提交 32213d8d 编写于 作者: A Andy Polyakov

sha[256|512]-586.pl: add more SIMD code paths.

上级 b4275915
......@@ -28,18 +28,31 @@
# May version, >60% over original. Add AVX+shrd code path, >25%
# improvement on Sandy Bridge over May version, 60% over original.
#
# May 2013.
#
# Replace AMD XOP code path with SSSE3 to cover more processors.
# (Biggest improvement coefficient is on upcoming Atom Silvermont,
# not shown.) Add AVX+BMI code path.
#
# Performance in clock cycles per processed byte (less is better):
#
# PIII P4 AMD K8 Core2 SB Atom Bldzr
# gcc 36 41 27 26 25 50 36
# icc 33 38 25 23 - - -
# x86 asm(*) 27/24 28 19/15.5 18/15.6 12.3 30/25 16.6
# x86_64 asm(**) 17.5 15.1 13.9 11.6 22 13.7
# gcc icc x86 asm(*) SIMD x86_64 asm(**)
# Pentium 46 57 40/38 - -
# PIII 36 33 27/24 - -
# P4 41 38 28 - 17.3
# AMD K8 27 25 19/15.5 - 14.9
# Core2 26 23 18/15.6 14.3 13.8
# Westmere 27 - 19/15.7 13.4 12.3
# Sandy Bridge 25 - 15.9 12.4 11.6
# Ivy Bridge 24 - 15.0 11.4 10.3
# Haswell 22 - 13.9 9.46 7.80
# Bulldozer 36 - 27/22 17.0 13.6
# VIA Nano 36 - 25/22 16.8 16.5
# Atom 50 - 30/25 21.9 18.9
#
# (*) numbers after slash are for unrolled loop, where available,
# otherwise best applicable such as AVX/XOP;
# (*) numbers after slash are for unrolled loop, where applicable;
# (**) x86_64 assembly performance is presented for reference
# purposes.
# purposes, results are best-available;
$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
push(@INC,"${dir}","${dir}../../perlasm");
......@@ -47,21 +60,23 @@ require "x86asm.pl";
&asm_init($ARGV[0],"sha512-586.pl",$ARGV[$#ARGV] eq "386");
$xmm=$ymm=0;
$xmm=$avx=0;
for (@ARGV) { $xmm=1 if (/-DOPENSSL_IA32_SSE2/); }
$ymm=1 if ($xmm &&
`$ENV{CC} -Wa,-v -c -o /dev/null -x assembler /dev/null 2>&1`
=~ /GNU assembler version ([2-9]\.[0-9]+)/ &&
$1>=2.19); # first version supporting AVX
if ($xmm && `$ENV{CC} -Wa,-v -c -o /dev/null -x assembler /dev/null 2>&1`
=~ /GNU assembler version ([2-9]\.[0-9]+)/) {
$avx = ($1>=2.19) + ($1>=2.22);
}
$ymm=1 if ($xmm && !$ymm && $ARGV[0] eq "win32n" &&
`nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/ &&
$1>=2.03); # first version supporting AVX
if ($xmm && !$avx && $ARGV[0] eq "win32n" &&
`nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/) {
$avx = ($1>=2.03) + ($1>=2.10);
}
$ymm=1 if ($xmm && !$ymm && $ARGV[0] eq "win32" &&
`ml 2>&1` =~ /Version ([0-9]+)\./ &&
$1>=10); # first version supporting AVX
if ($xmm && !$avx && $ARGV[0] eq "win32" &&
`ml 2>&1` =~ /Version ([0-9]+)\./) {
$avx = ($1>=10) + ($1>=11);
}
$unroll_after = 64*4; # If pre-evicted from L1P cache first spin of
# fully unrolled loop was measured to run about
......@@ -178,17 +193,21 @@ sub BODY_00_15() {
if (!$i386) {
&picmeup("edx","OPENSSL_ia32cap_P",$K256,&label("K256"));
&mov ("ecx",&DWP(0,"edx"));
&mov ("edx",&DWP(4,"edx"));
&mov ("ebx",&DWP(4,"edx"));
&test ("ecx",1<<20); # check for P4
&jnz (&label("loop"));
&test ("edx",1<<11); # check for XOP
&jnz (&label("XOP")) if ($ymm);
&and ("ecx",1<<30); # mask "Intel CPU" bit
&and ("edx",1<<28); # mask AVX bit
&or ("ecx","edx");
&and ("ebx",1<<28|1<<9); # mask AVX and SSSE3 bits
&or ("ecx","ebx");
&and ("ecx",1<<28|1<<30);
&cmp ("ecx",1<<28|1<<30);
&je (&label("AVX")) if ($ymm);
&je (&label("loop_shrd")) if (!$ymm);
if ($xmm) {
&je (&label("AVX")) if ($avx);
&test ("ebx",1<<9); # check for SSSE3
&jnz (&label("SSSE3"));
} else {
&je (&label("loop_shrd"));
}
if ($unroll_after) {
&sub ("eax","edi");
&cmp ("eax",$unroll_after);
......@@ -199,7 +218,7 @@ sub BODY_00_15() {
sub COMPACT_LOOP() {
my $suffix=shift;
&set_label("loop$suffix",16);
&set_label("loop$suffix",$suffix?32:16);
# copy input block to stack reversing byte and dword order
for($i=0;$i<4;$i++) {
&mov ("eax",&DWP($i*16+0,"edi"));
......@@ -292,7 +311,7 @@ my $suffix=shift;
&COMPACT_LOOP();
&mov ("esp",&DWP(12,"esp")); # restore sp
&function_end_A();
if (!$i386 && !$ymm) {
if (!$i386 && !$xmm) {
# ~20% improvement on Sandy Bridge
local *ror = sub { &shrd(@_[0],@_) };
&COMPACT_LOOP("_shrd");
......@@ -318,7 +337,11 @@ my $suffix=shift;
0x748f82ee,0x78a5636f,0x84c87814,0x8cc70208,
0x90befffa,0xa4506ceb,0xbef9a3f7,0xc67178f2 );
&data_word(@K256);
&data_word(0x00010203,0x04050607,0x08090a0b,0x0c0d0e0f);
&data_word(0x00010203,0x04050607,0x08090a0b,0x0c0d0e0f); # byte swap mask
&asciz("SHA256 block transform for x86, CRYPTOGAMS by <appro\@openssl.org>");
($a,$b,$c,$d,$e,$f,$g,$h)=(0..7); # offsets
sub off { &DWP(4*(((shift)-$i)&7),"esp"); }
if (!$i386 && $unroll_after) {
my @AH=($A,$K256);
......@@ -365,8 +388,6 @@ my @AH=($A,$K256);
&mov (&DWP(32+12*$i,"esp"),"ebx");
my ($t1,$t2) = ("ecx","esi");
my ($a,$b,$c,$d,$e,$f,$g,$h)=(0..7); # offsets
sub off { &DWP(4*(((shift)-$i)&7),"esp"); }
for ($i=0;$i<64;$i++) {
......@@ -472,15 +493,14 @@ my @AH=($A,$K256);
&mov ("esp",&DWP(96+12,"esp")); # restore sp
&function_end_A();
if ($ymm) {{{
}
if (!$i386 && $xmm) {{{
my @X = map("xmm$_",(0..3));
my ($t0,$t1,$t2,$t3) = map("xmm$_",(4..7));
my @AH = ($A,$T);
&set_label("XOP",16);
&set_label("SSSE3",32);
&lea ("esp",&DWP(-96,"esp"));
&vzeroall ();
# copy ctx->h[0-7] to A,B,C,D,E,F,G,H on stack
&mov ($AH[0],&DWP(0,"esi"));
&mov ($AH[1],&DWP(4,"esi"));
......@@ -500,135 +520,208 @@ my @AH = ($A,$T);
&mov ("edi",&DWP(96+4,"esp")); # inp
&mov (&DWP(24,"esp"),"ecx");
&mov (&DWP(28,"esp"),"esi");
&vmovdqa ($t3,&QWP(256,$K256));
&jmp (&label("grand_xop"));
&movdqa ($t3,&QWP(256,$K256));
&jmp (&label("grand_ssse3"));
&set_label("grand_xop",16);
&set_label("grand_ssse3",16);
# load input, reverse byte order, add K256[0..15], save to stack
&vmovdqu (@X[0],&QWP(0,"edi"));
&vmovdqu (@X[1],&QWP(16,"edi"));
&vmovdqu (@X[2],&QWP(32,"edi"));
&vmovdqu (@X[3],&QWP(48,"edi"));
&add ("edi",64);
&vpshufb (@X[0],@X[0],$t3);
&mov (&DWP(96+4,"esp"),"edi");
&vpshufb (@X[1],@X[1],$t3);
&vpshufb (@X[2],@X[2],$t3);
&vpaddd ($t0,@X[0],&QWP(0,$K256));
&vpshufb (@X[3],@X[3],$t3);
&vpaddd ($t1,@X[1],&QWP(16,$K256));
&vpaddd ($t2,@X[2],&QWP(32,$K256));
&vpaddd ($t3,@X[3],&QWP(48,$K256));
&vmovdqa (&QWP(32+0,"esp"),$t0);
&vmovdqa (&QWP(32+16,"esp"),$t1);
&vmovdqa (&QWP(32+32,"esp"),$t2);
&vmovdqa (&QWP(32+48,"esp"),$t3);
&jmp (&label("xop_00_47"));
&set_label("xop_00_47",16);
&movdqu (@X[0],&QWP(0,"edi"));
&movdqu (@X[1],&QWP(16,"edi"));
&movdqu (@X[2],&QWP(32,"edi"));
&movdqu (@X[3],&QWP(48,"edi"));
&add ("edi",64);
&pshufb (@X[0],$t3);
&mov (&DWP(96+4,"esp"),"edi");
&pshufb (@X[1],$t3);
&movdqa ($t0,&QWP(0,$K256));
&pshufb (@X[2],$t3);
&movdqa ($t1,&QWP(16,$K256));
&paddd ($t0,@X[0]);
&pshufb (@X[3],$t3);
&movdqa ($t2,&QWP(32,$K256));
&paddd ($t1,@X[1]);
&movdqa ($t3,&QWP(48,$K256));
&movdqa (&QWP(32+0,"esp"),$t0);
&paddd ($t2,@X[2]);
&movdqa (&QWP(32+16,"esp"),$t1);
&paddd ($t3,@X[3]);
&movdqa (&QWP(32+32,"esp"),$t2);
&movdqa (&QWP(32+48,"esp"),$t3);
&jmp (&label("ssse3_00_47"));
&set_label("ssse3_00_47",16);
&add ($K256,64);
sub XOP_00_47 () {
sub SSSE3_00_47 () {
my $j = shift;
my $body = shift;
my @X = @_;
my @insns = (&$body,&$body,&$body,&$body); # 120 instructions
&vpalignr ($t0,@X[1],@X[0],4); # X[1..4]
eval(shift(@insns));
&movdqa ($t0,@X[1]);
eval(shift(@insns)); # @
eval(shift(@insns));
&vpalignr ($t3,@X[3],@X[2],4); # X[9..12]
&movdqa ($t3,@X[3]);
eval(shift(@insns));
eval(shift(@insns));
&vprotd ($t1,$t0,14);
&palignr ($t0,@X[0],4); # X[1..4]
eval(shift(@insns));
eval(shift(@insns)); # @
eval(shift(@insns));
&vpsrld ($t0,$t0,3);
&vpaddd (@X[0],@X[0],$t3); # X[0..3] += X[9..12]
&palignr ($t3,@X[2],4); # X[9..12]
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns));
&movdqa ($t1,$t0);
eval(shift(@insns)); # @
eval(shift(@insns));
&vprotd ($t2,$t1,25-14);
&vpxor ($t0,$t0,$t1);
&movdqa ($t2,$t0);
eval(shift(@insns));
eval(shift(@insns));
&psrld ($t0,3);
eval(shift(@insns));
eval(shift(@insns)); # @
&paddd (@X[0],$t3); # X[0..3] += X[9..12]
eval(shift(@insns));
&vprotd ($t3,@X[3],13);
&vpxor ($t0,$t0,$t2); # sigma0(X[1..4])
eval(shift(@insns));
&psrld ($t2,7);
eval(shift(@insns));
&vpsrld ($t2,@X[3],10);
eval(shift(@insns));
eval(shift(@insns)); # @
eval(shift(@insns));
&vpaddd (@X[0],@X[0],$t0); # X[0..3] += sigma0(X[1..4])
&pshufd ($t3,@X[3],0b11111010); # X[14..15]
eval(shift(@insns));
eval(shift(@insns));
&vprotd ($t1,$t3,15-13);
&vpxor ($t3,$t3,$t2);
&pslld ($t1,32-18);
eval(shift(@insns));
eval(shift(@insns)); # @
&pxor ($t0,$t2);
eval(shift(@insns));
eval(shift(@insns));
&psrld ($t2,18-7);
eval(shift(@insns));
&vpxor ($t3,$t3,$t1); # sigma1(X[14..15])
eval(shift(@insns));
eval(shift(@insns)); # @
&pxor ($t0,$t1);
eval(shift(@insns));
eval(shift(@insns));
&pslld ($t1,18-7);
eval(shift(@insns));
&vpsrldq ($t3,$t3,8);
eval(shift(@insns));
eval(shift(@insns)); # @
&pxor ($t0,$t2);
eval(shift(@insns));
eval(shift(@insns));
&movdqa ($t2,$t3);
eval(shift(@insns));
&vpaddd (@X[0],@X[0],$t3); # X[0..1] += sigma1(X[14..15])
eval(shift(@insns));
eval(shift(@insns)); # @
&pxor ($t0,$t1); # sigma0(X[1..4])
eval(shift(@insns));
eval(shift(@insns));
&psrld ($t3,10);
eval(shift(@insns));
&vprotd ($t3,@X[0],13);
eval(shift(@insns));
eval(shift(@insns)); # @
&paddd (@X[0],$t0); # X[0..3] += sigma0(X[1..4])
eval(shift(@insns));
&vpsrld ($t2,@X[0],10);
eval(shift(@insns));
&psrlq ($t2,17);
eval(shift(@insns));
&vprotd ($t1,$t3,15-13);
eval(shift(@insns));
eval(shift(@insns)); # @
&pxor ($t3,$t2);
eval(shift(@insns));
&vpxor ($t3,$t3,$t2);
eval(shift(@insns));
&psrlq ($t2,19-17);
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns)); # @
&pxor ($t3,$t2);
eval(shift(@insns));
&vpxor ($t3,$t3,$t1); # sigma1(X[16..17])
eval(shift(@insns));
&pshufd ($t3,$t3,0b10000000);
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns)); # @
eval(shift(@insns));
&vpslldq ($t3,$t3,8); # 22 instructions
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns)); # @
eval(shift(@insns));
&vpaddd (@X[0],@X[0],$t3); # X[2..3] += sigma1(X[16..17])
&psrldq ($t3,8);
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns));
&paddd (@X[0],$t3); # X[0..1] += sigma1(X[14..15])
eval(shift(@insns)); # @
eval(shift(@insns));
&vpaddd ($t2,@X[0],&QWP(16*$j,$K256));
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns)); # @
eval(shift(@insns));
&pshufd ($t3,@X[0],0b01010000); # X[16..17]
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns));
&movdqa ($t2,$t3);
eval(shift(@insns)); # @
&psrld ($t3,10);
eval(shift(@insns));
&psrlq ($t2,17);
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns)); # @
&pxor ($t3,$t2);
eval(shift(@insns));
eval(shift(@insns));
&psrlq ($t2,19-17);
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns)); # @
&pxor ($t3,$t2);
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns));
&pshufd ($t3,$t3,0b00001000);
eval(shift(@insns));
eval(shift(@insns)); # @
&movdqa ($t2,&QWP(16*$j,$K256));
eval(shift(@insns));
eval(shift(@insns));
&pslldq ($t3,8);
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns)); # @
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns)); # @
&paddd (@X[0],$t3); # X[2..3] += sigma1(X[16..17])
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns));
&paddd ($t2,@X[0]);
eval(shift(@insns)); # @
foreach (@insns) { eval; } # remaining instructions
&vmovdqa (&QWP(32+16*$j,"esp"),$t2);
&movdqa (&QWP(32+16*$j,"esp"),$t2);
}
sub body_00_15 () {
(
'&mov ("ecx",$E);',
'&mov ("esi",&off($f));',
'&ror ($E,25-11);',
'&mov ("edi",&off($g));',
'&mov ("esi",&off($f));',
'&xor ($E,"ecx");',
'&mov ("edi",&off($g));',
'&xor ("esi","edi");',
'&ror ($E,11-6);',
'&and ("esi","ecx");',
......@@ -637,19 +730,19 @@ sub body_00_15 () {
'&xor ("edi","esi");', # Ch(e,f,g)
'&ror ($E,6);', # T = Sigma1(e)
'&mov ("ecx",$AH[0]);',
'&add ($E,"edi");', # T += Ch(e,f,g)
'&mov ("edi",&off($b));',
'&mov ("esi",$AH[0]);',
'&add ($E,&off($h));', # T += h
'&ror ("ecx",22-13);',
'&add ($E,"edi");', # T += Ch(e,f,g)
'&mov ("edi",&off($b));',
'&xor ("ecx",$AH[0]);',
'&mov (&off($a),$AH[0]);', # save $A, modulo-scheduled
'&xor ("ecx",$AH[0]);',
'&xor ($AH[0],"edi");', # a ^= b, (b^c) in next round
'&add ($E,&off($h));', # T += h
'&ror ("ecx",13-2);',
'&and ($AH[1],$AH[0]);', # (b^c) &= (a^b)
'&add ($E,&DWP(32+4*($i&15),"esp"));', # T += K[i]+X[i]
'&xor ("ecx","esi");',
'&add ($E,&DWP(32+4*($i&15),"esp"));', # T += K[i]+X[i]
'&xor ($AH[1],"edi");', # h = Maj(a,b,c) = Ch(a^b,c,b)
'&ror ("ecx",2);', # Sigma0(a)
......@@ -662,11 +755,11 @@ sub body_00_15 () {
}
for ($i=0,$j=0; $j<4; $j++) {
&XOP_00_47($j,\&body_00_15,@X);
&SSSE3_00_47($j,\&body_00_15,@X);
push(@X,shift(@X)); # rotate(@X)
}
&cmp (&DWP(16*$j,$K256),0x00010203);
&jne (&label("xop_00_47"));
&jne (&label("ssse3_00_47"));
for ($i=0; $i<16; ) {
foreach(body_00_15()) { eval; }
......@@ -708,16 +801,21 @@ sub body_00_15 () {
&mov (&DWP(28,"esp"),"edi");
&mov ("edi",&DWP(96+4,"esp")); # inp
&vmovdqa ($t3,&QWP(64,$K256));
&movdqa ($t3,&QWP(64,$K256));
&sub ($K256,3*64); # rewind K
&cmp ("edi",&DWP(96+8,"esp")); # are we done yet?
&jb (&label("grand_xop"));
&jb (&label("grand_ssse3"));
&mov ("esp",&DWP(96+12,"esp")); # restore sp
&vzeroall ();
&function_end_A();
&set_label("AVX",16);
if ($avx) {
&set_label("AVX",32);
if ($avx>1) {
&mov ("edx",&DWP(8,"edx"));
&and ("edx",1<<8|1<<3); # check for BMI2+BMI1
&cmp ("edx",1<<8|1<<3);
&je (&label("AVX_BMI"));
}
&lea ("esp",&DWP(-96,"esp"));
&vzeroall ();
# copy ctx->h[0-7] to A,B,C,D,E,F,G,H on stack
......@@ -742,7 +840,7 @@ sub body_00_15 () {
&vmovdqa ($t3,&QWP(256,$K256));
&jmp (&label("grand_avx"));
&set_label("grand_avx",16);
&set_label("grand_avx",32);
# load input, reverse byte order, add K256[0..15], save to stack
&vmovdqu (@X[0],&QWP(0,"edi"));
&vmovdqu (@X[1],&QWP(16,"edi"));
......@@ -809,12 +907,14 @@ my $j = shift;
my $body = shift;
my @X = @_;
my @insns = (&$body,&$body,&$body,&$body); # 120 instructions
my $insn;
foreach (Xupdate_AVX()) { # 31 instructions
eval;
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns));
eval($insn = shift(@insns));
eval(shift(@insns)) if ($insn =~ /rorx/ && @insns[0] =~ /rorx/);
}
&vpaddd ($t2,@X[0],&QWP(16*$j,$K256));
foreach (@insns) { eval; } # remaining instructions
......@@ -876,9 +976,150 @@ my @insns = (&$body,&$body,&$body,&$body); # 120 instructions
&mov ("esp",&DWP(96+12,"esp")); # restore sp
&vzeroall ();
&function_end_A();
}}}
if ($avx>1) {
sub bodyx_00_15 () { # +10%
(
'&rorx ("ecx",$E,6)',
'&rorx ("esi",$E,11)',
'&mov (&off($e),$E)', # save $E, modulo-scheduled
'&rorx ("edi",$E,25)',
'&xor ("ecx","esi")',
'&andn ("esi",$E,&off($g))',
'&xor ("ecx","edi")', # Sigma1(e)
'&and ($E,&off($f))',
'&mov (&off($a),$AH[0]);', # save $A, modulo-scheduled
'&or ($E,"esi")', # T = Ch(e,f,g)
'&rorx ("edi",$AH[0],2)',
'&rorx ("esi",$AH[0],13)',
'&lea ($E,&DWP(0,$E,"ecx"))', # T += Sigma1(e)
'&rorx ("ecx",$AH[0],22)',
'&xor ("esi","edi")',
'&mov ("edi",&off($b))',
'&xor ("ecx","esi")', # Sigma0(a)
'&xor ($AH[0],"edi")', # a ^= b, (b^c) in next round
'&add ($E,&off($h))', # T += h
'&and ($AH[1],$AH[0])', # (b^c) &= (a^b)
'&add ($E,&DWP(32+4*($i&15),"esp"))', # T += K[i]+X[i]
'&xor ($AH[1],"edi")', # h = Maj(a,b,c) = Ch(a^b,c,b)
'&add ("ecx",$E)', # h += T
'&add ($E,&off($d))', # d += T
'&lea ($AH[1],&DWP(0,$AH[1],"ecx"));'. # h += Sigma0(a)
'@AH = reverse(@AH); $i++;' # rotate(a,h)
);
}
&set_label("AVX_BMI",32);
&lea ("esp",&DWP(-96,"esp"));
&vzeroall ();
# copy ctx->h[0-7] to A,B,C,D,E,F,G,H on stack
&mov ($AH[0],&DWP(0,"esi"));
&mov ($AH[1],&DWP(4,"esi"));
&mov ("ecx",&DWP(8,"esi"));
&mov ("edi",&DWP(12,"esi"));
#&mov (&DWP(0,"esp"),$AH[0]);
&mov (&DWP(4,"esp"),$AH[1]);
&xor ($AH[1],"ecx"); # magic
&mov (&DWP(8,"esp"),"ecx");
&mov (&DWP(12,"esp"),"edi");
&mov ($E,&DWP(16,"esi"));
&mov ("edi",&DWP(20,"esi"));
&mov ("ecx",&DWP(24,"esi"));
&mov ("esi",&DWP(28,"esi"));
#&mov (&DWP(16,"esp"),$E);
&mov (&DWP(20,"esp"),"edi");
&mov ("edi",&DWP(96+4,"esp")); # inp
&mov (&DWP(24,"esp"),"ecx");
&mov (&DWP(28,"esp"),"esi");
&vmovdqa ($t3,&QWP(256,$K256));
&jmp (&label("grand_avx_bmi"));
&set_label("grand_avx_bmi",32);
# load input, reverse byte order, add K256[0..15], save to stack
&vmovdqu (@X[0],&QWP(0,"edi"));
&vmovdqu (@X[1],&QWP(16,"edi"));
&vmovdqu (@X[2],&QWP(32,"edi"));
&vmovdqu (@X[3],&QWP(48,"edi"));
&add ("edi",64);
&vpshufb (@X[0],@X[0],$t3);
&mov (&DWP(96+4,"esp"),"edi");
&vpshufb (@X[1],@X[1],$t3);
&vpshufb (@X[2],@X[2],$t3);
&vpaddd ($t0,@X[0],&QWP(0,$K256));
&vpshufb (@X[3],@X[3],$t3);
&vpaddd ($t1,@X[1],&QWP(16,$K256));
&vpaddd ($t2,@X[2],&QWP(32,$K256));
&vpaddd ($t3,@X[3],&QWP(48,$K256));
&vmovdqa (&QWP(32+0,"esp"),$t0);
&vmovdqa (&QWP(32+16,"esp"),$t1);
&vmovdqa (&QWP(32+32,"esp"),$t2);
&vmovdqa (&QWP(32+48,"esp"),$t3);
&jmp (&label("avx_bmi_00_47"));
&set_label("avx_bmi_00_47",16);
&add ($K256,64);
for ($i=0,$j=0; $j<4; $j++) {
&AVX_00_47($j,\&bodyx_00_15,@X);
push(@X,shift(@X)); # rotate(@X)
}
&cmp (&DWP(16*$j,$K256),0x00010203);
&jne (&label("avx_bmi_00_47"));
for ($i=0; $i<16; ) {
foreach(bodyx_00_15()) { eval; }
}
&mov ("esi",&DWP(96,"esp")); #ctx
#&mov ($AH[0],&DWP(0,"esp"));
&xor ($AH[1],"edi"); #&mov ($AH[1],&DWP(4,"esp"));
#&mov ("edi", &DWP(8,"esp"));
&mov ("ecx",&DWP(12,"esp"));
&add ($AH[0],&DWP(0,"esi"));
&add ($AH[1],&DWP(4,"esi"));
&add ("edi",&DWP(8,"esi"));
&add ("ecx",&DWP(12,"esi"));
&mov (&DWP(0,"esi"),$AH[0]);
&mov (&DWP(4,"esi"),$AH[1]);
&mov (&DWP(8,"esi"),"edi");
&mov (&DWP(12,"esi"),"ecx");
#&mov (&DWP(0,"esp"),$AH[0]);
&mov (&DWP(4,"esp"),$AH[1]);
&xor ($AH[1],"edi"); # magic
&mov (&DWP(8,"esp"),"edi");
&mov (&DWP(12,"esp"),"ecx");
#&mov ($E,&DWP(16,"esp"));
&mov ("edi",&DWP(20,"esp"));
&mov ("ecx",&DWP(24,"esp"));
&add ($E,&DWP(16,"esi"));
&add ("edi",&DWP(20,"esi"));
&add ("ecx",&DWP(24,"esi"));
&mov (&DWP(16,"esi"),$E);
&mov (&DWP(20,"esi"),"edi");
&mov (&DWP(20,"esp"),"edi");
&mov ("edi",&DWP(28,"esp"));
&mov (&DWP(24,"esi"),"ecx");
#&mov (&DWP(16,"esp"),$E);
&add ("edi",&DWP(28,"esi"));
&mov (&DWP(24,"esp"),"ecx");
&mov (&DWP(28,"esi"),"edi");
&mov (&DWP(28,"esp"),"edi");
&mov ("edi",&DWP(96+4,"esp")); # inp
&vmovdqa ($t3,&QWP(64,$K256));
&sub ($K256,3*64); # rewind K
&cmp ("edi",&DWP(96+8,"esp")); # are we done yet?
&jb (&label("grand_avx_bmi"));
&mov ("esp",&DWP(96+12,"esp")); # restore sp
&vzeroall ();
&function_end_A();
}
}
}}}
&function_end_B("sha256_block_data_order");
&asciz("SHA256 block transform for x86, CRYPTOGAMS by <appro\@openssl.org>");
&asm_finish();
......@@ -9,18 +9,30 @@
#
# SHA512 block transform for x86. September 2007.
#
# May 2013.
#
# Add SSSE3 code path, 20-25% improvement [over original SSE2 code].
#
# Performance in clock cycles per processed byte (less is better):
#
# PIII P4 AMD K8 Core2 SB Atom Bldzr
# gcc 75 116 54 66 58 126 121
# icc 77 95 55 57 - - -
# x86 asm 56 82 36 40 35 68 50
# SSE2 asm - 36.2 20.8 19.2 14.9 60(**) 17.1
# x86_64 asm(*) - 33 9.6 10.3 11.3 14.7 13.5
# gcc icc x86 asm SIMD(*) x86_64(**)
# Pentium 100 97 61 - -
# PIII 75 77 56 - -
# P4 116 95 82 34.6 30.8
# AMD K8 54 55 36 20.7 9.57
# Core2 66 57 40 15.9 9.97
# Westmere 70 - 38 12.2 9.58
# Sandy Bridge 58 - 35 11.9 11.2
# Ivy Bridge 50 - 33 11.5 8.17
# Haswell 46 - 29 11.3 7.66
# Bulldozer 121 - 50 14.0 13.5
# VIA Nano 91 - 52 33 14.7
# Atom 126 - 68 48(***) 14.7
#
# (*) x86_64 assembler performance is presented for reference
# purposes.
# (**) paddq is increadibly slow on Atom.
# (*) whichever best applicable.
# (**) x86_64 assembler performance is presented for reference
# purposes, the results are for integer-only code.
# (***) paddq is increadibly slow on Atom.
#
# IALU code-path is optimized for elder Pentiums. On vanilla Pentium
# performance improvement over compiler generated code reaches ~60%,
......@@ -82,7 +94,7 @@ sub BODY_00_15_sse2 {
&pand ("mm5",$E); # f&=e
&psllq ($E,23); # $E is sliding left
&movq ($A,"mm3") if ($phase<2);
&movq (&QWP(8*9,"esp"),"mm7") if ($phase>1); # save X[i]
&movq (&QWP(8*9,"esp"),"mm7") # save X[i]
&movq ("mm3","mm1"); # %mm3 is T1
&psrlq ("mm1",4);
&pxor ("mm5","mm6"); # Ch(e,f,g)
......@@ -100,11 +112,11 @@ sub BODY_00_15_sse2 {
&pxor ("mm3",$E); # T1=Sigma1_512(e)
&movq ($E,$Dsse2); # e = load d, e in next round
&movq ("mm5",$A); # %mm5 is sliding right
&paddq ("mm3","mm7"); # T1+=X[i]
&movq ("mm5",$A); # %mm5 is sliding right
&psrlq ("mm5",28);
&movq ("mm6",$A); # %mm6 is sliding left
&paddq ($E,"mm3"); # d += T1
&movq ("mm6",$A); # %mm6 is sliding left
&movq ("mm7","mm5");
&psllq ("mm6",25);
&movq ("mm1",$Bsse2); # load b
......@@ -290,19 +302,28 @@ sub BODY_00_15_x86 {
if ($sse2) {
&picmeup("edx","OPENSSL_ia32cap_P",$K512,&label("K512"));
&bt (&DWP(0,"edx"),26);
&jnc (&label("loop_x86"));
&mov ("ecx",&DWP(0,"edx"));
&test ("ecx",1<<26);
&jz (&label("loop_x86"));
&mov ("edx",&DWP(4,"edx"));
# load ctx->h[0-7]
&movq ($A,&QWP(0,"esi"));
&and ("ecx",1<<24); # XMM registers availability
&movq ("mm1",&QWP(8,"esi"));
&and ("edx",1<<9); # SSSE3 bit
&movq ($BxC,&QWP(16,"esi"));
&or ("ecx","edx");
&movq ("mm3",&QWP(24,"esi"));
&movq ($E,&QWP(32,"esi"));
&movq ("mm5",&QWP(40,"esi"));
&movq ("mm6",&QWP(48,"esi"));
&movq ("mm7",&QWP(56,"esi"));
&cmp ("ecx",1<<24|1<<9);
&je (&label("SSSE3"));
&sub ("esp",8*10);
&jmp (&label("loop_sse2"));
&set_label("loop_sse2",16);
#&movq ($Asse2,$A);
......@@ -316,32 +337,32 @@ if ($sse2) {
&movq ($Hsse2,"mm7");
&movq ("mm3",$A); # magic
&mov ("ecx",&DWP(0,"edi"));
&mov ("edx",&DWP(4,"edi"));
&mov ("eax",&DWP(0,"edi"));
&mov ("ebx",&DWP(4,"edi"));
&add ("edi",8);
&bswap ("ecx");
&bswap ("edx");
&mov (&DWP(8*9+4,"esp"),"ecx");
&mov (&DWP(8*9+0,"esp"),"edx");
&mov ("edx",15); # counter
&bswap ("eax");
&bswap ("ebx");
&jmp (&label("00_14_sse2"));
&set_label("00_14_sse2",16);
&movd ("mm1","eax");
&mov ("eax",&DWP(0,"edi"));
&movd ("mm7","ebx");
&mov ("ebx",&DWP(4,"edi"));
&add ("edi",8);
&bswap ("eax");
&movq ("mm7",&QWP(8*9,"esp")); # X[i]
&bswap ("ebx");
&mov (&DWP(8*8+4,"esp"),"eax");
&mov (&DWP(8*8+0,"esp"),"ebx");
&punpckldq("mm7","mm1");
&BODY_00_15_sse2();
&dec ("edx");
&jnz (&label("00_14_sse2"));
&movq ("mm7",&QWP(8*9,"esp")); # X[i]
&movd ("mm1","eax");
&movd ("mm7","ebx");
&punpckldq("mm7","mm1");
&BODY_00_15_sse2(1);
......@@ -429,6 +450,231 @@ if ($sse2) {
&mov ("esp",&DWP(8*10+12,"esp")); # restore sp
&emms ();
&function_end_A();
&set_label("SSSE3",32);
{ my ($cnt,$frame)=("ecx","edx");
my @X=map("xmm$_",(0..7));
my $j;
my $i=0;
&lea ($frame,&DWP(-64,"esp"));
&sub ("esp",256);
# fixed stack frame layout
#
# +0 A B C D E F G H # backing store
# +64 X[0]+K[i] .. X[15]+K[i] # XMM->MM xfer area
# +192 # XMM off-load ring buffer
# +256 # saved parameters
&movdqa (@X[1],&QWP(80*8,$K512)); # byte swap mask
&movdqu (@X[0],&QWP(0,"edi"));
&pshufb (@X[0],@X[1]);
for ($j=0;$j<8;$j++) {
&movdqa (&QWP(16*(($j-1)%4),$frame),@X[3]) if ($j>4); # off-load
&movdqa (@X[3],&QWP(16*($j%8),$K512));
&movdqa (@X[2],@X[1]) if ($j<7); # perpetuate byte swap mask
&movdqu (@X[1],&QWP(16*($j+1),"edi")) if ($j<7); # next input
&movdqa (@X[1],&QWP(16*(($j+1)%4),$frame)) if ($j==7);# restore @X[0]
&paddq (@X[3],@X[0]);
&pshufb (@X[1],@X[2]) if ($j<7);
&movdqa (&QWP(16*($j%8)-128,$frame),@X[3]); # xfer X[i]+K[i]
push(@X,shift(@X)); # rotate(@X)
}
#&jmp (&label("loop_ssse3"));
&nop ();
&set_label("loop_ssse3",32);
&movdqa (@X[2],&QWP(16*(($j+1)%4),$frame)); # pre-restore @X[1]
&movdqa (&QWP(16*(($j-1)%4),$frame),@X[3]); # off-load @X[3]
&lea ($K512,&DWP(16*8,$K512));
#&movq ($Asse2,$A); # off-load A-H
&movq ($Bsse2,"mm1");
&mov ("ebx","edi");
&movq ($Csse2,$BxC);
&lea ("edi",&DWP(128,"edi")); # advance input
&movq ($Dsse2,"mm3");
&cmp ("edi","eax");
#&movq ($Esse2,$E);
&movq ($Fsse2,"mm5");
&cmovb ("ebx","edi");
&movq ($Gsse2,"mm6");
&mov ("ecx",4); # loop counter
&pxor ($BxC,"mm1"); # magic
&movq ($Hsse2,"mm7");
&pxor ("mm3","mm3"); # magic
&jmp (&label("00_47_ssse3"));
sub BODY_00_15_ssse3 { # "phase-less" copy of BODY_00_15_sse2
(
'&movq ("mm1",$E)', # %mm1 is sliding right
'&movq ("mm7",&QWP(((-8*$i)%128)-128,$frame))',# X[i]+K[i]
'&pxor ("mm5","mm6")', # f^=g
'&psrlq ("mm1",14)',
'&movq (&QWP(8*($i+4)%64,"esp"),$E)', # modulo-scheduled save e
'&pand ("mm5",$E)', # f&=e
'&psllq ($E,23)', # $E is sliding left
'&paddq ($A,"mm3")', # [h+=Maj(a,b,c)]
'&movq ("mm3","mm1")', # %mm3 is T1
'&psrlq("mm1",4)',
'&pxor ("mm5","mm6")', # Ch(e,f,g)
'&pxor ("mm3",$E)',
'&psllq($E,23)',
'&pxor ("mm3","mm1")',
'&movq (&QWP(8*$i%64,"esp"),$A)', # modulo-scheduled save a
'&paddq("mm7","mm5")', # X[i]+=Ch(e,f,g)
'&pxor ("mm3",$E)',
'&psrlq("mm1",23)',
'&paddq("mm7",&QWP(8*($i+7)%64,"esp"))', # X[i]+=h
'&pxor ("mm3","mm1")',
'&psllq($E,4)',
'&pxor ("mm3",$E)', # T1=Sigma1_512(e)
'&movq ($E,&DWP(8*($i+3)%64,"esp"))', # e = load d, e in next round
'&paddq ("mm3","mm7")', # T1+=X[i]
'&movq ("mm5",$A)', # %mm5 is sliding right
'&psrlq("mm5",28)',
'&paddq ($E,"mm3")', # d += T1
'&movq ("mm6",$A)', # %mm6 is sliding left
'&movq ("mm7","mm5")',
'&psllq("mm6",25)',
'&movq ("mm1",&QWP(8*($i+1)%64,"esp"))', # load b
'&psrlq("mm5",6)',
'&pxor ("mm7","mm6")',
'&psllq("mm6",5)',
'&pxor ("mm7","mm5")',
'&pxor ($A,"mm1")', # a^b, b^c in next round
'&psrlq("mm5",5)',
'&pxor ("mm7","mm6")',
'&pand ($BxC,$A)', # (b^c)&(a^b)
'&psllq("mm6",6)',
'&pxor ("mm7","mm5")',
'&pxor ($BxC,"mm1")', # [h=]Maj(a,b,c)
'&pxor ("mm6","mm7")', # Sigma0_512(a)
'&movq ("mm5",&QWP(8*($i+5-1)%64,"esp"))', # pre-load f
'&paddq ($BxC,"mm6")', # h+=Sigma0(a)
'&movq ("mm6",&QWP(8*($i+6-1)%64,"esp"))', # pre-load g
'($A,$BxC) = ($BxC,$A); $i--;'
);
}
&set_label("00_47_ssse3",32);
for(;$j<16;$j++) {
my ($t0,$t2,$t1)=@X[2..4];
my @insns = (&BODY_00_15_ssse3(),&BODY_00_15_ssse3());
&movdqa ($t2,@X[5]);
&movdqa (@X[1],$t0); # restore @X[1]
&palignr ($t0,@X[0],8); # X[1..2]
&movdqa (&QWP(16*($j%4),$frame),@X[4]); # off-load @X[4]
&palignr ($t2,@X[4],8); # X[9..10]
&movdqa ($t1,$t0);
&psrlq ($t0,7);
&paddq (@X[0],$t2); # X[0..1] += X[9..10]
&movdqa ($t2,$t1);
&psrlq ($t1,1);
&psllq ($t2,64-8);
&pxor ($t0,$t1);
&psrlq ($t1,8-1);
&pxor ($t0,$t2);
&psllq ($t2,8-1);
&pxor ($t0,$t1);
&movdqa ($t1,@X[7]);
&pxor ($t0,$t2); # sigma0(X[1..2])
&movdqa ($t2,@X[7]);
&psrlq ($t1,6);
&paddq (@X[0],$t0); # X[0..1] += sigma0(X[1..2])
&movdqa ($t0,@X[7]);
&psrlq ($t2,19);
&psllq ($t0,64-61);
&pxor ($t1,$t2);
&psrlq ($t2,61-19);
&pxor ($t1,$t0);
&psllq ($t0,61-19);
&pxor ($t1,$t2);
&movdqa ($t2,&QWP(16*(($j+2)%4),$frame));# pre-restore @X[1]
&pxor ($t1,$t0); # sigma0(X[1..2])
&movdqa ($t0,&QWP(16*($j%8),$K512));
eval(shift(@insns));
&paddq (@X[0],$t1); # X[0..1] += sigma0(X[14..15])
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns));
eval(shift(@insns));
&paddq ($t0,@X[0]);
foreach(@insns) { eval; }
&movdqa (&QWP(16*($j%8)-128,$frame),$t0);# xfer X[i]+K[i]
push(@X,shift(@X)); # rotate(@X)
}
&lea ($K512,&DWP(16*8,$K512));
&dec ("ecx");
&jnz (&label("00_47_ssse3"));
&movdqa (@X[1],&QWP(0,$K512)); # byte swap mask
&lea ($K512,&DWP(-80*8,$K512)); # rewind
&movdqu (@X[0],&QWP(0,"ebx"));
&pshufb (@X[0],@X[1]);
for ($j=0;$j<8;$j++) { # load next or same block
my @insns = (&BODY_00_15_ssse3(),&BODY_00_15_ssse3());
&movdqa (&QWP(16*(($j-1)%4),$frame),@X[3]) if ($j>4); # off-load
&movdqa (@X[3],&QWP(16*($j%8),$K512));
&movdqa (@X[2],@X[1]) if ($j<7); # perpetuate byte swap mask
&movdqu (@X[1],&QWP(16*($j+1),"ebx")) if ($j<7); # next input
&movdqa (@X[1],&QWP(16*(($j+1)%4),$frame)) if ($j==7);# restore @X[0]
&paddq (@X[3],@X[0]);
&pshufb (@X[1],@X[2]) if ($j<7);
foreach(@insns) { eval; }
&movdqa (&QWP(16*($j%8)-128,$frame),@X[3]);# xfer X[i]+K[i]
push(@X,shift(@X)); # rotate(@X)
}
#&movq ($A,$Asse2); # load A-H
&movq ("mm1",$Bsse2);
&paddq ($A,"mm3"); # from BODY_00_15
#&movq ($BxC,$Csse2);
&movq ("mm3",$Dsse2);
#&movq ($E,$Esse2);
#&movq ("mm5",$Fsse2);
#&movq ("mm6",$Gsse2);
&movq ("mm7",$Hsse2);
&pxor ($BxC,"mm1"); # de-magic
&paddq ($A,&QWP(0,"esi"));
&paddq ("mm1",&QWP(8,"esi"));
&paddq ($BxC,&QWP(16,"esi"));
&paddq ("mm3",&QWP(24,"esi"));
&paddq ($E,&QWP(32,"esi"));
&paddq ("mm5",&QWP(40,"esi"));
&paddq ("mm6",&QWP(48,"esi"));
&paddq ("mm7",&QWP(56,"esi"));
&movq (&QWP(0,"esi"),$A);
&movq (&QWP(8,"esi"),"mm1");
&movq (&QWP(16,"esi"),$BxC);
&movq (&QWP(24,"esi"),"mm3");
&movq (&QWP(32,"esi"),$E);
&movq (&QWP(40,"esi"),"mm5");
&movq (&QWP(48,"esi"),"mm6");
&movq (&QWP(56,"esi"),"mm7");
&cmp ("edi","eax") # are we done yet?
&jb (&label("loop_ssse3"));
&mov ("esp",&DWP(64+12,$frame)); # restore sp
&emms ();
}
&function_end_A();
}
&set_label("loop_x86",16);
# copy input block to stack reversing byte and qword order
......@@ -655,6 +901,9 @@ if ($sse2) {
&data_word(0xfc657e2a,0x597f299c); # u64
&data_word(0x3ad6faec,0x5fcb6fab); # u64
&data_word(0x4a475817,0x6c44198c); # u64
&data_word(0x04050607,0x00010203); # byte swap
&data_word(0x0c0d0e0f,0x08090a0b); # mask
&function_end_B("sha512_block_data_order");
&asciz("SHA512 block transform for x86, CRYPTOGAMS by <appro\@openssl.org>");
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册