URL https://opencores.org/ocsvn/openrisc/openrisc/trunk

Subversion Repositories openrisc

[/] [openrisc/] [tags/] [gnu-src/] [newlib-1.18.0/] [newlib-1.18.0-or32-1.0rc1/] [newlib/] [libc/] [machine/] [x86_64/] [memcpy.S] - Blame information for rev 802

Go to most recent revision | Details | Compare with Previous | View Log


/*
 * ====================================================
 * Copyright (C) 2007 by Ellips BV. All rights reserved.
 *
 * Permission to use, copy, modify, and distribute this
 * software is freely granted, provided that this notice
 * is preserved.
 * ====================================================
 */
 
  #include "x86_64mach.h"
 
  .global SYM (memcpy)
  SOTYPE_FUNCTION(memcpy)
 
SYM (memcpy):
  movq    rdi, rax                /* Store destination in return value */
  cmpq    $16, rdx
  jb      byte_copy
 
  movq    rdi, r8                 /* Align destination on quad word boundary */
  andq    $7, r8
  jz      quadword_aligned
  movq    $8, rcx
  subq    r8, rcx
  subq    rcx, rdx
  rep     movsb
 
quadword_aligned:
  cmpq    $256, rdx
  jb      quadword_copy
 
  pushq    rax
  pushq    r12
  pushq    r13
  pushq    r14
 
  movq    rdx, rcx                /* Copy 128 bytes at a time with minimum cache polution */
  shrq    $7, rcx
 
  .p2align 4
loop:
  prefetchnta   768 (rsi)
  prefetchnta   832 (rsi)
 
  movq       (rsi), rax
  movq     8 (rsi), r8
  movq    16 (rsi), r9
  movq    24 (rsi), r10
  movq    32 (rsi), r11
  movq    40 (rsi), r12
  movq    48 (rsi), r13
  movq    56 (rsi), r14
 
  movntiq rax,    (rdi)
  movntiq r8 ,  8 (rdi)
  movntiq r9 , 16 (rdi)
  movntiq r10, 24 (rdi)
  movntiq r11, 32 (rdi)
  movntiq r12, 40 (rdi)
  movntiq r13, 48 (rdi)
  movntiq r14, 56 (rdi)
 
  movq     64 (rsi), rax
  movq     72 (rsi), r8
  movq     80 (rsi), r9
  movq     88 (rsi), r10
  movq     96 (rsi), r11
  movq    104 (rsi), r12
  movq    112 (rsi), r13
  movq    120 (rsi), r14
 
  movntiq rax,  64 (rdi)
  movntiq r8 ,  72 (rdi)
  movntiq r9 ,  80 (rdi)
  movntiq r10,  88 (rdi)
  movntiq r11,  96 (rdi)
  movntiq r12, 104 (rdi)
  movntiq r13, 112 (rdi)
  movntiq r14, 120 (rdi)
 
  leaq    128 (rsi), rsi
  leaq    128 (rdi), rdi
 
  dec     rcx
  jnz     loop
 
  sfence
  movq    rdx, rcx
  andq    $127, rcx
  rep     movsb
  popq    r14
  popq    r13
  popq    r12
  popq    rax
  ret
 
 
byte_copy:
  movq    rdx, rcx
  rep     movsb
  ret
 
 
quadword_copy:
  movq    rdx, rcx
  shrq    $3, rcx
  .p2align 4
  rep     movsq
  movq    rdx, rcx
  andq    $7, rcx
  rep     movsb                   /* Copy the remaining bytes */
  ret

Line No.	Rev	Author	Line
1	207	jeremybenn	`/*`
2			`* ====================================================`
3			`* Copyright (C) 2007 by Ellips BV. All rights reserved.`
4			`*`
5			`* Permission to use, copy, modify, and distribute this`
6			`* software is freely granted, provided that this notice`
7			`* is preserved.`
8			`* ====================================================`
9			`*/`
10
11			`#include "x86_64mach.h"`
12
13			`.global SYM (memcpy)`
14			`SOTYPE_FUNCTION(memcpy)`
15
16			`SYM (memcpy):`
17			`movq rdi, rax /* Store destination in return value */`
18			`cmpq $16, rdx`
19			`jb byte_copy`
20
21			`movq rdi, r8 /* Align destination on quad word boundary */`
22			`andq $7, r8`
23			`jz quadword_aligned`
24			`movq $8, rcx`
25			`subq r8, rcx`
26			`subq rcx, rdx`
27			`rep movsb`
28
29			`quadword_aligned:`
30			`cmpq $256, rdx`
31			`jb quadword_copy`
32
33			`pushq rax`
34			`pushq r12`
35			`pushq r13`
36			`pushq r14`
37
38			`movq rdx, rcx /* Copy 128 bytes at a time with minimum cache polution */`
39			`shrq $7, rcx`
40
41			`.p2align 4`
42			`loop:`
43			`prefetchnta 768 (rsi)`
44			`prefetchnta 832 (rsi)`
45
46			`movq (rsi), rax`
47			`movq 8 (rsi), r8`
48			`movq 16 (rsi), r9`
49			`movq 24 (rsi), r10`
50			`movq 32 (rsi), r11`
51			`movq 40 (rsi), r12`
52			`movq 48 (rsi), r13`
53			`movq 56 (rsi), r14`
54
55			`movntiq rax, (rdi)`
56			`movntiq r8 , 8 (rdi)`
57			`movntiq r9 , 16 (rdi)`
58			`movntiq r10, 24 (rdi)`
59			`movntiq r11, 32 (rdi)`
60			`movntiq r12, 40 (rdi)`
61			`movntiq r13, 48 (rdi)`
62			`movntiq r14, 56 (rdi)`
63
64			`movq 64 (rsi), rax`
65			`movq 72 (rsi), r8`
66			`movq 80 (rsi), r9`
67			`movq 88 (rsi), r10`
68			`movq 96 (rsi), r11`
69			`movq 104 (rsi), r12`
70			`movq 112 (rsi), r13`
71			`movq 120 (rsi), r14`
72
73			`movntiq rax, 64 (rdi)`
74			`movntiq r8 , 72 (rdi)`
75			`movntiq r9 , 80 (rdi)`
76			`movntiq r10, 88 (rdi)`
77			`movntiq r11, 96 (rdi)`
78			`movntiq r12, 104 (rdi)`
79			`movntiq r13, 112 (rdi)`
80			`movntiq r14, 120 (rdi)`
81
82			`leaq 128 (rsi), rsi`
83			`leaq 128 (rdi), rdi`
84
85			`dec rcx`
86			`jnz loop`
87
88			`sfence`
89			`movq rdx, rcx`
90			`andq $127, rcx`
91			`rep movsb`
92			`popq r14`
93			`popq r13`
94			`popq r12`
95			`popq rax`
96			`ret`
97
98
99			`byte_copy:`
100			`movq rdx, rcx`
101			`rep movsb`
102			`ret`
103
104
105			`quadword_copy:`
106			`movq rdx, rcx`
107			`shrq $3, rcx`
108			`.p2align 4`
109			`rep movsq`
110			`movq rdx, rcx`
111			`andq $7, rcx`
112			`rep movsb /* Copy the remaining bytes */`
113			`ret`

Browse

Tools

Subversion Repositories openrisc

[/] [openrisc/] [tags/] [gnu-src/] [newlib-1.18.0/] [newlib-1.18.0-or32-1.0rc1/] [newlib/] [libc/] [machine/] [x86_64/] [memcpy.S] - Blame information for rev 802