I hope the following text will not sound silly or to stupid for all of you who have much more experience than I in programming. I am new to CUDA and have written some code which, now successfully compiles, but I have some issues with debugging. I use visual studio 2005 and i started my project by using the template provided by nVidia. Well, now when I set the brake points right before the kernel function call and press step into when the executions reaches my break point, I get a message saying that “there is no source code available for the current location” and I can chose between pressing “ok” or “show disassembly”. If I just press ok and want to continue debugging the same message appears again. It’s the same both in EmuDebug and Debug.
I thought it only because I have written a bad code, and I tried to debug some of the examples provided by nVidia, but the same thing happens.
I would appreciate any kind of help.
I did that and the problem persists. I can’t see the results of simple operations like a=5. My deadline is approaching soon and I can’t see what’s going on in my project and why it’s not doing what it’s supposed to.
When I call the kernel with a pointer to a matrix allocated by cudaMalloc and initialize the matrix with 0’s. Everything is ok, but when I calculate some int value and say that matrix[i]=intValue; and after the for loop finishs I check if matrix element i got the desired intValue I jsut get a 0. It seems that the matrix just stays like I initializes it. Can anybody tell me why?