Return to addmul_1.asm CVS log

Up to [local] / OpenXM_contrib / gmp / mpn / x86 / pentium4 / sse2

Annotation of OpenXM_contrib/gmp/mpn/x86/pentium4/sse2/addmul_1.asm, Revision 1.1.1.1

1.1       ohara       1: dnl  Intel Pentium-4 mpn_addmul_1 -- Multiply a limb vector with a limb and add
                      2: dnl  the result to a second limb vector.
                      3:
                      4: dnl  Copyright 2001, 2002 Free Software Foundation, Inc.
                      5: dnl
                      6: dnl  This file is part of the GNU MP Library.
                      7: dnl
                      8: dnl  The GNU MP Library is free software; you can redistribute it and/or
                      9: dnl  modify it under the terms of the GNU Lesser General Public License as
                     10: dnl  published by the Free Software Foundation; either version 2.1 of the
                     11: dnl  License, or (at your option) any later version.
                     12: dnl
                     13: dnl  The GNU MP Library is distributed in the hope that it will be useful,
                     14: dnl  but WITHOUT ANY WARRANTY; without even the implied warranty of
                     15: dnl  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
                     16: dnl  Lesser General Public License for more details.
                     17: dnl
                     18: dnl  You should have received a copy of the GNU Lesser General Public
                     19: dnl  License along with the GNU MP Library; see the file COPYING.LIB.  If
                     20: dnl  not, write to the Free Software Foundation, Inc., 59 Temple Place -
                     21: dnl  Suite 330, Boston, MA 02111-1307, USA.
                     22:
                     23: include(`../config.m4')
                     24:
                     25:
                     26: C P4: 6 cycles/limb, unstable timing, at least on early Pentium4 silicon
                     27: C     (stepping 10).
                     28:
                     29:
                     30: C mp_limb_t mpn_addmul_1 (mp_ptr dst, mp_srcptr src, mp_size_t size,
                     31: C                         mp_limb_t multiplier);
                     32: C mp_limb_t mpn_addmul_1c (mp_ptr dst, mp_srcptr src, mp_size_t size,
                     33: C                          mp_limb_t multiplier, mp_limb_t carry);
                     34: C
                     35: C Only the carry limb propagation is on the dependent chain, but some other
                     36: C Pentium4 pipeline magic brings down performance to 6 cycles/l from the
                     37: C ideal 4 cycles/l.
                     38:
                     39: defframe(PARAM_CARRY,     20)
                     40: defframe(PARAM_MULTIPLIER,16)
                     41: defframe(PARAM_SIZE,      12)
                     42: defframe(PARAM_SRC,       8)
                     43: defframe(PARAM_DST,       4)
                     44:
                     45:        TEXT
                     46:        ALIGN(16)
                     47: PROLOGUE(mpn_addmul_1c)
                     48: deflit(`FRAME',0)
                     49:        movd    PARAM_CARRY, %mm0
                     50:        jmp     L(start_1c)
                     51: EPILOGUE()
                     52:
                     53: PROLOGUE(mpn_addmul_1)
                     54: deflit(`FRAME',0)
                     55:        pxor    %mm0, %mm0
                     56: L(start_1c):
                     57:        movl    PARAM_SRC, %eax
                     58:        movl    PARAM_SIZE, %ecx
                     59:        movl    PARAM_DST, %edx
                     60:        movd    PARAM_MULTIPLIER, %mm7
                     61:
                     62:        C eax   src, incrementing
                     63:        C ebx
                     64:        C ecx   loop counter, decrementing
                     65:        C edx   dst, incrementing
                     66:        C
                     67:        C mm0   carry, low 32-bits
                     68:        C mm7   multiplier
                     69:
                     70: L(loop):
                     71:        movd    (%eax), %mm1            C src
                     72:        leal    4(%eax), %eax
                     73:        movd    (%edx),%mm2             C dst
                     74:        pmuludq %mm7, %mm1
                     75:        paddq   %mm1, %mm2              C prod
                     76:        paddq   %mm2, %mm0              C carry
                     77:        subl    $1, %ecx
                     78:        movd    %mm0, (%edx)            C result
                     79:        psrlq   $32, %mm0
                     80:        leal    4(%edx), %edx
                     81:        jnz     L(loop)
                     82:
                     83:        movd    %mm0, %eax
                     84:        emms
                     85:        ret
                     86:
                     87: EPILOGUE()