Actually I'd have to clarify how it "doesn't work": the siesta binary compiles, 
but segfaults on any test in Tests.


From: Grigory Shamov 
<[email protected]<mailto:[email protected]>>
Reply-To: "[email protected]<mailto:[email protected]>" 
<[email protected]<mailto:[email protected]>>
Date: Tuesday, 9 June, 2015 11:38 AM
To: "[email protected]<mailto:[email protected]>" 
<[email protected]<mailto:[email protected]>>
Subject: [SIESTA-L] compiling with Intel Compilers v 14 - 15?

Hi All,

I am having difficulty building the recent SIESTA 3.2 pl 5 with newer versions 
of Intel compilers. I use NetCDF and OpenMPI 1.6.5, built with the same 
compilers with usual flags. However, optimization level –O2 doesn't work for 
newer Intel compilers versions (14.0.2.144, 15.0.0.090) but works with the old 
Intel (12.1.5.339). -O1 works.

Has anyone tried using the new Intel 14, Intel 15 Fortrans to build SIESTA; did 
it work for you? Are there any compiler options to make it to compile with a 
decent optimization level?

Thank you in advance!

===== my arch.make goes below ====
FPP=
FPP_OUTPUT=
FC=mpif90
RANLIB=ranlib

SYS=nag

SP_KIND=4
DP_KIND=8
KINDS=$(SP_KIND) $(DP_KIND)

FFLAGS=-g -O2  -xSSE4.2 -msse4.2
FPPFLAGS= -DMPI -DFC_HAVE_FLUSH -DFC_HAVE_ABORT -DGRID_DP -DPHI_GRID_SP -DCDF
LDFLAGS=

ARFLAGS_EXTRA=

FCFLAGS_fixed_f=
FCFLAGS_free_f90=
FPPFLAGS_fixed_F=
FPPFLAGS_free_F90=

BLAS_LIBS=-mkl=sequential
LAPACK_LIBS=-mkl=sequential
BLACS_LIBS=-L$(MKLROOT)/lib/intel64 -lmkl_blacs_openmpi_lp64
SCALAPACK_LIBS=-L$(MKLROOT)/lib/intel64 -lmkl_scalapack_lp64

COMP_LIBS=dc_lapack.a

NETCDF_LIBS=-L/global/software/netcdf/netcdf-4.1.3-intel14-66 -lnetcdff -lnetcdf
NETCDF_INTERFACE=-I/global/software/netcdf/netcdf-4.1.3-intel14-66/include
NETCDF_INCFLAGS=-I/global/software/netcdf/netcdf-4.1.3-intel14-66/include


--
Grigory Shamov
Westgrid/ComputeCanada Site Lead
University of Manitoba
E2-588 EITC Building,
(204) 474-9625


Responder a