Return to mul_1.asm CVS log

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

Annotation of OpenXM_contrib/gmp/mpn/arm/mul_1.asm, Revision 1.1.1.1

1.1       ohara       1: dnl  ARM mpn_mul_1 -- Multiply a limb vector with a limb and store the result
                      2: dnl  in a second limb vector.
                      3: dnl  Contributed by Robert Harley.
                      4:
                      5: dnl  Copyright 1998, 2000, 2001 Free Software Foundation, Inc.
                      6:
                      7: dnl  This file is part of the GNU MP Library.
                      8:
                      9: dnl  The GNU MP Library is free software; you can redistribute it and/or modify
                     10: dnl  it under the terms of the GNU Lesser General Public License as published
                     11: dnl  by the Free Software Foundation; either version 2.1 of the License, or (at
                     12: dnl  your option) any later version.
                     13:
                     14: dnl  The GNU MP Library is distributed in the hope that it will be useful, but
                     15: dnl  WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
                     16: dnl  or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public
                     17: dnl  License for more details.
                     18:
                     19: dnl  You should have received a copy of the GNU Lesser General Public License
                     20: dnl  along with the GNU MP Library; see the file COPYING.LIB.  If not, write to
                     21: dnl  the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
                     22: dnl  MA 02111-1307, USA.
                     23:
                     24: include(`../config.m4')
                     25:
                     26: C This runs at 7.75 cycles/limb in the StrongARM.
                     27:
                     28: define(`rp',`r0')
                     29: define(`up',`r1')
                     30: define(`n',`r2')
                     31: define(`v',`r3')
                     32:
                     33:
                     34: ASM_START()
                     35: PROLOGUE(mpn_mul_1)
                     36:        stmfd   sp!, { r8, r9, lr }
                     37:        ands    ip, n, #1
                     38:        beq     L(skip1)
                     39:        ldr     lr, [up], #4
                     40:        umull   r9, ip, v, lr
                     41:        str     r9, [rp], #4
                     42: L(skip1):
                     43:        tst     n, #2
                     44:        beq     L(skip2)
                     45:        mov     r8, ip
                     46:        ldmia   up!, { ip, lr }
                     47:        mov     r9, #0
                     48:        umlal   r8, r9, v, ip
                     49:        mov     ip, #0
                     50:        umlal   r9, ip, v, lr
                     51:        stmia   rp!, { r8, r9 }
                     52: L(skip2):
                     53:        bics    n, n, #3
                     54:        beq     L(return)
                     55:        stmfd   sp!, { r6, r7 }
                     56: L(mul_1_loop):
                     57:        mov     r6, ip
                     58:        ldmia   up!, { r8, r9, ip, lr }
                     59:        ldr     r7, [rp, #12]                   C cache allocate
                     60:        mov     r7, #0
                     61:        umlal   r6, r7, v, r8
                     62:        mov     r8, #0
                     63:        umlal   r7, r8, v, r9
                     64:        mov     r9, #0
                     65:        umlal   r8, r9, v, ip
                     66:        mov     ip, #0
                     67:        umlal   r9, ip, v, lr
                     68:        subs    n, n, #4
                     69:        stmia   rp!, { r6, r7, r8, r9 }
                     70:        bne     L(mul_1_loop)
                     71:        ldmfd   sp!, { r6, r7 }
                     72: L(return):
                     73:        mov     r0, ip
                     74:        ldmfd   sp!, { r8, r9, pc }
                     75: EPILOGUE(mpn_mul_1)