Return to addmul_1.asm CVS log

Up to [local] / OpenXM_contrib / gmp / mpn / powerpc32

Annotation of OpenXM_contrib/gmp/mpn/powerpc32/addmul_1.asm, Revision 1.1.1.2

1.1       maekawa     1: dnl PowerPC-32 mpn_addmul_1 -- Multiply a limb vector with a limb and add
                      2: dnl the result to a second limb vector.
                      3:
1.1.1.2 ! ohara       4: dnl Copyright 1995, 1997, 1998, 2000, 2001, 2002 Free Software Foundation, Inc.
1.1       maekawa     5:
                      6: dnl This file is part of the GNU MP Library.
                      7:
                      8: dnl The GNU MP Library is free software; you can redistribute it and/or modify
                      9: dnl it under the terms of the GNU Lesser General Public License as published by
                     10: dnl the Free Software Foundation; either version 2.1 of the License, or (at your
                     11: dnl option) any later version.
                     12:
                     13: dnl The GNU MP Library is distributed in the hope that it will be useful, but
                     14: dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
                     15: dnl or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public
                     16: dnl License for more details.
                     17:
                     18: dnl You should have received a copy of the GNU Lesser General Public License
                     19: dnl along with the GNU MP Library; see the file COPYING.LIB.  If not, write to
                     20: dnl the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
                     21: dnl MA 02111-1307, USA.
                     22:
                     23:
                     24: dnl INPUT PARAMETERS
                     25: dnl res_ptr    r3
                     26: dnl s1_ptr     r4
                     27: dnl size       r5
                     28: dnl s2_limb    r6
                     29:
1.1.1.2 ! ohara      30: dnl This is optimized for the PPC604.  It has not been tuned for PPC601,
        !            31: dnl PPC603, PPC750 (G3), 7400 (G4), 7450 (newer G4).
        !            32: dnl
        !            33: dnl Loop Analysis for the 604:
        !            34: dnl 12 mem insn
        !            35: dnl 8 serializing insn
        !            36: dnl 8 int multiply
        !            37: dnl 25 int reg write
        !            38: dnl 9 int ops (8 of which serialize)
        !            39: dnl
        !            40: dnl The multiply insns need 16 cycles/4limb.
        !            41: dnl The integer register writes will need 13 cycles/4limb.
        !            42: dnl All-in-all, it should be possible to get to 4 cycles/limb,
        !            43: dnl but that will require some clever FPNOPS and BNOPS for exact
        !            44: dnl issue control.
1.1       maekawa    45:
                     46: include(`../config.m4')
                     47:
                     48: ASM_START()
                     49: PROLOGUE(mpn_addmul_1)
                     50:        cmpi    cr0,r5,9        C more than 9 limbs?
1.1.1.2 ! ohara      51:        bgt     cr0,L(big)      C branch if more than 9 limbs
1.1       maekawa    52:
                     53:        mtctr   r5
                     54:        lwz     r0,0(r4)
                     55:        mullw   r7,r0,r6
                     56:        mulhwu  r10,r0,r6
                     57:        lwz     r9,0(r3)
                     58:        addc    r8,r7,r9
                     59:        addi    r3,r3,-4
1.1.1.2 ! ohara      60:        bdz     L(end)
        !            61: L(loop):
1.1       maekawa    62:        lwzu    r0,4(r4)
                     63:        stwu    r8,4(r3)
                     64:        mullw   r8,r0,r6
                     65:        adde    r7,r8,r10
                     66:        mulhwu  r10,r0,r6
                     67:        lwz     r9,4(r3)
                     68:        addze   r10,r10
                     69:        addc    r8,r7,r9
1.1.1.2 ! ohara      70:        bdnz    L(loop)
        !            71: L(end):        stw     r8,4(r3)
1.1       maekawa    72:        addze   r3,r10
                     73:        blr
                     74:
1.1.1.2 ! ohara      75: L(big):        stmw    r30,-32(r1)
1.1       maekawa    76:        addi    r5,r5,-1
                     77:        srwi    r0,r5,2
                     78:        mtctr   r0
                     79:
                     80:        lwz     r7,0(r4)
                     81:        mullw   r8,r7,r6
                     82:        mulhwu  r0,r7,r6
                     83:        lwz     r7,0(r3)
                     84:        addc    r8,r8,r7
                     85:        stw     r8,0(r3)
                     86:
1.1.1.2 ! ohara      87: L(loopU):
1.1       maekawa    88:        lwz     r7,4(r4)
                     89:        lwz     r12,8(r4)
                     90:        lwz     r30,12(r4)
                     91:        lwzu    r31,16(r4)
                     92:        mullw   r8,r7,r6
                     93:        mullw   r9,r12,r6
                     94:        mullw   r10,r30,r6
                     95:        mullw   r11,r31,r6
                     96:        adde    r8,r8,r0        C add cy_limb
                     97:        mulhwu  r0,r7,r6
                     98:        lwz     r7,4(r3)
                     99:        adde    r9,r9,r0
                    100:        mulhwu  r0,r12,r6
                    101:        lwz     r12,8(r3)
                    102:        adde    r10,r10,r0
                    103:        mulhwu  r0,r30,r6
                    104:        lwz     r30,12(r3)
                    105:        adde    r11,r11,r0
                    106:        mulhwu  r0,r31,r6
                    107:        lwz     r31,16(r3)
                    108:        addze   r0,r0           C new cy_limb
                    109:        addc    r8,r8,r7
                    110:        stw     r8,4(r3)
                    111:        adde    r9,r9,r12
                    112:        stw     r9,8(r3)
                    113:        adde    r10,r10,r30
                    114:        stw     r10,12(r3)
                    115:        adde    r11,r11,r31
                    116:        stwu    r11,16(r3)
1.1.1.2 ! ohara     117:        bdnz    L(loopU)
1.1       maekawa   118:
                    119:        andi.   r31,r5,3
                    120:        mtctr   r31
1.1.1.2 ! ohara     121:        beq     cr0,L(endx)
1.1       maekawa   122:
1.1.1.2 ! ohara     123: L(loopE):
1.1       maekawa   124:        lwzu    r7,4(r4)
                    125:        mullw   r8,r7,r6
                    126:        adde    r8,r8,r0        C add cy_limb
                    127:        mulhwu  r0,r7,r6
                    128:        lwz     r7,4(r3)
                    129:        addze   r0,r0           C new cy_limb
                    130:        addc    r8,r8,r7
                    131:        stwu    r8,4(r3)
1.1.1.2 ! ohara     132:        bdnz    L(loopE)
        !           133: L(endx):
1.1       maekawa   134:        addze   r3,r0
                    135:        lmw     r30,-32(r1)
                    136:        blr
                    137: EPILOGUE(mpn_addmul_1)