Improve arm64 support. (#676)

* Improve Arm64 support.

Make library compile and run on Arm64 8.2 (Nvidia Xavier arm+cuda)
`configure` is regenerated

* remove simd from utils.cpp

acinclude/ax_check_cpu.m4

@@ -0,0 +1,25 @@
+
+# serial 1
+
+AC_DEFUN([AX_CPU_ARCH], [
+
+AC_MSG_CHECKING([for cpu arch])
+AC_CANONICAL_TARGET
+
+case $target in
+  amd64-* | x86_64-*)
+    ARCH_CPUFLAGS="-msse4 -mpopcnt"
+    ARCH_CXXFLAGS="-m64"
+    ;;
+  aarch64*-*)
+dnl This is an arch for Nvidia Xavier a proper detection would be nice.
+    ARCH_CPUFLAGS="-march=armv8.2-a"
+    ;;
+  *) ;;
+esac
+AC_MSG_RESULT([$target CPUFLAGS+=$ARCH_CPUFLAGS CXXFLAGS+=$ARCH_CXXFLAGS])
+
+AC_SUBST(ARCH_CPUFLAGS)
+AC_SUBST(ARCH_CXXFLAGS)
+
+])dnl

configure

@@ -626,6 +626,12 @@ ac_includes_default="\
 
 ac_header_list=
 ac_subst_vars='LTLIBOBJS
+ARCH_CXXFLAGS
+ARCH_CPUFLAGS
+target_os
+target_vendor
+target_cpu
+target
 LAPACK_LIBS
 OPENMP_LDFLAGS
 BLAS_LIBS
@@ -686,6 +692,7 @@ infodir
 docdir
 oldincludedir
 includedir
+runstatedir
 localstatedir
 sharedstatedir
 sysconfdir
@@ -767,6 +774,7 @@ datadir='${datarootdir}'
 sysconfdir='${prefix}/etc'
 sharedstatedir='${prefix}/com'
 localstatedir='${prefix}/var'
+runstatedir='${localstatedir}/run'
 includedir='${prefix}/include'
 oldincludedir='/usr/include'
 docdir='${datarootdir}/doc/${PACKAGE_TARNAME}'
@@ -1019,6 +1027,15 @@ do
   | -silent | --silent | --silen | --sile | --sil)
     silent=yes ;;
 
+  -runstatedir | --runstatedir | --runstatedi | --runstated \
+  | --runstate | --runstat | --runsta | --runst | --runs \
+  | --run | --ru | --r)
+    ac_prev=runstatedir ;;
+  -runstatedir=* | --runstatedir=* | --runstatedi=* | --runstated=* \
+  | --runstate=* | --runstat=* | --runsta=* | --runst=* | --runs=* \
+  | --run=* | --ru=* | --r=*)
+    runstatedir=$ac_optarg ;;
+
   -sbindir | --sbindir | --sbindi | --sbind | --sbin | --sbi | --sb)
     ac_prev=sbindir ;;
   -sbindir=* | --sbindir=* | --sbindi=* | --sbind=* | --sbin=* \
@@ -1156,7 +1173,7 @@ fi
 for ac_var in	exec_prefix prefix bindir sbindir libexecdir datarootdir \
 		datadir sysconfdir sharedstatedir localstatedir includedir \
 		oldincludedir docdir infodir htmldir dvidir pdfdir psdir \
-		libdir localedir mandir
+		libdir localedir mandir runstatedir
 do
   eval ac_val=\$$ac_var
   # Remove trailing slashes.
@@ -1309,6 +1326,7 @@ Fine tuning of the installation directories:
   --sysconfdir=DIR        read-only single-machine data [PREFIX/etc]
   --sharedstatedir=DIR    modifiable architecture-independent data [PREFIX/com]
   --localstatedir=DIR     modifiable single-machine data [PREFIX/var]
+  --runstatedir=DIR       modifiable per-process data [LOCALSTATEDIR/run]
   --libdir=DIR            object code libraries [EPREFIX/lib]
   --includedir=DIR        C header files [PREFIX/include]
   --oldincludedir=DIR     C header files for non-gcc [/usr/include]
@@ -1329,6 +1347,7 @@ _ACEOF
 System types:
   --build=BUILD     configure for building on BUILD [guessed]
   --host=HOST       cross-compile to build programs to run on HOST [BUILD]
+  --target=TARGET   configure for building compilers for TARGET [HOST]
 _ACEOF
 fi
 
@@ -5455,6 +5474,8 @@ main ()
     if (*(data + i) != *(data3 + i))
       return 14;
   close (fd);
+  free (data);
+  free (data3);
   return 0;
 }
 _ACEOF
@@ -6814,6 +6835,68 @@ if test "x$ax_lapack_ok" == "xno"; then
   as_fn_error $? "An implementation of LAPACK is required but none was found." "$LINENO" 5
 fi
 
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking target system type" >&5
+$as_echo_n "checking target system type... " >&6; }
+if ${ac_cv_target+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test "x$target_alias" = x; then
+  ac_cv_target=$ac_cv_host
+else
+  ac_cv_target=`$SHELL "$ac_aux_dir/config.sub" $target_alias` ||
+    as_fn_error $? "$SHELL $ac_aux_dir/config.sub $target_alias failed" "$LINENO" 5
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_target" >&5
+$as_echo "$ac_cv_target" >&6; }
+case $ac_cv_target in
+*-*-*) ;;
+*) as_fn_error $? "invalid value of canonical target" "$LINENO" 5;;
+esac
+target=$ac_cv_target
+ac_save_IFS=$IFS; IFS='-'
+set x $ac_cv_target
+shift
+target_cpu=$1
+target_vendor=$2
+shift; shift
+# Remember, the first character of IFS is used to create $*,
+# except with old shells:
+target_os=$*
+IFS=$ac_save_IFS
+case $target_os in *\ *) target_os=`echo "$target_os" | sed 's/ /-/g'`;; esac
+
+
+# The aliases save the names the user supplied, while $host etc.
+# will get canonicalized.
+test -n "$target_alias" &&
+  test "$program_prefix$program_suffix$program_transform_name" = \
+    NONENONEs,x,x, &&
+  program_prefix=${target_alias}-
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for cpu arch" >&5
+$as_echo_n "checking for cpu arch... " >&6; }
+
+
+case $target in
+  amd64-* | x86_64-*)
+    ARCH_CPUFLAGS="-msse4 -mpopcnt"
+    ARCH_CXXFLAGS="-m64"
+    ;;
+  aarch64*-*)
+    ARCH_CPUFLAGS="-march=armv8.2-a"
+    ;;
+  *) ;;
+esac
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $target CPUFLAGS+=$ARCH_CPUFLAGS CXXFLAGS+=$ARCH_CXXFLAGS" >&5
+$as_echo "$target CPUFLAGS+=$ARCH_CPUFLAGS CXXFLAGS+=$ARCH_CXXFLAGS" >&6; }
+
+
+
+
+
 
 ac_config_files="$ac_config_files makefile.inc"
 

configure.ac

@@ -64,6 +64,7 @@ if test "x$ax_lapack_ok" == "xno"; then
   AC_MSG_ERROR([An implementation of LAPACK is required but none was found.])
 fi
 
+AX_CPU_ARCH
 
 AC_CONFIG_FILES([makefile.inc])
 AC_OUTPUT

makefile.inc.in

@@ -9,8 +9,8 @@ CXXCPP       = @CXXCPP@
 # TODO: Investigate the LAPACKE wrapper for LAPACK, which defines the correct
 #   type for FORTRAN integers.
 CPPFLAGS     = -DFINTEGER=int @CPPFLAGS@
-CXXFLAGS     = -fPIC @OPENMP_CXXFLAGS@ -m64 -Wno-sign-compare @CXXFLAGS@
-CPUFLAGS     = -msse4 -mpopcnt
+CXXFLAGS     = -fPIC @OPENMP_CXXFLAGS@ @ARCH_CXXFLAGS@ -Wno-sign-compare @CXXFLAGS@
+CPUFLAGS     = @ARCH_CPUFLAGS@
 LDFLAGS      = @OPENMP_LDFLAGS@ @LDFLAGS@
 LIBS         = @BLAS_LIBS@ @LAPACK_LIBS@ @LIBS@
 PYTHONCFLAGS = @PYTHON_CFLAGS@ -I@NUMPY_INCLUDE@

utils.cpp

@@ -21,8 +21,6 @@
 
 #include <omp.h>
 
-#include <immintrin.h>
-
 #include <algorithm>
 #include <vector>