dnl Contributed to the GNU project by Torbjorn Granlund and Marco Bodrato.
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 This is an attempt at a sublsh1_n for x86-32, not relying on sse2 insns. The
C innerloop is 2*3-way unrolled, which is best we can do with the available
C registers. It seems tricky to use the same structure for rsblsh1_n, since we
C cannot feed carry between operations there.
C cycles/limb
C P5
C P6 model 0-8,10-12
C P6 model 9 (Banias)
C P6 model 13 (Dothan)
C P4 model 0 (Willamette)
C P4 model 1 (?)
C P4 model 2 (Northwood)
C P4 model 3 (Prescott)
C P4 model 4 (Nocona)
C Intel Atom 6.75
C AMD K6
C AMD K7
C AMD K8
C This is a basic sublsh1_n for k7, atom, and perhaps some other x86-32
C processors. It uses 2*4-way unrolling, for good reasons.
C
C Breaking carry recurrency might be a good idea. We would then need separate
C registers for the shift carryandadd/subtract carry, which in turn would
C force us to 2*2-way unrolling.
dnl re-use parameter space
define(VAR_COUNT,`PARAM_SIZE')
define(SAVE_EBX,`PARAM_SRC')
define(SAVE_EBP,`PARAM_DST')
ASM_START()
TEXT ALIGN(8)
PROLOGUE(mpn_sublsh1_n_ip1)
deflit(`FRAME',0)
define(`rp', `%edi')
define(`up', `%esi')
mov PARAM_SIZE, %eax C size push up FRAME_pushl() push rp FRAME_pushl() xor %edx, %edx mov PARAM_SRC, up mov PARAM_DST, rp mov %ebx, SAVE_EBX mov %eax, %ebx shr $3, %eax
not %eax C count = -(size\8)-i and $7, %ebx C size % 8 jz L(exact)
L(oop): ifdef(`CPU_P6',` shr %edx') C restore 2nd saved carry bit mov (up), %ecx adc %ecx, %ecx
rcr %edx C restore 1st saved carry bit lea 4(up), up
sbb %ecx, (rp) lea 4(rp), rp adc %edx, %edx C save a carry bit in edx ifdef(`CPU_P6',` adc %edx, %edx') C save another carry bit in edx dec %ebx jnz L(oop)
L(exact): inc %eax jz L(end) mov %eax, VAR_COUNT mov %ebp, SAVE_EBP
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.