Problem with make & Intel's MKL

Discussion created by jski on Oct 13, 2008

I'm working to get numbers comparing double_precision_optimized_matmult vs. MKL's cblas_dgemm().

When I try to add the MKL code to double_precision_optimized_matmult and make it, I get an error resulting from BRCC's preprocessor phase can't find MKL_INT in /opt/intel/mkl/ - even though it's included via:

#include <mkl_cblas.h>

If I edit /usr/local/amdbrook/samples/utils/build/config.mk, these apply to the g++ phase of the build.

How do I instruct BRCC to resolve the includes to mkl_types.h?