I’m trying to do cross-compile on Nsight EE to Jetson AGX Xavier.
And I tested cuda runtime project sample code to run and debug.
Local implementation has no issue with build, run and debug.
And Remote build and run are OK, but debug has some warning messages.
warning: Cuda API error detected: cudaLaunchKernel returned (0xd1)
Anyway, It reachs to end of the application.
And this is the console log of remote debug.
nsight_SampleCode_remote_debug.txt (4.2 KB)
And I tested one more project with CUDA Sample is MatrixMul.
I create cuda runtime project and copy codes from /usr/local/cuda-10.2/samples/0_Simple/matrixMul/matrixMul.cu
It works on local run and debug without any issue.
But there are problems on remote implementations.
This message prints repeatedly when it run on remote.
Error! Matrix[00005]=0.00000000, ref=3.19999981 error term is > 1.000000E-06
This message prints repeatedly when it debug on remote.
warning: Cuda API error detected: cudaLaunchKernel returned (0xd1)
Those are logs.
nsight_MatrixMul_remote_run.txt (1.5 KB)
nsight_MatrixMul_remote_debug.txt (5.1 KB)
Please, Check what is the problem?