The bug for tensorRT 1.0

I use the tensorRT 1.0 on tegra x1. Because the tensorRT not support same layre,I must use caffe after get output from tensorRT. But I find use booth tensorrt and caffe,the tensorRT will get error. When I use the protobuf form apt-get,the programe get the error as follow:

./…/…/externals/protobuf/aarch64/10.0/include/google/protobuf/generated_message_util.h:84:
const string& google::protobuf::internal::GetEmptyStringAlreadyInited():
Assertion `empty_string_ != __null’ failed.

I solved the problem throth using protobuf v3.0.0. But I get a Segmentation fault when I init the tensrtRT interface. How can I solve this problem.

Hi,

This bug is a known issue. The reason is conflict protobuf.

We already fixed this issue in TensorRT2.1 and packed it into JetPack3.1.
Please wait for our announcement for the next JetPack release. It should be soon.

Thanks.

Hi,

TensorRT2.1 is available now!
Please check JetPack SDK | NVIDIA Developer