dnl AMD64 mpn_mullo_basecase optimised for Intel Haswell.
dnl Contributed to the GNU project by Torbjörn Granlund.
dnl Copyright 2008, 2009, 2011-2013 Free Software Foundation, Inc.
dnl This file is part of the GNU MP Library.
dnl
dnl The GNU MP Library is free software; you can redistribute it and/or modify
dnl it under the terms of either:
dnl
dnl * the GNU Lesser General Public License as published by the Free
dnl Software Foundation; either version 3 of the License, or (at your
dnl option) any later version.
dnl
dnl or
dnl
dnl * the GNU General Public License as published by the Free Software
dnl Foundation; either version 2 of the License, or (at your option) any
dnl later version.
dnl
dnl or both in parallel, as here.
dnl
dnl The GNU MP Library is distributed in the hope that it will be useful, but
dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
dnl for more details.
dnl
dnl You should have received copies of the GNU General Public License and the
dnl GNU Lesser General Public License along with the GNU MP Library. If not,
dnl see https://www.gnu.org/licenses/.
include(`../config.m4')
C cycles/limb mul_2 addmul_2
C AMD K8,K9 n/a n/a
C AMD K10 n/a n/a
C AMD bull n/a n/a
C AMD pile n/a n/a
C AMD steam ? ?
C AMD bobcat n/a n/a
C AMD jaguar ? ?
C Intel P4 n/a n/a
C Intel core n/a n/a
C Intel NHM n/a n/a
C Intel SBR n/a n/a
C Intel IBR n/a n/a
C Intel HWL 1.86 2.15
C Intel BWL ? ?
C Intel atom n/a n/a
C VIA nano n/a n/a
C The inner loops of thiscode are the result of running a code generation and
C optimisation tool suite written by David Harvey and Torbjörn Granlund.
C TODO
C * Implement proper cor2, replacing current cor0.
C * Micro-optimise.
mov 16(%rsp), up C restore `up' to beginning lea 16(vp), vp lea 24(rp,n,8), rp C put back rp to old rp + 2 add $2, n jl L(outer)
pop %r15 pop %r14
jnz L(cor0)
L(cor1):mov (vp), v0 mov 8(vp), v1 mov (up), %rdx
mulx( v0, %r12, %rbp) C u0 x v2 add (rp), %r12 C FIXME: rp[0] still available in reg? adc %rax, %rbp mov 8(up), %r10
imul v0, %r10
imul v1, %rdx mov %r12, (rp) add %r10, %rdx add %rbp, %rdx mov %rdx, 8(rp) pop %rax C deallocate `up' copy pop %r13 pop %r12 pop %rbp pop %rbx
FUNC_EXIT() ret
L(cor0):mov (vp), %r11
imul (up), %r11 add %rax, %r11 mov %r11, (rp) pop %rax C deallocate `up' copy pop %r13 pop %r12 pop %rbp pop %rbx
FUNC_EXIT() ret
ALIGN(16)
L(small): cmp $2, n
jae L(gt1)
L(n1): imul (vp), %rdx mov %rdx, (rp)
FUNC_EXIT() ret
L(gt1): ja L(gt2)
L(n2): mov (vp), %r9
mulx( %r9, %rax, %rdx) mov %rax, (rp) mov 8(up), %rax
imul %r9, %rax add %rax, %rdx mov 8(vp), %r9 mov (up), %rcx
imul %r9, %rcx add %rcx, %rdx mov %rdx, 8(rp)
FUNC_EXIT() ret
L(gt2):
L(n3): mov (vp), %r9
mulx( %r9, %rax, %r10) C u0 x v0 mov %rax, (rp) mov 8(up), %rdx
mulx( %r9, %rax, %rdx) C u1 x v0
imul 16(up), %r9 C u2 x v0 add %rax, %r10 adc %rdx, %r9 mov 8(vp), %r11 mov (up), %rdx
mulx( %r11, %rax, %rdx) C u0 x v1 add %rax, %r10 adc %rdx, %r9
imul 8(up), %r11 C u1 x v1 add %r11, %r9 mov %r10, 8(rp) mov 16(vp), %r10 mov (up), %rax
imul %rax, %r10 C u0 x v2 add %r10, %r9 mov %r9, 16(rp)
FUNC_EXIT() ret
EPILOGUE()
Messung V0.5
¤ Dauer der Verarbeitung: 0.14 Sekunden
(vorverarbeitet)
¤
Die Informationen auf dieser Webseite wurden
nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit,
noch Qualität der bereit gestellten Informationen zugesichert.
Bemerkung:
Die farbliche Syntaxdarstellung und die Messung sind noch experimentell.