I ran into issues while trying to integrate CUDA code into an existing codebase, which uses STLport (on Windows).
Basically, just by adding STLport headers on the INCLUDE path, nvcc compilation breaks, with errors from cudafe such as:
…\STLport-5.0.2\cstdlib(140): error : function “abs(long long)” has already been defined
c:\program files\nvidia gpu computing toolkit\cuda\v5.0\include\math_functions.h(7907): error : function “hypot(float, float)” has already been defined
…STLport-5.0.2\stl/_string.c(655): error : member “stlp_std::basic_string::npos” has already been defined
I used CUDA 5.0.35, VS2010 and tried STLport 5.0.2 and 5.2.1. All fail the same way.
Is there anyone having the same issues or anyone using STLport with CUDA and NOT having those issues?