source/libs/gmp/gmp-src/mpn/x86_64/sublsh1_n.asm

   1 dnl  AMD64 mpn_sublsh1_n -- rp[] = up[] - (vp[] << 1)
   2
   3 dnl  Copyright 2003, 2005-2007, 2011, 2012 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
  34 C            cycles/limb
  35 C AMD K8,K9      2.2
  36 C AMD K10        2.2
  37 C Intel P4      12.75
  38 C Intel core2    3.45
  39 C Intel corei    ?
  40 C Intel atom     ?
  41 C VIA nano       3.25
  42
  43 C Sometimes speed degenerates, supposedly related to that some operand
  44 C alignments cause cache conflicts.
  45
  46 C The speed is limited by decoding/issue bandwidth.  There are 26 instructions
  47 C in the loop, which corresponds to 26/3/4 = 2.167 c/l.
  48
  49 C INPUT PARAMETERS
  50 define(`rp',`%rdi')
  51 define(`up',`%rsi')
  52 define(`vp',`%rdx')
  53 define(`n', `%rcx')
  54
  55 ABI_SUPPORT(DOS64)
  56 ABI_SUPPORT(STD64)
  57
  58 ASM_START()
  59         TEXT
  60         ALIGN(16)
  61 PROLOGUE(mpn_sublsh1_n)
  62         FUNC_ENTRY(4)
  63         push    %rbx
  64         push    %rbp
  65
  66         mov     (vp), %r8
  67         mov     R32(n), R32(%rax)
  68         lea     (rp,n,8), rp
  69         lea     (up,n,8), up
  70         lea     (vp,n,8), vp
  71         neg     n
  72         xor     R32(%rbp), R32(%rbp)
  73         and     $3, R32(%rax)
  74         je      L(b00)
  75         cmp     $2, R32(%rax)
  76         jc      L(b01)
  77         je      L(b10)
  78
  79 L(b11): add     %r8, %r8
  80         mov     8(vp,n,8), %r9
  81         adc     %r9, %r9
  82         mov     16(vp,n,8), %r10
  83         adc     %r10, %r10
  84         sbb     R32(%rax), R32(%rax)    C save scy
  85         mov     (up,n,8), %rbp
  86         mov     8(up,n,8), %rbx
  87         sub     %r8, %rbp
  88         sbb     %r9, %rbx
  89         mov     %rbp, (rp,n,8)
  90         mov     %rbx, 8(rp,n,8)
  91         mov     16(up,n,8), %rbp
  92         sbb     %r10, %rbp
  93         mov     %rbp, 16(rp,n,8)
  94         sbb     R32(%rbp), R32(%rbp)    C save acy
  95         add     $3, n
  96         jmp     L(ent)
  97
  98 L(b10): add     %r8, %r8
  99         mov     8(vp,n,8), %r9
 100         adc     %r9, %r9
 101         sbb     R32(%rax), R32(%rax)    C save scy
 102         mov     (up,n,8), %rbp
 103         mov     8(up,n,8), %rbx
 104         sub     %r8, %rbp
 105         sbb     %r9, %rbx
 106         mov     %rbp, (rp,n,8)
 107         mov     %rbx, 8(rp,n,8)
 108         sbb     R32(%rbp), R32(%rbp)    C save acy
 109         add     $2, n
 110         jmp     L(ent)
 111
 112 L(b01): add     %r8, %r8
 113         sbb     R32(%rax), R32(%rax)    C save scy
 114         mov     (up,n,8), %rbp
 115         sub     %r8, %rbp
 116         mov     %rbp, (rp,n,8)
 117         sbb     R32(%rbp), R32(%rbp)    C save acy
 118         inc     n
 119 L(ent): jns     L(end)
 120
 121         ALIGN(16)
 122 L(top): add     R32(%rax), R32(%rax)    C restore scy
 123
 124         mov     (vp,n,8), %r8
 125 L(b00): adc     %r8, %r8
 126         mov     8(vp,n,8), %r9
 127         adc     %r9, %r9
 128         mov     16(vp,n,8), %r10
 129         adc     %r10, %r10
 130         mov     24(vp,n,8), %r11
 131         adc     %r11, %r11
 132
 133         sbb     R32(%rax), R32(%rax)    C save scy
 134         add     R32(%rbp), R32(%rbp)    C restore acy
 135
 136         mov     (up,n,8), %rbp
 137         mov     8(up,n,8), %rbx
 138         sbb     %r8, %rbp
 139         sbb     %r9, %rbx
 140         mov     %rbp, (rp,n,8)
 141         mov     %rbx, 8(rp,n,8)
 142         mov     16(up,n,8), %rbp
 143         mov     24(up,n,8), %rbx
 144         sbb     %r10, %rbp
 145         sbb     %r11, %rbx
 146         mov     %rbp, 16(rp,n,8)
 147         mov     %rbx, 24(rp,n,8)
 148
 149         sbb     R32(%rbp), R32(%rbp)    C save acy
 150         add     $4, n
 151         js      L(top)
 152
 153 L(end): add     R32(%rbp), R32(%rax)
 154         neg     R32(%rax)
 155
 156         pop     %rbp
 157         pop     %rbx
 158         FUNC_EXIT()
 159         ret
 160 EPILOGUE()