xref: /netbsd-src/external/lgpl3/gmp/dist/mpn/ia64/copyi.asm (revision ce54336801cf28877c3414aa2fcb251dddd543a2)
1dnl  IA-64 mpn_copyi -- copy limb vector, incrementing.
2
3dnl  Contributed to the GNU project by Torbjorn Granlund.
4
5dnl  Copyright 2001, 2002, 2004 Free Software Foundation, Inc.
6
7dnl  This file is part of the GNU MP Library.
8dnl
9dnl  The GNU MP Library is free software; you can redistribute it and/or modify
10dnl  it under the terms of either:
11dnl
12dnl    * the GNU Lesser General Public License as published by the Free
13dnl      Software Foundation; either version 3 of the License, or (at your
14dnl      option) any later version.
15dnl
16dnl  or
17dnl
18dnl    * the GNU General Public License as published by the Free Software
19dnl      Foundation; either version 2 of the License, or (at your option) any
20dnl      later version.
21dnl
22dnl  or both in parallel, as here.
23dnl
24dnl  The GNU MP Library is distributed in the hope that it will be useful, but
25dnl  WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
26dnl  or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
27dnl  for more details.
28dnl
29dnl  You should have received copies of the GNU General Public License and the
30dnl  GNU Lesser General Public License along with the GNU MP Library.  If not,
31dnl  see https://www.gnu.org/licenses/.
32
33include(`../config.m4')
34
35C         cycles/limb
36C Itanium:    1
37C Itanium 2:  0.5
38
39C INPUT PARAMETERS
40C rp = r32
41C sp = r33
42C n = r34
43
44ASM_START()
45PROLOGUE(mpn_copyi)
46	.prologue
47	.save ar.lc, r2
48	.body
49ifdef(`HAVE_ABI_32',
50`	addp4		r32 = 0, r32
51	addp4		r33 = 0, r33
52	sxt4		r34 = r34
53	;;
54')
55{.mmi
56	nop		0
57	nop		0
58	mov.i		r2 = ar.lc
59}
60{.mmi
61	and		r14 = 3, r34
62	cmp.ge		p14, p15 = 3, r34
63	add		r34 = -4, r34
64	;;
65}
66{.mmi
67	cmp.eq		p8, p0 = 1, r14
68	cmp.eq		p10, p0 = 2, r14
69	cmp.eq		p12, p0 = 3, r14
70}
71{.bbb
72  (p8)	br.dptk		.Lb01
73  (p10)	br.dptk		.Lb10
74  (p12)	br.dptk		.Lb11
75}
76
77.Lb00:	C  n = 0, 4, 8, 12, ...
78  (p14)	br.dptk		.Ls00
79	;;
80	add		r21 = 8, r33
81	ld8		r16 = [r33], 16
82	shr		r15 = r34, 2
83	;;
84	ld8		r17 = [r21], 16
85	mov.i		ar.lc = r15
86	ld8		r18 = [r33], 16
87	add		r20 = 8, r32
88	;;
89	ld8		r19 = [r21], 16
90	br.cloop.dptk	.Loop
91	;;
92	br.sptk		.Lend
93	;;
94
95.Lb01:	C  n = 1, 5, 9, 13, ...
96	add		r21 = 0, r33
97	add		r20 = 0, r32
98	add		r33 = 8, r33
99	add		r32 = 8, r32
100	;;
101	ld8		r19 = [r21], 16
102	shr		r15 = r34, 2
103  (p14)	br.dptk		.Ls01
104	;;
105	ld8		r16 = [r33], 16
106	mov.i		ar.lc = r15
107	;;
108	ld8		r17 = [r21], 16
109	ld8		r18 = [r33], 16
110	br.sptk		.Li01
111	;;
112
113.Lb10:	C  n = 2,6, 10, 14, ...
114	add		r21 = 8, r33
115	add		r20 = 8, r32
116	ld8		r18 = [r33], 16
117	shr		r15 = r34, 2
118	;;
119	ld8		r19 = [r21], 16
120	mov.i		ar.lc = r15
121  (p14)	br.dptk		.Ls10
122	;;
123	ld8		r16 = [r33], 16
124	ld8		r17 = [r21], 16
125	br.sptk		.Li10
126	;;
127
128.Lb11:	C  n = 3, 7, 11, 15, ...
129	add		r21 = 0, r33
130	add		r20 = 0, r32
131	add		r33 = 8, r33
132	add		r32 = 8, r32
133	;;
134	ld8		r17 = [r21], 16
135	shr		r15 = r34, 2
136	;;
137	ld8		r18 = [r33], 16
138	mov.i		ar.lc = r15
139	ld8		r19 = [r21], 16
140  (p14)	br.dptk		.Ls11
141	;;
142	ld8		r16 = [r33], 16
143	br.sptk		.Li11
144	;;
145
146	ALIGN(32)
147.Loop:
148.Li00:
149{.mmb
150	st8		[r32] = r16, 16
151	ld8		r16 = [r33], 16
152	nop.b		0
153}
154.Li11:
155{.mmb
156	st8		[r20] = r17, 16
157	ld8		r17 = [r21], 16
158	nop.b		0
159	;;
160}
161.Li10:
162{.mmb
163	st8		[r32] = r18, 16
164	ld8		r18 = [r33], 16
165	nop.b		0
166}
167.Li01:
168{.mmb
169	st8		[r20] = r19, 16
170	ld8		r19 = [r21], 16
171	br.cloop.dptk	.Loop
172	;;
173}
174.Lend:	st8		[r32] = r16, 16
175.Ls11:	st8		[r20] = r17, 16
176	;;
177.Ls10:	st8		[r32] = r18, 16
178.Ls01:	st8		[r20] = r19, 16
179.Ls00:	mov.i		ar.lc = r2
180	br.ret.sptk.many b0
181EPILOGUE()
182ASM_END()
183