Skylake performance results.

[openssl.git] / crypto / sha / asm / sha1-x86_64.pl

diff --git a/crypto/sha/asm/sha1-x86_64.pl b/crypto/sha/asm/sha1-x86_64.pl
index 9aa128ed3aef4218171952a895cd69bffd9eb2b5..d6b0722f31af0ddc16ab2758cb88f36da412d118 100755 (executable)
--- a/crypto/sha/asm/sha1-x86_64.pl
+++ b/crypto/sha/asm/sha1-x86_64.pl
@@ -73,6 +73,7 @@
 # Sandy Bridge 7.70            6.10/+26%       4.99/+54%
 # Ivy Bridge   6.06            4.67/+30%       4.60/+32%
 # Haswell      5.45            4.15/+31%       3.57/+53%
+# Skylake      5.18            4.06/+28%       3.54/+46%
 # Bulldozer    9.11            5.95/+53%
 # VIA Nano     9.32            7.15/+30%
 # Atom         10.3            9.17/+12%
@@ -107,6 +108,10 @@ if (!$avx && $win64 && ($flavour =~ /masm/ || $ENV{ASM} =~ /ml64/) &&
        $avx = ($1>=10) + ($1>=11);
 }
 
+if (!$avx && `$ENV{CC} -v 2>&1` =~ /(^clang version|based on LLVM) ([2-9]\.[0-9]+)/) {
+       $avx = ($2>=3.0) + ($2>3.0);
+}
+
 $shaext=1;     ### set to zero if compiling for 1.0.1
 $avx=1         if (!$shaext && $avx);
 
@@ -1831,7 +1836,9 @@ se_handler:
 
        jmp     .Lcommon_seh_tail
 .size  se_handler,.-se_handler
+___
 
+$code.=<<___ if ($shaext);
 .type  shaext_handler,\@abi-omnipotent
 .align 16
 shaext_handler:
@@ -1864,7 +1871,9 @@ shaext_handler:
 
        jmp     .Lcommon_seh_tail
 .size  shaext_handler,.-shaext_handler
+___
 
+$code.=<<___;
 .type  ssse3_handler,\@abi-omnipotent
 .align 16
 ssse3_handler:
@@ -1987,9 +1996,13 @@ $code.=<<___;
 .LSEH_info_sha1_block_data_order:
        .byte   9,0,0,0
        .rva    se_handler
+___
+$code.=<<___ if ($shaext);
 .LSEH_info_sha1_block_data_order_shaext:
        .byte   9,0,0,0
        .rva    shaext_handler
+___
+$code.=<<___;
 .LSEH_info_sha1_block_data_order_ssse3:
        .byte   9,0,0,0
        .rva    ssse3_handler