compiler ignore kernel function

Accelerated Computing CUDA CUDA Programming and Performance
1

does anybody has meet this problem like me.
*.cpp compile is OK
*.cu compile is OK
linking… OK
it is mean that build solution is OK
no error no warning :)

but
compiler did not jump to kernel function (ignore this function) <img src=‘http://hqnveipbwb20/public/style_emoticons/<#EMO_DIR#>/crying.gif’ class=‘bbc_emoticon’ alt=‘:’(’ />

  • .cu file
    after setup alocate and copy memory
    //call kernel function
    funtion<<dimGrid, dimBlock>>(parameter);
    return data

i have made some cuda program in visual c++ 2005 and it work well :)
but this is the first time i meet this problem. <img src=‘http://hqnveipbwb20/public/style_emoticons/<#EMO_DIR#>/crying.gif’ class=‘bbc_emoticon’ alt=‘:’(’ />

2

If the compiler detects that the method doesn’t return anything (no global-memory field is modified, typically), it will just ignore the call. Try to check the .cubin file and/or the ptx, to see if it is generated.

3

thank you tim-timmy. :D

I had solved this problem. :D

I forget that i use larger threads than i have External Media

geforce 8800GT limited threads on a block is 512. :)

but i use too much 546. >.<