Sorry, this post has been removed.
the solution i found was that to be sure that all values used in the calculations were doubles and NO integers * doubles
Thanks in advance.
Sorry, this post has been removed.
the solution i found was that to be sure that all values used in the calculations were doubles and NO integers * doubles
Thanks in advance.
Hi,
I’m not sure if it is the actual reason of the error you encounter, but there is at least a problem in your code at the following level:
double SrReal[512], SrImag[512];
double ft[512*512];
Those two lines define some per-thread automatic variables in your kernel, that are expected to be allocated in the thread’s local memory. Those 3 tables only are accountable for a bit more than 2MB of memory. And according to the CUDA_C_Programming_Guide, Appendix F page 137 (149/171), local memory is limited to 16KB for compute capability 1.x and 512KB for compute capability 2.x… Here, you are requesting far more than those limits.
Just a guess…