|
Packit |
5c3484 |
dnl X86-64 mpn_addmul_1 and mpn_submul_1 optimised for Intel Sandy Bridge.
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
dnl Contributed to the GNU project by Torbjörn Granlund.
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
dnl Copyright 2003-2005, 2007, 2008, 2011-2013 Free Software Foundation, Inc.
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
dnl This file is part of the GNU MP Library.
|
|
Packit |
5c3484 |
dnl
|
|
Packit |
5c3484 |
dnl The GNU MP Library is free software; you can redistribute it and/or modify
|
|
Packit |
5c3484 |
dnl it under the terms of either:
|
|
Packit |
5c3484 |
dnl
|
|
Packit |
5c3484 |
dnl * the GNU Lesser General Public License as published by the Free
|
|
Packit |
5c3484 |
dnl Software Foundation; either version 3 of the License, or (at your
|
|
Packit |
5c3484 |
dnl option) any later version.
|
|
Packit |
5c3484 |
dnl
|
|
Packit |
5c3484 |
dnl or
|
|
Packit |
5c3484 |
dnl
|
|
Packit |
5c3484 |
dnl * the GNU General Public License as published by the Free Software
|
|
Packit |
5c3484 |
dnl Foundation; either version 2 of the License, or (at your option) any
|
|
Packit |
5c3484 |
dnl later version.
|
|
Packit |
5c3484 |
dnl
|
|
Packit |
5c3484 |
dnl or both in parallel, as here.
|
|
Packit |
5c3484 |
dnl
|
|
Packit |
5c3484 |
dnl The GNU MP Library is distributed in the hope that it will be useful, but
|
|
Packit |
5c3484 |
dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
|
|
Packit |
5c3484 |
dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
|
|
Packit |
5c3484 |
dnl for more details.
|
|
Packit |
5c3484 |
dnl
|
|
Packit |
5c3484 |
dnl You should have received copies of the GNU General Public License and the
|
|
Packit |
5c3484 |
dnl GNU Lesser General Public License along with the GNU MP Library. If not,
|
|
Packit |
5c3484 |
dnl see https://www.gnu.org/licenses/.
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
include(`../config.m4')
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
C cycles/limb
|
|
Packit |
5c3484 |
C AMD K8,K9
|
|
Packit |
5c3484 |
C AMD K10
|
|
Packit |
5c3484 |
C AMD bull
|
|
Packit |
5c3484 |
C AMD pile
|
|
Packit |
5c3484 |
C AMD steam
|
|
Packit |
5c3484 |
C AMD bobcat
|
|
Packit |
5c3484 |
C AMD jaguar
|
|
Packit |
5c3484 |
C Intel P4
|
|
Packit |
5c3484 |
C Intel core
|
|
Packit |
5c3484 |
C Intel NHM
|
|
Packit |
5c3484 |
C Intel SBR 3.24 (average, fluctuating in 3.20-3.57)
|
|
Packit |
5c3484 |
C Intel IBR 3.04
|
|
Packit |
5c3484 |
C Intel HWL
|
|
Packit |
5c3484 |
C Intel BWL
|
|
Packit |
5c3484 |
C Intel atom
|
|
Packit |
5c3484 |
C VIA nano
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
C The loop of this code is the result of running a code generation and
|
|
Packit |
5c3484 |
C optimization tool suite written by David Harvey and Torbjörn Granlund.
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
define(`rp', `%rdi') C rcx
|
|
Packit |
5c3484 |
define(`up', `%rsi') C rdx
|
|
Packit |
5c3484 |
define(`n_param', `%rdx') C r8
|
|
Packit |
5c3484 |
define(`v0', `%rcx') C r9
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
define(`n', `%rbx')
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
define(`I',`$1')
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
ifdef(`OPERATION_addmul_1',`
|
|
Packit |
5c3484 |
define(`ADDSUB', `add')
|
|
Packit |
5c3484 |
define(`func', `mpn_addmul_1')
|
|
Packit |
5c3484 |
')
|
|
Packit |
5c3484 |
ifdef(`OPERATION_submul_1',`
|
|
Packit |
5c3484 |
define(`ADDSUB', `sub')
|
|
Packit |
5c3484 |
define(`func', `mpn_submul_1')
|
|
Packit |
5c3484 |
')
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
ABI_SUPPORT(DOS64)
|
|
Packit |
5c3484 |
ABI_SUPPORT(STD64)
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
MULFUNC_PROLOGUE(mpn_addmul_1 mpn_submul_1)
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
IFDOS(` define(`up', ``%rsi'')') dnl
|
|
Packit |
5c3484 |
IFDOS(` define(`rp', ``%rcx'')') dnl
|
|
Packit |
5c3484 |
IFDOS(` define(`v0', ``%r9'')') dnl
|
|
Packit |
5c3484 |
IFDOS(` define(`r9', ``rdi'')') dnl
|
|
Packit |
5c3484 |
IFDOS(` define(`n_param',``%r8'')') dnl
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
ASM_START()
|
|
Packit |
5c3484 |
TEXT
|
|
Packit |
5c3484 |
ALIGN(32)
|
|
Packit |
5c3484 |
PROLOGUE(func)
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
IFDOS(``push %rsi '')
|
|
Packit |
5c3484 |
IFDOS(``push %rdi '')
|
|
Packit |
5c3484 |
IFDOS(``mov %rdx, %rsi '')
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
mov (up), %rax
|
|
Packit |
5c3484 |
push %rbx
|
|
Packit |
5c3484 |
lea (up,n_param,8), up
|
|
Packit |
5c3484 |
lea (rp,n_param,8), rp
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
test $1, R8(n_param)
|
|
Packit |
5c3484 |
jnz L(b13)
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
L(b02): xor R32(%r11), R32(%r11)
|
|
Packit |
5c3484 |
test $2, R8(n_param)
|
|
Packit |
5c3484 |
jnz L(b2)
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
L(b0): mov $1, R32(n)
|
|
Packit |
5c3484 |
sub n_param, n
|
|
Packit |
5c3484 |
mul v0
|
|
Packit |
5c3484 |
mov %rdx, %r9
|
|
Packit |
5c3484 |
mov -8(rp,n,8), %r8
|
|
Packit |
5c3484 |
jmp L(e0)
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
ALIGN(16)
|
|
Packit |
5c3484 |
L(b2): mov $-1, n
|
|
Packit |
5c3484 |
sub n_param, n
|
|
Packit |
5c3484 |
mul v0
|
|
Packit |
5c3484 |
mov 8(rp,n,8), %r8
|
|
Packit |
5c3484 |
mov %rdx, %r9
|
|
Packit |
5c3484 |
jmp L(e2)
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
ALIGN(16)
|
|
Packit |
5c3484 |
L(b13): xor R32(%r9), R32(%r9)
|
|
Packit |
5c3484 |
test $2, R8(n_param)
|
|
Packit |
5c3484 |
jnz L(b3)
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
L(b1): mov $2, R32(n)
|
|
Packit |
5c3484 |
sub n_param, n
|
|
Packit |
5c3484 |
jns L(1)
|
|
Packit |
5c3484 |
mul v0
|
|
Packit |
5c3484 |
mov -16(rp,n,8), %r10
|
|
Packit |
5c3484 |
mov %rdx, %r11
|
|
Packit |
5c3484 |
jmp L(e1)
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
ALIGN(16)
|
|
Packit |
5c3484 |
L(b3): xor R32(n), R32(n)
|
|
Packit |
5c3484 |
sub n_param, n
|
|
Packit |
5c3484 |
mul v0
|
|
Packit |
5c3484 |
mov (rp,n,8), %r10
|
|
Packit |
5c3484 |
jmp L(e3)
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
ALIGN(32)
|
|
Packit |
5c3484 |
L(top): mul v0
|
|
Packit |
5c3484 |
mov -16(rp,n,8), %r10
|
|
Packit |
5c3484 |
ADDSUB %r11, %r8
|
|
Packit |
5c3484 |
mov %rdx, %r11
|
|
Packit |
5c3484 |
adc $0, %r9
|
|
Packit |
5c3484 |
mov %r8, -24(rp,n,8)
|
|
Packit |
5c3484 |
L(e1): ADDSUB %rax, %r10
|
|
Packit |
5c3484 |
mov -8(up,n,8), %rax
|
|
Packit |
5c3484 |
adc $0, %r11
|
|
Packit |
5c3484 |
mul v0
|
|
Packit |
5c3484 |
ADDSUB %r9, %r10
|
|
Packit |
5c3484 |
mov %rdx, %r9
|
|
Packit |
5c3484 |
mov -8(rp,n,8), %r8
|
|
Packit |
5c3484 |
adc $0, %r11
|
|
Packit |
5c3484 |
mov %r10, -16(rp,n,8)
|
|
Packit |
5c3484 |
L(e0): ADDSUB %rax, %r8
|
|
Packit |
5c3484 |
adc $0, %r9
|
|
Packit |
5c3484 |
mov (up,n,8), %rax
|
|
Packit |
5c3484 |
mul v0
|
|
Packit |
5c3484 |
mov (rp,n,8), %r10
|
|
Packit |
5c3484 |
ADDSUB %r11, %r8
|
|
Packit |
5c3484 |
mov %r8, -8(rp,n,8)
|
|
Packit |
5c3484 |
adc $0, %r9
|
|
Packit |
5c3484 |
L(e3): mov %rdx, %r11
|
|
Packit |
5c3484 |
ADDSUB %rax, %r10
|
|
Packit |
5c3484 |
mov 8(up,n,8), %rax
|
|
Packit |
5c3484 |
adc $0, %r11
|
|
Packit |
5c3484 |
mul v0
|
|
Packit |
5c3484 |
mov 8(rp,n,8), %r8
|
|
Packit |
5c3484 |
ADDSUB %r9, %r10
|
|
Packit |
5c3484 |
mov %rdx, %r9
|
|
Packit |
5c3484 |
mov %r10, (rp,n,8)
|
|
Packit |
5c3484 |
adc $0, %r11
|
|
Packit |
5c3484 |
L(e2): ADDSUB %rax, %r8
|
|
Packit |
5c3484 |
adc $0, %r9
|
|
Packit |
5c3484 |
mov 16(up,n,8), %rax
|
|
Packit |
5c3484 |
add $4, n
|
|
Packit |
5c3484 |
jnc L(top)
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
L(end): mul v0
|
|
Packit |
5c3484 |
mov I(-8(rp),-16(rp,n,8)), %r10
|
|
Packit |
5c3484 |
ADDSUB %r11, %r8
|
|
Packit |
5c3484 |
mov %rdx, %r11
|
|
Packit |
5c3484 |
adc $0, %r9
|
|
Packit |
5c3484 |
mov %r8, I(-16(rp),-24(rp,n,8))
|
|
Packit |
5c3484 |
ADDSUB %rax, %r10
|
|
Packit |
5c3484 |
adc $0, %r11
|
|
Packit |
5c3484 |
ADDSUB %r9, %r10
|
|
Packit |
5c3484 |
adc $0, %r11
|
|
Packit |
5c3484 |
mov %r10, I(-8(rp),-16(rp,n,8))
|
|
Packit |
5c3484 |
mov %r11, %rax
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
pop %rbx
|
|
Packit |
5c3484 |
IFDOS(``pop %rdi '')
|
|
Packit |
5c3484 |
IFDOS(``pop %rsi '')
|
|
Packit |
5c3484 |
ret
|
|
Packit |
5c3484 |
|
|
Packit |
5c3484 |
ALIGN(16)
|
|
Packit |
5c3484 |
L(1): mul v0
|
|
Packit |
5c3484 |
ADDSUB %rax, -8(rp)
|
|
Packit |
5c3484 |
mov %rdx, %rax
|
|
Packit |
5c3484 |
adc $0, %rax
|
|
Packit |
5c3484 |
pop %rbx
|
|
Packit |
5c3484 |
IFDOS(``pop %rdi '')
|
|
Packit |
5c3484 |
IFDOS(``pop %rsi '')
|
|
Packit |
5c3484 |
ret
|
|
Packit |
5c3484 |
EPILOGUE()
|
|
Packit |
5c3484 |
ASM_END()
|
|
rpm-build |
c3cd4f |
CF_PROT
|