Hello,
I have been trying to compile binary for Jetson Nano (4GB) on my host machine which is an Ubuntu 20.04 distribution. I have tried my hands on over vectorAdd example and following are the steps which I followed to create two binaries:
-
Binary 1 - To run on host machine i.e. Ubuntu 20.04
Inside the vectorAdd sample directory on host:
Command - make
Result - Binary executed correctly, result was pass. -
Binary 2 - To run on Jetson Nano
Inside the vectorAdd sample directory on host:
Command - make TARGET_ARCH=aarch64 TARGET_OS=linux SMS=“52 53” HOST_COMPILER=/usr/bin/aarch64-linux-gnu-g++
Result -
[Vector addition of 50000 elements]
Failed to allocate device vector A (error code initialization error)!
I have tried many permutation and combination of different parameters but still unable to run binary on Jetson Nano.
Tried NVCC command as well:
nvcc -m64 -arch=compute_53 -code=sm_53 -target-dir=aarch64-linux --compiler-bindir=/usr/bin/aarch64-linux-gnu-g++ vectorAdd.cu -o vectorAdd
But still the result remains same. Binary is not able to invoke the GPU.
Can anyone please state me the problem so that I can look at that specific area and sort this out.
Regards,
Alark Mishra