Created by: jedibobo

报错： Scanning dependencies of target dequantize_op [ 45%] Building CXX object paddle/fluid/operators/CMakeFiles/eye_op.dir/eye_op.cc.o [ 45%] Building CXX object paddle/fluid/operators/CMakeFiles/dequantize_op.dir/dequantize_op.cc.o [ 45%] Linking CXX static library libeye_op.a [ 45%] Linking CXX static library libdequantize_op.a [ 45%] Built target dequantize_op [ 45%] Built target eye_op [ 46%] Building NVCC (Device) object paddle/fluid/operators/CMakeFiles/conv_transpose_op.dir/conv_transpose_op_generated_conv_transpose_op.cu.o [ 46%] Building NVCC (Device) object paddle/fluid/operators/CMakeFiles/conv_transpose_op.dir/conv_transpose_op_generated_conv_transpose_cudnn_op.cu.o nvcc error : 'cicc' died due to signal 9 (Kill signal) CMake Error at activation_op_generated_activation_op.cu.o.Release.cmake:275 (message): Error generating file /home/dlinano/Paddle/build/paddle/fluid/operators/CMakeFiles/activation_op.dir//./activation_op_generated_activation_op.cu.o

paddle/fluid/operators/CMakeFiles/activation_op.dir/build.make:63: recipe for target 'paddle/fluid/operators/CMakeFiles/activation_op.dir/activation_op_generated_activation_op.cu.o' failed make[2]: *** [paddle/fluid/operators/CMakeFiles/activation_op.dir/activation_op_generated_activation_op.cu.o] Error 1 CMakeFiles/Makefile2:46195: recipe for target 'paddle/fluid/operators/CMakeFiles/activation_op.dir/all' failed make[1]: *** [paddle/fluid/operators/CMakeFiles/activation_op.dir/all] Error 2 make[1]: *** 正在等待未完成的任务.... Scanning dependencies of target conv_transpose_op [ 46%] Building CXX object paddle/fluid/operators/CMakeFiles/conv_transpose_op.dir/conv_transpose_op.cc.o [ 46%] Linking CXX static library libconv_transpose_op.a [ 46%] Built target conv_transpose_op Scanning dependencies of target arg_min_op [ 46%] Building CXX object paddle/fluid/operators/CMakeFiles/arg_min_op.dir/arg_min_op.cc.o [ 46%] Linking CXX static library libarg_min_op.a [ 46%] Built target arg_min_op

环境：JetPack4.3 Found Paddle host system: ubuntu, version: 18.04.3 -- Found Paddle host system's CPU: 4 cores -- CXX compiler: /usr/bin/c++, version: GNU 7.5.0 -- C compiler: /usr/bin/cc, version: GNU 7.5.0

Cmake命令： cmake ..
-DWITH_CONTRIB=OFF
-DWITH_MKL=OFF
-DWITH_MKLDNN=OFF
-DWITH_TESTING=OFF
-DCMAKE_BUILD_TYPE=Release
-DON_INFER=ON
-DWITH_PYTHON=ON
-DWITH_XBYAK=OFF
-DWITH_NV_JETSON=ON
-DPY_VERSION=3.6
-DTENSORRT_ROOT=/usr/lib/aarch64-linux-gnu/
-DCUDA_ARCH_NAME=All

为什么不使用低版本： 1.6.3试过，运行Paddle Inference给的demo上结果错误，得到的回复是model和lib版本要一致。

官方文档的0.0.0的也试过：https://paddle-inference-dist.cdn.bcebos.com/temp_data/paddlepaddle_gpu-0.0.0-cp36-cp36m-linux_aarch64.whl 这个链接给出的版本不支持Paddle-TensorRT https://github.com/PaddlePaddle/Paddle/issues/20786相同报错。

1.8.x版本会报cpuid.h的错误，在issue里面找到的结果是2.0.0-alpha0已经修复

最后恳请大佬给我Jetson nano上部署paddle模型指条路 谢谢！