1dnl  Alpha ev67 mpn_popcount -- mpn bit population count.
2
3dnl  Copyright 2003, 2005 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 ev67: 1.5 cycles/limb
35
36
37C unsigned long mpn_popcount (mp_srcptr src, mp_size_t size);
38C
39C This schedule seems necessary for the full 1.5 c/l, the IQ can't quite hide
40C all latencies, the addq's must be deferred to the next iteration.
41C
42C Since we need just 3 instructions per limb, further unrolling could approach
43C 1.0 c/l.
44C
45C The main loop processes two limbs at a time.  An odd size is handled by
46C processing src[0] at the start.  If the size is even that result is
47C discarded, and src[0] is repeated by the main loop.
48C
49
50ASM_START()
51PROLOGUE(mpn_popcount)
52
53	C r16	src
54	C r17	size
55
56	ldq	r0, 0(r16)		C L0  src[0]
57	and	r17, 1, r8		C U1  1 if size odd
58	srl	r17, 1, r17		C U0  size, limb pairs
59
60	s8addq	r8, r16, r16		C L1  src++ if size odd
61	ctpop	r0, r0			C U0
62	beq	r17, L(one)		C U1  if size==1
63
64	cmoveq	r8, r31, r0		C L   discard first limb if size even
65	clr	r3			C L
66
67	clr	r4			C L
68	unop				C U
69	unop				C L
70	unop				C U
71
72
73	ALIGN(16)
74L(top):
75	C r0	total accumulating
76	C r3	pop 0
77	C r4	pop 1
78	C r16	src, incrementing
79	C r17	size, decrementing
80
81	ldq	r1, 0(r16)		C L
82	ldq	r2, 8(r16)		C L
83	lda	r16, 16(r16)		C U
84	lda	r17, -1(r17)		C U
85
86	addq	r0, r3, r0		C L
87	addq	r0, r4, r0		C L
88	ctpop	r1, r3			C U0
89	ctpop	r2, r4			C U0
90
91	ldl	r31, 512(r16)		C L	prefetch
92	bne	r17, L(top)		C U
93
94
95	addq	r0, r3, r0		C L
96	addq	r0, r4, r0		C U
97L(one):
98	ret	r31, (r26), 1		C L0
99
100EPILOGUE()
101ASM_END()
102