[Resolved] Invalid

Developer Tools Nsight Compute
1

I am trying to use the NsightCompute-2019.3 CLI for profiling the matrixMul sample, but the profiler goes to a loop and only stops after ~280 ==PROF== Profiling "MatrixMulCUDA" - 2: 0%....50%....100% - 47 passes messages and then dumps all the output making it hard to use.
If I press Ctrl + C after one run, i can see the results properly. Am I missing something or is this a bug?

Driver 430.26
CUDA 10.2
GPU: GTX 1660 (sm_75)

Here is a sample of what I am trying to do.

/usr/local/cuda/samples/0_Simple/matrixMul$ nv-nsight-cu-cli ./matrixMul
[Matrix Multiply Using CUDA] - Starting...
==PROF== Connected to process 28512
GPU Device 0: "GeForce GTX 1660" with compute capability 7.5

MatrixA(320,320), MatrixB(640,320)
Computing result using CUDA Kernel...
==PROF== Profiling "MatrixMulCUDA" - 1: 0%....50%....100% - 47 passes
done
==PROF== Profiling "MatrixMulCUDA" - 2: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 3: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 4: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 5: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 6: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 7: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 8: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 9: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 10: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 11: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 12: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 13: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 14: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 15: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 16: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 17: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 18: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 19: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 20: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 21: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 22: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 23: 0%....50%....100% - 47 passes
==PROF== Profiling "MatrixMulCUDA" - 24: 0%....50%....100% - 47 pass
<....Only stops after ~280 runs>
2

I should have looked at the code. It is running the matrixmul function 300 times.