added FMA4 intrinsics for 1hv complex code path

246a6d87 · Alexander Heinecke · fbb0e6eb · 246a6d87 · 246a6d87
Commit 246a6d87 authored Jan 04, 2013 by Alexander Heinecke
--- a/ELPA_2011.12.Intrinsics/src/elpa2_kernels/elpa2_tum_kernels_complex_sse-avx_1hv.cpp
+++ b/ELPA_2011.12.Intrinsics/src/elpa2_kernels/elpa2_tum_kernels_complex_sse-avx_1hv.cpp
--- a/ELPA_2011.12.Intrinsics/test/Makefile
+++ b/ELPA_2011.12.Intrinsics/test/Makefile
@@ -21,8 +21,8 @@ LIBS = -mkl=sequential -L$(MKL_HOME) -lmkl_scalapack_lp64 -lmkl_blacs_intelmpi_l
 #F90OPT=$(F90) -msse3
 #CC=gcc -O3
 #CCOPT=$(CC) -msse3 -funsafe-loop-optimizations -funsafe-math-optimizations -ftree-vect-loop-version -ftree-vectorize
-#CC=icc -O3
-#CCOPT=$(CC) -msse3
+##CC=icc -O3
+##CCOPT=$(CC) -msse3
 #MKL_HOME=/opt/intel/mkl/lib/intel64
 #LIBS = -mkl=sequential -L$(MKL_HOME) -lmkl_scalapack_lp64 -lmkl_blacs_intelmpi_lp64
 #
@@ -34,7 +34,9 @@ LIBS = -mkl=sequential -L$(MKL_HOME) -lmkl_scalapack_lp64 -lmkl_blacs_intelmpi_l
 #CC=gcc -O3
 #F90OPT=$(F90) -msse3
 #CCOPT=$(CC) -funsafe-loop-optimizations -funsafe-math-optimizations -ftree-vect-loop-version -ftree-vectorize -mfma4 -mxop -march=bdver1 -D__USE_AVX128__
-#LIBS = -L/opt/acml5.0.0/gfortran64_fma4/lib/ -lacml -lgfortran libscalapack.a
+##CCOPT=$(CC) -funsafe-loop-optimizations -funsafe-math-optimizations -ftree-vect-loop-version -ftree-vectorize -mfma4 -mxop -march=bdver1
+##LIBS = -L/opt/acml5.0.0/gfortran64_fma4/lib/ -lacml -lgfortran libscalapack.a
+#LIBS = -L/lrz/sys/libraries/acml/5.2.0/ifort64_fma4_mp/lib -lacml_mp -lgfortran libscalapack.a
 #
 # ------------------------------------------------------------------------------
 # Settings for Intel Fortran (Linux) old 11.x Toolchain, !!!!!!! do not use !!!!!!: