David M. Cooke <cookedm@physics.mcmaster.ca> [2007-11-27 05:54]:
I'm guessing that numpy.distutils isn't able to tell whether your compiler is intele or not. From numpy/distutils/fcompiler/ intel.py, the command run to find the version is
$ ifort -FI -V -c dummy.f -o dummy.o
where dummy.f is just a file with a dummy subroutine in it. Could you do this and give us the output?
Hello David, No Itanium here, but this is the output of the 32-bit version. ============================================================================== $ ifort -FI -V -c foo.f -o foo.o Intel(R) Fortran Compiler for applications running on IA-32, Version 10.1 Build 20070913 Package ID: l_fc_p_10.1.008 Copyright (C) 1985-2007 Intel Corporation. All rights reserved. FOR NON-COMMERCIAL USE ONLY Intel Fortran 10.0-2024 ================================================================================ There's a distutils problem using Intel's MKL in NumPy-1.04 and NumPy SVN 4510 (and below). In MKL 9.1 & 10.0, mkl_lapack32 & mkl_lapack64 no longer exist. They were replaced by /32/mkl_lapack and /64/mkl_lapack. As a result, numpy/distutils/system_info.py needs to be changed: class lapack_mkl_info(mkl_info): def calc_info(self): mkl = get_info('mkl') if not mkl: return if sys.platform == 'win32': lapack_libs = self.get_libs('lapack_libs',['mkl_lapack']) else: #lapack_libs = self.get_libs('lapack_libs',['mkl_lapack32','mkl_lapack64']) lapack_libs = self.get_libs('lapack_libs',['mkl_lapack']) If this change is not made NumPy builds, but the MKL is not used. I reported this problem earlier on the NumPy list. -rex