Cuda debugging (Parallel Nsight) Debugging Cuda Kernel

Hi All,

After installing Parallel Nsight 2.1 (for Cuda 4.1 RC2 version), I get Connected and Disconnected pop up, immediately. I have done the general settings for running NSight on my Visual studio 2010. I wish to debug on localhost itself.

I have a few questions:

  1. I have read through sections of the Parallel NSight user manual, It didn’t help too much for this issue.
  2. Can I start local debugging directly with NSight-> Start Cuda Debugging for starting to debug my project OR should I click on this option only (& everytime) I step into the kernel ?
  3. The other way I tried is, I started with normal CPU debugging, just when I hit the kernel, selected step 2) and got the following (screenshot): (This time it said ‘Connected to Cuda Debugger’):


And as soon I pressed (F10, default) for debugging it said “Debugger cannot run Cuda kernel”.


Has anyone else faced this problem ?

Could you please let me know the procedure for cuda kernel invocation. I just want to debug the kernel. Where am I going wrong ?

Best Regards,
[/b]Shyam B

I had one problem regarding parallel nsight debugging. I used opencv 2.2 and cuda toolkit 4.1 to make my code. My code works fine when I give normal debugging, but when I used cuda debugging I got opencv error. I also attached the error msg. Please any one help me.