Annotation of OpenXM_contrib/gmp/mpn/x86/pentium/diveby3.asm, Revision 1.1.1.1
1.1 maekawa 1: dnl Intel P5 mpn_divexact_by3 -- mpn division by 3, expecting no remainder.
2: dnl
3: dnl P5: 15.0 cycles/limb
4:
5:
6: dnl Copyright (C) 2000 Free Software Foundation, Inc.
7: dnl
8: dnl This file is part of the GNU MP Library.
9: dnl
10: dnl The GNU MP Library is free software; you can redistribute it and/or
11: dnl modify it under the terms of the GNU Lesser General Public License as
12: dnl published by the Free Software Foundation; either version 2.1 of the
13: dnl License, or (at your option) any later version.
14: dnl
15: dnl The GNU MP Library is distributed in the hope that it will be useful,
16: dnl but WITHOUT ANY WARRANTY; without even the implied warranty of
17: dnl MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18: dnl Lesser General Public License for more details.
19: dnl
20: dnl You should have received a copy of the GNU Lesser General Public
21: dnl License along with the GNU MP Library; see the file COPYING.LIB. If
22: dnl not, write to the Free Software Foundation, Inc., 59 Temple Place -
23: dnl Suite 330, Boston, MA 02111-1307, USA.
24:
25:
26: include(`../config.m4')
27:
28:
29: C mp_limb_t mpn_divexact_by3c (mp_ptr dst, mp_srcptr src, mp_size_t size,
30: C mp_limb_t carry);
31:
32: defframe(PARAM_CARRY,16)
33: defframe(PARAM_SIZE, 12)
34: defframe(PARAM_SRC, 8)
35: defframe(PARAM_DST, 4)
36:
37: dnl multiplicative inverse of 3, modulo 2^32
38: deflit(INVERSE_3, 0xAAAAAAAB)
39:
40: dnl ceil(b/3), ceil(b*2/3) and floor(b*2/3) where b=2^32
41: deflit(ONE_THIRD_CEIL, 0x55555556)
42: deflit(TWO_THIRDS_CEIL, 0xAAAAAAAB)
43: deflit(TWO_THIRDS_FLOOR, 0xAAAAAAAA)
44:
45: .text
46: ALIGN(8)
47:
48: PROLOGUE(mpn_divexact_by3c)
49: deflit(`FRAME',0)
50:
51: movl PARAM_SRC, %ecx
52: movl PARAM_SIZE, %edx
53:
54: decl %edx
55: jnz L(two_or_more)
56:
57: movl (%ecx), %edx
58: movl PARAM_CARRY, %eax C risk of cache bank clash here
59:
60: movl PARAM_DST, %ecx
61: subl %eax, %edx
62:
63: sbbl %eax, %eax C 0 or -1
64:
65: imull $INVERSE_3, %edx, %edx
66:
67: negl %eax C 0 or 1
68: cmpl $ONE_THIRD_CEIL, %edx
69:
70: sbbl $-1, %eax C +1 if edx>=ceil(b/3)
71: cmpl $TWO_THIRDS_CEIL, %edx
72:
73: sbbl $-1, %eax C +1 if edx>=ceil(b*2/3)
74: movl %edx, (%ecx)
75:
76: ret
77:
78:
79: L(two_or_more):
80: C eax
81: C ebx
82: C ecx src
83: C edx size-1
84: C esi
85: C edi
86: C ebp
87:
88: pushl %ebx FRAME_pushl()
89: pushl %esi FRAME_pushl()
90:
91: pushl %edi FRAME_pushl()
92: pushl %ebp FRAME_pushl()
93:
94: movl PARAM_DST, %edi
95: movl PARAM_CARRY, %esi
96:
97: movl (%ecx), %eax C src low limb
98: xorl %ebx, %ebx
99:
100: sub %esi, %eax
101: movl $TWO_THIRDS_FLOOR, %esi
102:
103: leal (%ecx,%edx,4), %ecx C &src[size-1]
104: leal (%edi,%edx,4), %edi C &dst[size-1]
105:
106: adcl $0, %ebx C carry, 0 or 1
107: negl %edx C -(size-1)
108:
109:
110: C The loop needs a source limb ready at the top, which leads to one limb
111: C handled separately at the end, and the special case above for size==1.
112: C There doesn't seem to be any scheduling that would keep the speed but move
113: C the source load and carry subtract up to the top.
114: C
115: C The destination cache line prefetching adds 1 cycle to the loop but is
116: C considered worthwhile. The slowdown is a factor of 1.07, but will prevent
117: C repeated write-throughs if the destination isn't in L1. A version using
118: C an outer loop to prefetch only every 8 limbs (a cache line) proved to be
119: C no faster, due to unavoidable branch mispreditions in the inner loop.
120: C
121: C setc is 2 cycles on P54, so an adcl is used instead. If the movl $0,%ebx
122: C could be avoided then the src limb fetch could pair up and save a cycle.
123: C This would probably mean going to a two limb loop with the carry limb
124: C alternately positive or negative, since an sbbl %ebx,%ebx will leave a
125: C value which is in the opposite sense to the preceding sbbl/adcl %ebx,%eax.
126: C
127: C A register is used for TWO_THIRDS_FLOOR because a cmp can't be done as
128: C "cmpl %edx, $n" with the immediate as the second operand.
129: C
130: C The "4" source displacement is in the loop rather than the setup because
131: C this gets L(top) aligned to 8 bytes at no cost.
132:
133: ALIGN(8)
134: L(top):
135: C eax source limb, carry subtracted
136: C ebx carry (0 or 1)
137: C ecx &src[size-1]
138: C edx counter, limbs, negative
139: C esi TWO_THIRDS_FLOOR
140: C edi &dst[size-1]
141: C ebp scratch (result limb)
142:
143: imull $INVERSE_3, %eax, %ebp
144:
145: cmpl $ONE_THIRD_CEIL, %ebp
146: movl (%edi,%edx,4), %eax C dst cache line prefetch
147:
148: sbbl $-1, %ebx C +1 if ebp>=ceil(b/3)
149: cmpl %ebp, %esi
150:
151: movl 4(%ecx,%edx,4), %eax C next src limb
152:
153: sbbl %ebx, %eax C and further -1 if ebp>=ceil(b*2/3)
154: movl $0, %ebx
155:
156: adcl $0, %ebx C new carry
157: movl %ebp, (%edi,%edx,4)
158:
159: incl %edx
160: jnz L(top)
161:
162:
163:
164: imull $INVERSE_3, %eax, %edx
165:
166: cmpl $ONE_THIRD_CEIL, %edx
167: movl %edx, (%edi)
168:
169: sbbl $-1, %ebx C +1 if edx>=ceil(b/3)
170: cmpl $TWO_THIRDS_CEIL, %edx
171:
172: sbbl $-1, %ebx C +1 if edx>=ceil(b*2/3)
173: popl %ebp
174:
175: movl %ebx, %eax
176: popl %edi
177:
178: popl %esi
179: popl %ebx
180:
181: ret
182:
183: EPILOGUE()
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>