1dnl  AMD64 mpn_sublsh1_n -- rp[] = up[] - (vp[] << 1)
2
3dnl  Copyright 2003, 2005-2007, 2011, 2012 Free Software Foundation, Inc.
4
5dnl  This file is part of the GNU MP Library.
6dnl
7dnl  The GNU MP Library is free software; you can redistribute it and/or modify
8dnl  it under the terms of either:
9dnl
10dnl    * the GNU Lesser General Public License as published by the Free
11dnl      Software Foundation; either version 3 of the License, or (at your
12dnl      option) any later version.
13dnl
14dnl  or
15dnl
16dnl    * the GNU General Public License as published by the Free Software
17dnl      Foundation; either version 2 of the License, or (at your option) any
18dnl      later version.
19dnl
20dnl  or both in parallel, as here.
21dnl
22dnl  The GNU MP Library is distributed in the hope that it will be useful, but
23dnl  WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
24dnl  or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
25dnl  for more details.
26dnl
27dnl  You should have received copies of the GNU General Public License and the
28dnl  GNU Lesser General Public License along with the GNU MP Library.  If not,
29dnl  see https://www.gnu.org/licenses/.
30
31include(`../config.m4')
32
33
34C	     cycles/limb
35C AMD K8,K9	 2.2
36C AMD K10	 2.2
37C Intel P4	12.75
38C Intel core2	 3.45
39C Intel corei	 ?
40C Intel atom	 ?
41C VIA nano	 3.25
42
43C Sometimes speed degenerates, supposedly related to that some operand
44C alignments cause cache conflicts.
45
46C The speed is limited by decoding/issue bandwidth.  There are 26 instructions
47C in the loop, which corresponds to 26/3/4 = 2.167 c/l.
48
49C INPUT PARAMETERS
50define(`rp',`%rdi')
51define(`up',`%rsi')
52define(`vp',`%rdx')
53define(`n', `%rcx')
54
55ABI_SUPPORT(DOS64)
56ABI_SUPPORT(STD64)
57
58ASM_START()
59	TEXT
60	ALIGN(16)
61PROLOGUE(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
79L(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
98L(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
112L(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
119L(ent):	jns	L(end)
120
121	ALIGN(16)
122L(top):	add	R32(%rax), R32(%rax)	C restore scy
123
124	mov	(vp,n,8), %r8
125L(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
153L(end):	add	R32(%rbp), R32(%rax)
154	neg	R32(%rax)
155
156	pop	%rbp
157	pop	%rbx
158	FUNC_EXIT()
159	ret
160EPILOGUE()
161