I recently updated my Cuda SDK from 5.0 to 5.5.
Now when compiling, I get many many errors like:
device_functions.h(2442): error: function “__syncthreads_or” has already been defined
Seems like all functions in this file are recognised as already defined.
I am working under Debian Wheezy, my GPU is GTX460. The command that gives the error is
nvcc -I/usr/cuda-5.5/include -I…/…/workspace-kepler/global -arch=sm_20 -c src/main.cu -o obj/src/main.o
This file main.cu contains the main routine as well as the CUDA-code. I made sure to not include any of the CUDA-header files.
Thanks in advance