can nsight run directly on tx1 to do native development

for some reasons ,i can’t use a host PC do remote development on tx1.

so ,I wanna to ask ,

can Nsignt installed directly on TX1,and so I can develop Cuda project directly on TX1?

or does there have any method to do native cuda development on TX1 with out using a host PC?

I read All the document and forum but there is not any usefull information.

thanks a lot .

By default, CUDA Toolkit 8.0 is installed to the Jetson by JetPack-L4T. This provides the toolchain for native CUDA development on Jetson (ARM aarch64), amongst others including:

  • compiler (nvcc)
  • linker (nvlink)
  • debugger (cuda-gdb)
  • profiler (nvprof)

These are located under /usr/local/cuda in addition to headers and libs for application development.

At this time, Nsight GUI is not provided for ARM to natively run on Jetson. However, as mentioned above, all the CUDA tools that Nsight uses underneath are provided. If you are looking for a way to manage your project, I recommend cmake.