Cuda streams do not run concurrently when using the convolve function

[echoGee]

System information (version)

• OpenCV => 3.4.0
• Operating System / Platform => Windows 64 Bit
• Compiler => Visual Studio 2015

Detailed description

Using streams to run multiple different convolutions do not run concurrently . I have not used any compiler flags such as –default-stream per-thread

Steps to reproduce

The code is in a .cpp file.

Ptr<cuda::Convolution> convolver;
cuda::Stream s1, s2, s3, s4, s5, s6, s7, s8, s9, s10;
convolver->convolve(im_32fc1_d, x, a_d, false, s1);

convolver->convolve(im_32fc1_d, xy, b_d, false, s2);

convolver->convolve(im_32fc1_d, xx, c_d, false, s3);

convolver->convolve(im_32fc1_d, yy, d_d, false, s4);

convolver->convolve(im_32fc1_d, xy, e_d, false, s5);

[tdakhran]

@echoGee did you check "Enable concurrent kernel profiling" checkbox in Settings tab in Nvidia Visual Profiler?

[echoGee]

@tarkook : I see its reference in http://docs.nvidia.com/cuda/profiler-users-guide/index.html. I could not find that option in the Visual profiler. Any pointers?