Commit 79c5ee3b authored by Peter W. Draper's avatar Peter W. Draper
Browse files

Merge branch 'master' into initial_partitions

parents 77c32527 ad76757d
......@@ -148,10 +148,14 @@ if test "$enable_opt" = "yes" ; then
ac_test_CFLAGS="yes"
CFLAGS="$old_CFLAGS $CFLAGS"
# Check SSE & AVX support (some overlap with AX_CC_MAXOPT).
# Check SSE & AVX support (some overlap with AX_CC_MAXOPT).
# Don't use the SIMD_FLAGS result with Intel compilers. The -x<code>
# value from AX_CC_MAXOPT should be sufficient.
AX_EXT
if test "$SIMD_FLAGS" != ""; then
CFLAGS="$CFLAGS $SIMD_FLAGS"
if test "$ax_cv_c_compiler_vendor" != "intel"; then
CFLAGS="$CFLAGS $SIMD_FLAGS"
fi
fi
fi
......
......@@ -49,7 +49,7 @@
# modified version of the Autoconf Macro, you may extend this special
# exception to the GPL to apply to your modified version as well.
#serial 5
#serial 6
AC_DEFUN([AX_APPEND_FLAG],
[dnl
......@@ -59,7 +59,7 @@ AS_VAR_SET_IF(FLAGS,[
AS_CASE([" AS_VAR_GET(FLAGS) "],
[*" $1 "*], [AC_RUN_LOG([: FLAGS already contains $1])],
[
AS_VAR_APPEND(FLAGS," $1")
AS_VAR_APPEND(FLAGS,[" $1"])
AC_RUN_LOG([: FLAGS="$FLAGS"])
])
],
......
......@@ -125,7 +125,7 @@ if test "$ac_test_CFLAGS" != "set"; then
*1?6[[aef]]?:*:*:*|*2?6[[5cef]]?:*:*:*) icc_flags="-xSSE4.2 -xS -xT -xB -xK" ;;
*2?6[[ad]]?:*:*:*) icc_flags="-xAVX -SSE4.2 -xS -xT -xB -xK" ;;
*3?6[[ae]]?:*:*:*) icc_flags="-xCORE-AVX-I -xAVX -SSE4.2 -xS -xT -xB -xK" ;;
*3?6[[cf]]?:*:*:*|*4?6[[56]]?:*:*:*) icc_flags="-xCORE-AVX2 -xCORE-AVX-I -xAVX -SSE4.2 -xS -xT -xB -xK" ;;
*3?6[[cf]]?:*:*:*|*4?6[[56]]?:*:*:*|*4?6[[ef]]?:*:*:*) icc_flags="-xCORE-AVX2 -xCORE-AVX-I -xAVX -SSE4.2 -xS -xT -xB -xK" ;;
*000?f[[346]]?:*:*:*|?f[[346]]?:*:*:*|f[[346]]?:*:*:*) icc_flags="-xSSE3 -xP -xO -xN -xW -xK" ;;
*00??f??:*:*:*|??f??:*:*:*|?f??:*:*:*|f??:*:*:*) icc_flags="-xSSE2 -xN -xW -xK" ;;
esac ;;
......
......@@ -8,23 +8,35 @@
#
# DESCRIPTION
#
# Find supported SIMD extensions by requesting cpuid. When an SIMD
# Find supported SIMD extensions by requesting cpuid. When a SIMD
# extension is found, the -m"simdextensionname" is added to SIMD_FLAGS if
# compiler supports it. For example, if "sse2" is available, then "-msse2"
# compiler supports it. For example, if "sse2" is available then "-msse2"
# is added to SIMD_FLAGS.
#
# Find other supported CPU extensions by requesting cpuid. When a
# processor extension is found, the -m"extensionname" is added to
# CPUEXT_FLAGS if compiler supports it. For example, if "bmi2" is
# available then "-mbmi2" is added to CPUEXT_FLAGS.
#
# This macro calls:
#
# AC_SUBST(SIMD_FLAGS)
# AC_SUBST(CPUEXT_FLAGS)
#
# And defines:
#
# HAVE_MMX / HAVE_SSE / HAVE_SSE2 / HAVE_SSE3 / HAVE_SSSE3 / HAVE_SSE4.1 / HAVE_SSE4.2 / HAVE_AVX
# HAVE_RDRND / HAVE_BMI1 / HAVE_BMI2 / HAVE_ADX / HAVE_MPX
# HAVE_PREFETCHWT1 / HAVE_ABM / HAVE_MMX / HAVE_SSE / HAVE_SSE2
# HAVE_SSE3 / HAVE_SSSE3 / HAVE_SSE4_1 / HAVE_SSE4_2 / HAVE_SSE4a
# HAVE_SHA / HAVE_AES / HAVE_AVX / HAVE_FMA3 / HAVE_FMA4 / HAVE_XOP
# HAVE_AVX2 / HAVE_AVX512_F / HAVE_AVX512_CD / HAVE_AVX512_PF
# HAVE_AVX512_ER / HAVE_AVX512_VL / HAVE_AVX512_BW / HAVE_AVX512_DQ
# HAVE_AVX512_IFMA / HAVE_AVX512_VBMI
#
# LICENSE
#
# Copyright (c) 2007 Christophe Tournayre <turn3r@users.sourceforge.net>
# Copyright (c) 2013 Michael Petch <mpetch@capp-sysware.com>
# Copyright (c) 2013,2015 Michael Petch <mpetch@capp-sysware.com>
#
# Copying and distribution of this file, with or without modification, are
# permitted in any medium without royalty provided the copyright notice
......@@ -32,14 +44,18 @@
# warranty.
# SWIFT modifications. Intel compilers accept these flags, but warn about
# the order of the flags when more than one is used. Given that reverse
# the check order and break on the highest accepted.
# the order of the flags when more than one is used. Given that we just
# set SIMD_FLAGS to the most specific value, rather than all accepted ones.
#serial 13
#serial 15
AC_DEFUN([AX_EXT],
[
AC_REQUIRE([AC_CANONICAL_HOST])
AC_REQUIRE([AC_PROG_CC])
CPUEXT_FLAGS=""
SIMD_FLAGS=""
case $host_cpu in
powerpc*)
......@@ -58,199 +74,234 @@ AC_DEFUN([AX_EXT],
fi
;;
i[[3456]]86*|x86_64*|amd64*)
AC_REQUIRE([AX_GCC_X86_CPUID])
AC_REQUIRE([AX_GCC_X86_CPUID_COUNT])
AC_REQUIRE([AX_GCC_X86_AVX_XGETBV])
AX_GCC_X86_CPUID(0x00000001)
ecx=0
edx=0
if test "$ax_cv_gcc_x86_cpuid_0x00000001" != "unknown";
eax_cpuid0=0
AX_GCC_X86_CPUID(0x00000000)
if test "$ax_cv_gcc_x86_cpuid_0x00000000" != "unknown";
then
ecx=`echo $ax_cv_gcc_x86_cpuid_0x00000001 | cut -d ":" -f 3`
edx=`echo $ax_cv_gcc_x86_cpuid_0x00000001 | cut -d ":" -f 4`
eax_cpuid0=`echo $ax_cv_gcc_x86_cpuid_0x00000000 | cut -d ":" -f 1`
fi
AC_CACHE_CHECK([whether mmx is supported], [ax_cv_have_mmx_ext],
[
ax_cv_have_mmx_ext=no
if test "$((0x$edx>>23&0x01))" = 1; then
ax_cv_have_mmx_ext=yes
fi
])
AC_CACHE_CHECK([whether sse is supported], [ax_cv_have_sse_ext],
[
ax_cv_have_sse_ext=no
if test "$((0x$edx>>25&0x01))" = 1; then
ax_cv_have_sse_ext=yes
fi
])
eax_cpuid80000000=0
AX_GCC_X86_CPUID(0x80000000)
if test "$ax_cv_gcc_x86_cpuid_0x80000000" != "unknown";
then
eax_cpuid80000000=`echo $ax_cv_gcc_x86_cpuid_0x80000000 | cut -d ":" -f 1`
fi
AC_CACHE_CHECK([whether sse2 is supported], [ax_cv_have_sse2_ext],
[
ax_cv_have_sse2_ext=no
if test "$((0x$edx>>26&0x01))" = 1; then
ax_cv_have_sse2_ext=yes
ecx_cpuid1=0
edx_cpuid1=0
if test "$((0x$eax_cpuid0))" -ge 1 ; then
AX_GCC_X86_CPUID(0x00000001)
if test "$ax_cv_gcc_x86_cpuid_0x00000001" != "unknown";
then
ecx_cpuid1=`echo $ax_cv_gcc_x86_cpuid_0x00000001 | cut -d ":" -f 3`
edx_cpuid1=`echo $ax_cv_gcc_x86_cpuid_0x00000001 | cut -d ":" -f 4`
fi
])
fi
AC_CACHE_CHECK([whether sse3 is supported], [ax_cv_have_sse3_ext],
[
ax_cv_have_sse3_ext=no
if test "$((0x$ecx&0x01))" = 1; then
ax_cv_have_sse3_ext=yes
ebx_cpuid7=0
ecx_cpuid7=0
if test "$((0x$eax_cpuid0))" -ge 7 ; then
AX_GCC_X86_CPUID_COUNT(0x00000007, 0x00)
if test "$ax_cv_gcc_x86_cpuid_0x00000007" != "unknown";
then
ebx_cpuid7=`echo $ax_cv_gcc_x86_cpuid_0x00000007 | cut -d ":" -f 2`
ecx_cpuid7=`echo $ax_cv_gcc_x86_cpuid_0x00000007 | cut -d ":" -f 3`
fi
])
fi
AC_CACHE_CHECK([whether ssse3 is supported], [ax_cv_have_ssse3_ext],
[
ax_cv_have_ssse3_ext=no
if test "$((0x$ecx>>9&0x01))" = 1; then
ax_cv_have_ssse3_ext=yes
ecx_cpuid80000001=0
edx_cpuid80000001=0
if test "$((0x$eax_cpuid80000000))" -ge "$((0x80000001))" ; then
AX_GCC_X86_CPUID(0x80000001)
if test "$ax_cv_gcc_x86_cpuid_0x80000001" != "unknown";
then
ecx_cpuid80000001=`echo $ax_cv_gcc_x86_cpuid_0x80000001 | cut -d ":" -f 3`
edx_cpuid80000001=`echo $ax_cv_gcc_x86_cpuid_0x80000001 | cut -d ":" -f 4`
fi
])
fi
AC_CACHE_CHECK([whether sse4.1 is supported], [ax_cv_have_sse41_ext],
AC_CACHE_VAL([ax_cv_have_mmx_os_support_ext],
[
ax_cv_have_sse41_ext=no
if test "$((0x$ecx>>19&0x01))" = 1; then
ax_cv_have_sse41_ext=yes
fi
ax_cv_have_mmx_os_support_ext=yes
])
AC_CACHE_CHECK([whether sse4.2 is supported], [ax_cv_have_sse42_ext],
[
ax_cv_have_sse42_ext=no
if test "$((0x$ecx>>20&0x01))" = 1; then
ax_cv_have_sse42_ext=yes
fi
])
ax_cv_have_none_os_support_ext=yes
AC_CACHE_CHECK([whether avx is supported by processor], [ax_cv_have_avx_cpu_ext],
AC_CACHE_VAL([ax_cv_have_sse_os_support_ext],
[
ax_cv_have_avx_cpu_ext=no
if test "$((0x$ecx>>28&0x01))" = 1; then
ax_cv_have_avx_cpu_ext=yes
ax_cv_have_sse_os_support_ext=no,
if test "$((0x$edx_cpuid1>>25&0x01))" = 1; then
AC_LANG_PUSH([C])
AC_TRY_RUN([
#include <signal.h>
#include <stdlib.h>
/* No way at ring1 to ring3 in protected mode to check the CR0 and CR4
control registers directly. Execute an SSE instruction.
If it raises SIGILL then OS doesn't support SSE based instructions */
void sig_handler(int signum){ exit(1); }
int main(){
signal(SIGILL, sig_handler);
/* SSE instruction xorps %xmm0,%xmm0 */
__asm__ __volatile__ (".byte 0x0f, 0x57, 0xc0");
return 0;
}],
ax_cv_have_sse_os_support_ext=yes,
ax_cv_have_sse_os_support_ext=no,
ax_cv_have_sse_os_support_ext=no)
AC_LANG_POP([C])
fi
])
if test x"$ax_cv_have_avx_cpu_ext" = x"yes"; then
xgetbv_eax=0
if test "$((0x$ecx_cpuid1>>28&0x01))" = 1; then
AX_GCC_X86_AVX_XGETBV(0x00000000)
xgetbv_eax="0"
if test x"$ax_cv_gcc_x86_avx_xgetbv_0x00000000" != x"unknown"; then
xgetbv_eax=`echo $ax_cv_gcc_x86_avx_xgetbv_0x00000000 | cut -d ":" -f 1`
fi
AC_CACHE_CHECK([whether avx is supported by operating system], [ax_cv_have_avx_ext],
AC_CACHE_VAL([ax_cv_have_avx_os_support_ext],
[
ax_cv_have_avx_ext=no
if test "$((0x$ecx>>27&0x01))" = 1; then
ax_cv_have_avx_os_support_ext=no
if test "$((0x$ecx_cpuid1>>27&0x01))" = 1; then
if test "$((0x$xgetbv_eax&0x6))" = 6; then
ax_cv_have_avx_ext=yes
ax_cv_have_avx_os_support_ext=yes
fi
fi
])
if test x"$ax_cv_have_avx_ext" = x"no"; then
AC_MSG_WARN([Your processor supports AVX, but your operating system doesn't])
fi
fi
checkmore=yes
if test "$ax_cv_have_avx_ext" = yes; then
AX_CHECK_COMPILE_FLAG(-mavx, ax_cv_support_avx_ext=yes, [])
if test x"$ax_cv_support_avx_ext" = x"yes"; then
SIMD_FLAGS="$SIMD_FLAGS -mavx"
AC_DEFINE(HAVE_AVX,,[Support AVX (Advanced Vector Extensions) instructions])
checkmore=no
else
AC_MSG_WARN([Your processor supports avx instructions but not your compiler, can you try another compiler?])
fi
fi
if test $checkmore = yes -a "$ax_cv_have_sse42_ext" = yes; then
AX_CHECK_COMPILE_FLAG(-msse4.2, ax_cv_support_sse42_ext=yes, [])
if test x"$ax_cv_support_sse42_ext" = x"yes"; then
SIMD_FLAGS="$SIMD_FLAGS -msse4.2"
AC_DEFINE(HAVE_SSE4_2,,[Support SSSE4.2 (Streaming SIMD Extensions 4.2) instructions])
checkmore=no
else
AC_MSG_WARN([Your processor supports sse4.2 instructions but not your compiler, can you try another compiler?])
fi
fi
if test $checkmore = yes -a "$ax_cv_have_sse41_ext" = yes; then
AX_CHECK_COMPILE_FLAG(-msse4.1, ax_cv_support_sse41_ext=yes, [])
if test x"$ax_cv_support_sse41_ext" = x"yes"; then
SIMD_FLAGS="$SIMD_FLAGS -msse4.1"
AC_DEFINE(HAVE_SSE4_1,,[Support SSSE4.1 (Streaming SIMD Extensions 4.1) instructions])
checkmore=no
else
AC_MSG_WARN([Your processor supports sse4.1 instructions but not your compiler, can you try another compiler?])
fi
fi
if test $checkmore = yes -a "$ax_cv_have_ssse3_ext" = yes; then
AX_CHECK_COMPILE_FLAG(-mssse3, ax_cv_support_ssse3_ext=yes, [])
if test x"$ax_cv_support_ssse3_ext" = x"yes"; then
SIMD_FLAGS="$SIMD_FLAGS -mssse3"
AC_DEFINE(HAVE_SSSE3,,[Support SSSE3 (Supplemental Streaming SIMD Extensions 3) instructions])
checkmore=no
else
AC_MSG_WARN([Your processor supports ssse3 instructions but not your compiler, can you try another compiler?])
fi
fi
if test $checkmore = yes -a "$ax_cv_have_sse3_ext" = yes; then
AX_CHECK_COMPILE_FLAG(-msse3, ax_cv_support_sse3_ext=yes, [])
if test x"$ax_cv_support_sse3_ext" = x"yes"; then
SIMD_FLAGS="$SIMD_FLAGS -msse3"
AC_DEFINE(HAVE_SSE3,,[Support SSE3 (Streaming SIMD Extensions 3) instructions])
checkmore=no
else
AC_MSG_WARN([Your processor supports sse3 instructions but not your compiler, can you try another compiler?])
fi
fi
if test $checkmore = yes -a "$ax_cv_have_sse2_ext" = yes; then
AX_CHECK_COMPILE_FLAG(-msse2, ax_cv_support_sse2_ext=yes, [])
if test x"$ax_cv_support_sse2_ext" = x"yes"; then
SIMD_FLAGS="$SIMD_FLAGS -msse2"
AC_DEFINE(HAVE_SSE2,,[Support SSE2 (Streaming SIMD Extensions 2) instructions])
checkmore=no
else
AC_MSG_WARN([Your processor supports sse2 instructions but not your compiler, can you try another compiler?])
fi
fi
if test $checkmore = yes -a "$ax_cv_have_sse_ext" = yes; then
AX_CHECK_COMPILE_FLAG(-msse, ax_cv_support_sse_ext=yes, [])
if test x"$ax_cv_support_sse_ext" = x"yes"; then
SIMD_FLAGS="$SIMD_FLAGS -msse"
AC_DEFINE(HAVE_SSE,,[Support SSE (Streaming SIMD Extensions) instructions])
checkmore=no
else
AC_MSG_WARN([Your processor supports sse instructions but not your compiler, can you try another compiler?])
fi
fi
if test $checkmore = yes -a "$ax_cv_have_mmx_ext" = yes; then
AX_CHECK_COMPILE_FLAG(-mmmx, ax_cv_support_mmx_ext=yes, [])
if test x"$ax_cv_support_mmx_ext" = x"yes"; then
SIMD_FLAGS="$SIMD_FLAGS -mmmx"
AC_DEFINE(HAVE_MMX,,[Support mmx instructions])
else
AC_MSG_WARN([Your processor supports mmx instructions but not your compiler, can you try another compiler?])
AC_CACHE_VAL([ax_cv_have_avx512_os_support_ext],
[
ax_cv_have_avx512_os_support_ext=no
if test "$ax_cv_have_avx_os_support_ext" = yes; then
if test "$((0x$xgetbv_eax&0xe6))" = "$((0xe6))"; then
ax_cv_have_avx512_os_support_ext=yes
fi
fi
fi
])
for ac_instr_info dnl
in "none;rdrnd;RDRND;ecx_cpuid1,30;-mrdrnd;HAVE_RDRND;CPUEXT_FLAGS" dnl
"none;bmi1;BMI1;ebx_cpuid7,3;-mbmi;HAVE_BMI1;CPUEXT_FLAGS" dnl
"none;bmi2;BMI2;ebx_cpuid7,8;-mbmi2;HAVE_BMI2;CPUEXT_FLAGS" dnl
"none;adx;ADX;ebx_cpuid7,19;-madx;HAVE_ADX;CPUEXT_FLAGS" dnl
"none;mpx;MPX;ebx_cpuid7,14;-mmpx;HAVE_MPX;CPUEXT_FLAGS" dnl
"none;prefetchwt1;PREFETCHWT1;ecx_cpuid7,0;-mprefetchwt1;HAVE_PREFETCHWT1;CPUEXT_FLAGS" dnl
"none;abm;ABM;ecx_cpuid80000001,5;-mabm;HAVE_ABM;CPUEXT_FLAGS" dnl
"mmx;mmx;MMX;edx_cpuid1,23;-mmmx;HAVE_MMX;SIMD_FLAGS" dnl
"sse;sse;SSE;edx_cpuid1,25;-msse;HAVE_SSE;SIMD_FLAGS" dnl
"sse;sse2;SSE2;edx_cpuid1,26;-msse2;HAVE_SSE2;SIMD_FLAGS" dnl
"sse;sse3;SSE3;ecx_cpuid1,1;-msse3;HAVE_SSE3;SIMD_FLAGS" dnl
"sse;ssse3;SSSE3;ecx_cpuid1,9;-mssse3;HAVE_SSSE3;SIMD_FLAGS" dnl
"sse;sse41;SSE4.1;ecx_cpuid1,19;-msse4.1;HAVE_SSE4_1;SIMD_FLAGS" dnl
"sse;sse42;SSE4.2;ecx_cpuid1,20;-msse4.2;HAVE_SSE4_2;SIMD_FLAGS" dnl
"sse;sse4a;SSE4a;ecx_cpuid80000001,6;-msse4a;HAVE_SSE4a;SIMD_FLAGS" dnl
"sse;sha;SHA;ebx_cpuid7,29;-msha;HAVE_SHA;SIMD_FLAGS" dnl
"sse;aes;AES;ecx_cpuid1,25;-maes;HAVE_AES;SIMD_FLAGS" dnl
"avx;avx;AVX;ecx_cpuid1,28;-mavx;HAVE_AVX;SIMD_FLAGS" dnl
"avx;fma3;FMA3;ecx_cpuid1,12;-mfma;HAVE_FMA3;SIMD_FLAGS" dnl
"avx;fma4;FMA4;ecx_cpuid80000001,16;-mfma4;HAVE_FMA4;SIMD_FLAGS" dnl
"avx;xop;XOP;ecx_cpuid80000001,11;-mxop;HAVE_XOP;SIMD_FLAGS" dnl
"avx;avx2;AVX2;ebx_cpuid7,5;-mavx2;HAVE_AVX2;SIMD_FLAGS" dnl
"avx512;avx512f;AVX512-F;ebx_cpuid7,16;-mavx512f;HAVE_AVX512_F;SIMD_FLAGS" dnl
"avx512;avx512cd;AVX512-CD;ebx_cpuid7,28;-mavx512cd;HAVE_AVX512_CD;SIMD_FLAGS" dnl
"avx512;avx512pf;AVX512-PF;ebx_cpuid7,26;-mavx512pf;HAVE_AVX512_PF;SIMD_FLAGS" dnl
"avx512;avx512er;AVX512-ER;ebx_cpuid7,27;-mavx512er;HAVE_AVX512_ER;SIMD_FLAGS" dnl
"avx512;avx512vl;AVX512-VL;ebx_cpuid7,31;-mavx512vl;HAVE_AVX512_VL;SIMD_FLAGS" dnl
"avx512;avx512bw;AVX512-BW;ebx_cpuid7,30;-mavx512bw;HAVE_AVX512_BW;SIMD_FLAGS" dnl
"avx512;avx512dq;AVX512-DQ;ebx_cpuid7,17;-mavx512dq;HAVE_AVX512_DQ;SIMD_FLAGS" dnl
"avx512;avx512ifma;AVX512-IFMA;ebx_cpuid7,21;-mavx512ifma;HAVE_AVX512_IFMA;SIMD_FLAGS" dnl
"avx512;avx512vbmi;AVX512-VBMI;ecx_cpuid7,1;-mavx512vbmi;HAVE_AVX512_VBMI;SIMD_FLAGS" dnl
#
do ac_instr_os_support=$(eval echo \$ax_cv_have_$(echo $ac_instr_info | cut -d ";" -f 1)_os_support_ext)
ac_instr_acvar=$(echo $ac_instr_info | cut -d ";" -f 2)
ac_instr_shortname=$(echo $ac_instr_info | cut -d ";" -f 3)
ac_instr_chk_loc=$(echo $ac_instr_info | cut -d ";" -f 4)
ac_instr_chk_reg=0x$(eval echo \$$(echo $ac_instr_chk_loc | cut -d "," -f 1))
ac_instr_chk_bit=$(echo $ac_instr_chk_loc | cut -d "," -f 2)
ac_instr_compiler_flags=$(echo $ac_instr_info | cut -d ";" -f 5)
ac_instr_have_define=$(echo $ac_instr_info | cut -d ";" -f 6)
ac_instr_flag_type=$(echo $ac_instr_info | cut -d ";" -f 7)
AC_CACHE_CHECK([whether ${ac_instr_shortname} is supported by the processor], [ax_cv_have_${ac_instr_acvar}_cpu_ext],
[
eval ax_cv_have_${ac_instr_acvar}_cpu_ext=no
if test "$((${ac_instr_chk_reg}>>${ac_instr_chk_bit}&0x01))" = 1 ; then
eval ax_cv_have_${ac_instr_acvar}_cpu_ext=yes
fi
])
if test x"$(eval echo \$ax_cv_have_${ac_instr_acvar}_cpu_ext)" = x"yes"; then
AC_CACHE_CHECK([whether ${ac_instr_shortname} is supported by the processor and OS], [ax_cv_have_${ac_instr_acvar}_ext],
[
eval ax_cv_have_${ac_instr_acvar}_ext=no
if test x"${ac_instr_os_support}" = x"yes"; then
eval ax_cv_have_${ac_instr_acvar}_ext=yes
fi
])
if test "$(eval echo \$ax_cv_have_${ac_instr_acvar}_ext)" = yes; then
AX_CHECK_COMPILE_FLAG(${ac_instr_compiler_flags}, eval ax_cv_support_${ac_instr_acvar}_ext=yes,
eval ax_cv_support_${ac_instr_acvar}_ext=no)
if test x"$(eval echo \$ax_cv_support_${ac_instr_acvar}_ext)" = x"yes"; then
eval ${ac_instr_flag_type}=\"${ac_instr_compiler_flags}\"
AC_DEFINE_UNQUOTED([${ac_instr_have_define}])
else
AC_MSG_WARN([Your processor and OS supports ${ac_instr_shortname} instructions but not your compiler, can you try another compiler?])
fi
else
if test x"${ac_instr_os_support}" = x"no"; then
AC_CACHE_VAL(ax_cv_support_${ac_instr_acvar}_ext, eval ax_cv_support_${ac_instr_acvar}_ext=no)
AC_MSG_WARN([Your processor supports ${ac_instr_shortname}, but your OS doesn't])
fi
fi
else
AC_CACHE_VAL(ax_cv_have_${ac_instr_acvar}_ext, eval ax_cv_have_${ac_instr_acvar}_ext=no)
AC_CACHE_VAL(ax_cv_support_${ac_instr_acvar}_ext, eval ax_cv_support_${ac_instr_acvar}_ext=no)
fi
done
;;
esac
AH_TEMPLATE([HAVE_RDRND],[Define to 1 to support Digital Random Number Generator])
AH_TEMPLATE([HAVE_BMI1],[Define to 1 to support Bit Manipulation Instruction Set 1])
AH_TEMPLATE([HAVE_BMI2],[Define to 1 to support Bit Manipulation Instruction Set 2])
AH_TEMPLATE([HAVE_ADX],[Define to 1 to support Multi-Precision Add-Carry Instruction Extensions])
AH_TEMPLATE([HAVE_MPX],[Define to 1 to support Memory Protection Extensions])
AH_TEMPLATE([HAVE_PREFETCHWT1],[Define to 1 to support Prefetch Vector Data Into Caches WT1])
AH_TEMPLATE([HAVE_ABM],[Define to 1 to support Advanced Bit Manipulation])
AH_TEMPLATE([HAVE_MMX],[Define to 1 to support Multimedia Extensions])
AH_TEMPLATE([HAVE_SSE],[Define to 1 to support Streaming SIMD Extensions])
AH_TEMPLATE([HAVE_SSE2],[Define to 1 to support Streaming SIMD Extensions])
AH_TEMPLATE([HAVE_SSE3],[Define to 1 to support Streaming SIMD Extensions 3])
AH_TEMPLATE([HAVE_SSSE3],[Define to 1 to support Supplemental Streaming SIMD Extensions 3])
AH_TEMPLATE([HAVE_SSE4_1],[Define to 1 to support Streaming SIMD Extensions 4.1])
AH_TEMPLATE([HAVE_SSE4_2],[Define to 1 to support Streaming SIMD Extensions 4.2])
AH_TEMPLATE([HAVE_SSE4a],[Define to 1 to support AMD Streaming SIMD Extensions 4a])
AH_TEMPLATE([HAVE_SHA],[Define to 1 to support Secure Hash Algorithm Extension])
AH_TEMPLATE([HAVE_AES],[Define to 1 to support Advanced Encryption Standard New Instruction Set (AES-NI)])
AH_TEMPLATE([HAVE_AVX],[Define to 1 to support Advanced Vector Extensions])
AH_TEMPLATE([HAVE_FMA3],[Define to 1 to support Fused Multiply-Add Extensions 3])
AH_TEMPLATE([HAVE_FMA4],[Define to 1 to support Fused Multiply-Add Extensions 4])
AH_TEMPLATE([HAVE_XOP],[Define to 1 to support eXtended Operations Extensions])
AH_TEMPLATE([HAVE_AVX2],[Define to 1 to support Advanced Vector Extensions 2])
AH_TEMPLATE([HAVE_AVX512_F],[Define to 1 to support AVX-512 Foundation Extensions])
AH_TEMPLATE([HAVE_AVX512_CD],[Define to 1 to support AVX-512 Conflict Detection Instructions])
AH_TEMPLATE([HAVE_AVX512_PF],[Define to 1 to support AVX-512 Conflict Prefetch Instructions])
AH_TEMPLATE([HAVE_AVX512_ER],[Define to 1 to support AVX-512 Exponential & Reciprocal Instructions])
AH_TEMPLATE([HAVE_AVX512_VL],[Define to 1 to support AVX-512 Vector Length Extensions])
AH_TEMPLATE([HAVE_AVX512_BW],[Define to 1 to support AVX-512 Byte and Word Instructions])
AH_TEMPLATE([HAVE_AVX512_DQ],[Define to 1 to support AVX-512 Doubleword and Quadword Instructions])
AH_TEMPLATE([HAVE_AVX512_IFMA],[Define to 1 to support AVX-512 Integer Fused Multiply Add Instructions])
AH_TEMPLATE([HAVE_AVX512_VBMI],[Define to 1 to support AVX-512 Vector Byte Manipulation Instructions])
AC_SUBST(SIMD_FLAGS)
AC_SUBST(CPUEXT_FLAGS)
])
......@@ -54,7 +54,7 @@
# modified version of the Autoconf Macro, you may extend this special
# exception to the GPL to apply to your modified version as well.
#serial 1
#serial 2
AC_DEFUN([AX_GCC_X86_AVX_XGETBV],
[AC_REQUIRE([AC_PROG_CC])
......@@ -64,7 +64,7 @@ AC_CACHE_CHECK(for x86-AVX xgetbv $1 output, ax_cv_gcc_x86_avx_xgetbv_$1,
int op = $1, eax, edx;
FILE *f;
/* Opcodes for xgetbv */
__asm__(".byte 0x0f, 0x01, 0xd0"
__asm__ __volatile__ (".byte 0x0f, 0x01, 0xd0"
: "=a" (eax), "=d" (edx)
: "c" (op));
f = fopen("conftest_xgetbv", "w"); if (!f) return 1;
......
......@@ -5,13 +5,15 @@
# SYNOPSIS
#
# AX_GCC_X86_CPUID(OP)
# AX_GCC_X86_CPUID_COUNT(OP, COUNT)
#
# DESCRIPTION
#
# On Pentium and later x86 processors, with gcc or a compiler that has a
# compatible syntax for inline assembly instructions, run a small program
# that executes the cpuid instruction with input OP. This can be used to
# detect the CPU type.
# detect the CPU type. AX_GCC_X86_CPUID_COUNT takes an additional COUNT
# parameter that gets passed into register ECX before calling cpuid.
#
# On output, the values of the eax, ebx, ecx, and edx registers are stored
# as hexadecimal strings as "eax:ebx:ecx:edx" in the cache variable
......@@ -28,6 +30,7 @@
#
# Copyright (c) 2008 Steven G. Johnson <stevenj@alum.mit.edu>
# Copyright (c) 2008 Matteo Frigo
# Copyright (c) 2015 Michael Petch <mpetch@capp-sysware.com>
#
# This program is free software: you can redistribute it and/or modify it
# under the terms of the GNU General Public License as published by the
......@@ -55,18 +58,25 @@
# modified version of the Autoconf Macro, you may extend this special
# exception to the GPL to apply to your modified version as well.
#serial 7
#serial 9
AC_DEFUN([AX_GCC_X86_CPUID],
[AX_GCC_X86_CPUID_COUNT($1, 0)
])
AC_DEFUN([AX_GCC_X86_CPUID_COUNT],
[AC_REQUIRE([AC_PROG_CC])
AC_LANG_PUSH([C])
AC_CACHE_CHECK(for x86 cpuid $1 output, ax_cv_gcc_x86_cpuid_$1,
[AC_RUN_IFELSE([AC_LANG_PROGRAM([#include <stdio.h>], [
int op = $1, eax, ebx, ecx, edx;
int op = $1, level = $2, eax, ebx, ecx, edx;
FILE *f;
__asm__("cpuid"
: "=a" (eax), "=b" (ebx), "=c" (ecx), "=d" (edx)
: "a" (op));
__asm__ __volatile__ ("xchg %%ebx, %1\n"
"cpuid\n"
"xchg %%ebx, %1\n"
: "=a" (eax), "=r" (ebx), "=c" (ecx), "=d" (edx)
: "a" (op), "2" (level));
f = fopen("conftest_cpuid", "w"); if (!f) return 1;
fprintf(f, "%x:%x:%x:%x\n", eax, ebx, ecx, edx);
fclose(f);
......
......@@ -4,7 +4,7 @@
#
# SYNOPSIS
#
# DX_INIT_DOXYGEN(PROJECT-NAME, DOXYFILE-PATH, [OUTPUT-DIR])
# DX_INIT_DOXYGEN(PROJECT-NAME, [DOXYFILE-PATH], [OUTPUT-DIR], ...)
# DX_DOXYGEN_FEATURE(ON|OFF)
# DX_DOT_FEATURE(ON|OFF)
# DX_HTML_FEATURE(ON|OFF)
......@@ -45,14 +45,19 @@
# Once all the feature defaults have been specified, call DX_INIT_DOXYGEN
# with the following parameters: a one-word name for the project for use
# as a filename base etc., an optional configuration file name (the
# default is 'Doxyfile', the same as Doxygen's default), and an optional
# output directory name (the default is 'doxygen-doc').
# default is '$(srcdir)/Doxyfile', the same as Doxygen's default), and an
# optional output directory name (the default is 'doxygen-doc'). To run
# doxygen multiple times for different configuration files and output
# directories provide more parameters: the second, forth, sixth, etc
# parameter are configuration file names and the third, fifth, seventh,
# etc parameter are output directories. No checking is done to catch
# duplicates.
#
# Automake Support
#
# The following is a template aminclude.am file for use with Automake.
# Make targets and variables values are controlled by the various
# DX_COND_* conditionals set by autoconf.
# The DX_RULES substitution can be used to add all needed rules to the
# Makefile. Note that this is a substitution without being a variable:
# only the @DX_RULES@ syntax will work.
#
# The provided targets are:
#
......@@ -61,9 +66,7 @@
# doxygen-run: Run doxygen, which will generate some of the
# documentation (HTML, CHM, CHI, MAN, RTF, XML)
# but will not do the post processing required
# for the rest of it (PS, PDF, and some MAN).
#
# doxygen-man: Rename some doxygen generated man pages.
# for the rest of it (PS, PDF).
#
# doxygen-ps: Generate doxygen PostScript documentation.
#
......@@ -84,178 +87,17 @@
#
# Then add this variable to MOSTLYCLEANFILES.
#
# ----- begin aminclude.am -------------------------------------
#
# ## --------------------------------- ##
# ## Format-independent Doxygen rules. ##
# ## --------------------------------- ##
#
# if DX_COND_doc
#
# ## ------------------------------- ##
# ## Rules specific for HTML output. ##
# ## ------------------------------- ##
#
# if DX_COND_html
#
# DX_CLEAN_HTML = @DX_DOCDIR@/html
#
# endif DX_COND_html
#
# ## ------------------------------ ##
# ## Rules specific for CHM output. ##
# ## ------------------------------ ##
#
# if DX_COND_chm
#
# DX_CLEAN_CHM = @DX_DOCDIR@/chm
#
# if DX_COND_chi
#
# DX_CLEAN_CHI = @DX_DOCDIR@/@PACKAGE@.chi
#
# endif DX_COND_chi
#
# endif DX_COND_chm
#
# ## ------------------------------ ##
# ## Rules specific for MAN output. ##
# ## ------------------------------ ##
#
# if DX_COND_man
#
# DX_CLEAN_MAN = @DX_DOCDIR@/man
#
# endif DX_COND_man
#
# ## ------------------------------ ##