?

Thank you!

caffe session prob shape (1000,) Top1: n02123045 tabby, tabby cat Top5: ['n02123045 tabby, tabby cat', 'n02124075 Egyptian cat', 'n02123159 tiger cat', 'n02119789 kit fox, Vulpes macrotis', 'n02119022 red fox, Vulpes vulpes'] tensorflow session F tensorflow/stream_executor/cuda/cuda_driver.cc:302] current context was not created by the StreamExecutor cuda_driver API: 0x2c3a910; a CUDA runtime call was likely performed without using a StreamExecutor context Aborted (core dumped)

oliviaguest commented 8 years ago

Me too - although I just get:

python vgg16.py 
Segmentation fault (core dumped)
loganwilliams commented 7 years ago

I was able to convert the model successfully by temporarily disabling GPU usage in TensorFlow.

olivernina commented 7 years ago

I got it working by compiling caffe on cpu mode. Good luck

stgzr commented 7 years ago

Same issue,

F tensorflow/stream_executor/cuda/cuda_driver.cc:316] current context was not created by the StreamExecutor cuda_driver API: 0x42aa310; a CUDA runtime call was likely performed without using a StreamExecutor context Aborted (core dumped)

caffe.__version__ = 1.0.0-rc3 tensorflow.__version__ = 0.9.0 GPU: Tesla K40c GPU Driver version: 352.39 CUDA 7.5

SudhakarReddyPeddinti commented 7 years ago

I faced the same issue. Apparently the issue arises because of the GPU sharing between caffe and Tensorflow. Caffe uses the GPU first to inference the test cat.JPG and does not allow the Tensorflow to run. I have solved this issue by making the Caffe to run only on CPU while compiling.

Summary: update Makefile.config with CPU_ONLY := 1 and run make to run Caffe with CPU only