github.com/aergoio/aergo@v1.3.1/libtool/src/gmp-6.1.2/mpn/x86/atom/cnd_add_n.asm (about)

     1  dnl  X86 mpn_cnd_add_n optimised for Intel Atom.
     2  
     3  dnl  Copyright 2013 Free Software Foundation, Inc.
     4  
     5  dnl  This file is part of the GNU MP Library.
     6  dnl
     7  dnl  The GNU MP Library is free software; you can redistribute it and/or modify
     8  dnl  it under the terms of either:
     9  dnl
    10  dnl    * the GNU Lesser General Public License as published by the Free
    11  dnl      Software Foundation; either version 3 of the License, or (at your
    12  dnl      option) any later version.
    13  dnl
    14  dnl  or
    15  dnl
    16  dnl    * the GNU General Public License as published by the Free Software
    17  dnl      Foundation; either version 2 of the License, or (at your option) any
    18  dnl      later version.
    19  dnl
    20  dnl  or both in parallel, as here.
    21  dnl
    22  dnl  The GNU MP Library is distributed in the hope that it will be useful, but
    23  dnl  WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
    24  dnl  or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
    25  dnl  for more details.
    26  dnl
    27  dnl  You should have received copies of the GNU General Public License and the
    28  dnl  GNU Lesser General Public License along with the GNU MP Library.  If not,
    29  dnl  see https://www.gnu.org/licenses/.
    30  
    31  include(`../config.m4')
    32  
    33  C			    cycles/limb
    34  C P5				 ?
    35  C P6 model 0-8,10-12		 ?
    36  C P6 model 9   (Banias)		 ?
    37  C P6 model 13  (Dothan)		 ?
    38  C P4 model 0-1 (Willamette)	 ?
    39  C P4 model 2   (Northwood)	 ?
    40  C P4 model 3-4 (Prescott)	 ?
    41  C Intel atom			 4.67
    42  C AMD K6			 ?
    43  C AMD K7			 ?
    44  C AMD K8			 ?
    45  
    46  
    47  define(`rp',  `%edi')
    48  define(`up',  `%esi')
    49  define(`vp',  `%ebp')
    50  define(`n',   `%ecx')
    51  define(`cnd', `20(%esp)')
    52  
    53  ASM_START()
    54  	TEXT
    55  	ALIGN(16)
    56  PROLOGUE(mpn_cnd_add_n)
    57  	push	%edi
    58  	push	%esi
    59  	push	%ebx
    60  	push	%ebp
    61  
    62  	mov	cnd, %eax		C make cnd into a mask (1)
    63  	mov	24(%esp), rp
    64  	neg	%eax			C make cnd into a mask (1)
    65  	mov	28(%esp), up
    66  	sbb	%eax, %eax		C make cnd into a mask (1)
    67  	mov	32(%esp), vp
    68  	mov	%eax, cnd		C make cnd into a mask (1)
    69  	mov	36(%esp), n
    70  
    71  	xor	%edx, %edx
    72  
    73  	shr	$1, n
    74  	jnc	L(top)
    75  
    76  	mov	0(vp), %eax
    77  	and	cnd, %eax
    78  	lea	4(vp), vp
    79  	add	0(up), %eax
    80  	lea	4(rp), rp
    81  	lea	4(up), up
    82  	sbb	%edx, %edx
    83  	mov	%eax, -4(rp)
    84  	inc	n
    85  	dec	n
    86  	je	L(end)
    87  
    88  L(top):	sbb	%edx, %edx
    89  	mov	0(vp), %eax
    90  	and	cnd, %eax
    91  	lea	8(vp), vp
    92  	lea	8(rp), rp
    93  	mov	-4(vp), %ebx
    94  	and	cnd, %ebx
    95  	add	%edx, %edx
    96  	adc	0(up), %eax
    97  	lea	8(up), up
    98  	mov	%eax, -8(rp)
    99  	adc	-4(up), %ebx
   100  	dec	n
   101  	mov	%ebx, -4(rp)
   102  	jne	L(top)
   103  
   104  L(end):	mov	$0, %eax
   105  	adc	%eax, %eax
   106  
   107  	pop	%ebp
   108  	pop	%ebx
   109  	pop	%esi
   110  	pop	%edi
   111  	ret
   112  EPILOGUE()
   113  ASM_END()