![[BACK]](/icons/cvsweb/back.gif){kind=icon}
Return to addmul_1.S CVS log ![[TXT]](/icons/cvsweb/text.gif){kind=icon}
![[DIR]](/icons/cvsweb/dir.gif){kind=icon}
Up to [local] / OpenXM_contrib / gmp / mpn / pa64|
Return to addmul_1.S CVS log | Up to [local] / OpenXM_contrib / gmp / mpn / pa64 |
1.1 maekawa 1: ; HP-PA 2.0 64-bit __gmpn_addmul_1 -- Multiply a limb vector with a limb and
2: ; add the result to a second limb vector.
3:
4: ; Copyright (C) 1998, 1999, 2000 Free Software Foundation, Inc.
5:
6: ; This file is part of the GNU MP Library.
7:
8: ; The GNU MP Library is free software; you can redistribute it and/or modify
9: ; it under the terms of the GNU Lesser General Public License as published by
10: ; the Free Software Foundation; either version 2.1 of the License, or (at your
11: ; option) any later version.
12:
13: ; The GNU MP Library is distributed in the hope that it will be useful, but
14: ; WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
15: ; or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public
16: ; License for more details.
17:
18: ; You should have received a copy of the GNU Lesser General Public License
19: ; along with the GNU MP Library; see the file COPYING.LIB. If not, write to
20: ; the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
21: ; MA 02111-1307, USA.
22:
23: ; INPUT PARAMETERS
24: #define rptr %r26
25: #define sptr %r25
26: #define size %r24
27: #define s2limb -56(%r30)
28:
29: ; This runs at 11 cycles/limb on a PA8000. It might be possible to make
30: ; it faster, but the PA8000 pipeline is not publically documented and it
31: ; is very complex to reverse engineer
32:
33: #define t1 %r19
34: #define rlimb %r20
35: #define hi %r21
36: #define lo %r22
37: #define m0 %r28
38: #define m1 %r3
39: #define cylimb %r29
40: #define t3 %r4
41: #define t2 %r6
42: #define t5 %r23
43: #define t4 %r31
44: .level 2.0n
45: .code
46: .export __gmpn_addmul_1,entry
47: __gmpn_addmul_1
48: .proc
49: .callinfo frame=128,no_calls
50: .entry
51: fldd -56(%r30),%fr5 ; s2limb passed on stack
52: ldo 128(%r30),%r30
53: add %r0,%r0,cylimb ; clear cy and cylimb
54:
55: std %r3,-96(%r30)
56: std %r4,-88(%r30)
57: std %r5,-80(%r30)
58: std %r6,-72(%r30)
59: depdi,z 1,31,1,%r5
60:
61: fldd 0(sptr),%fr4
62: ldo 8(sptr),sptr
63:
64: xmpyu %fr5R,%fr4R,%fr6
65: fstd %fr6,-128(%r30)
66: xmpyu %fr5R,%fr4L,%fr7
67: fstd %fr7,-120(%r30)
68: xmpyu %fr5L,%fr4R,%fr8
69: fstd %fr8,-112(%r30)
70: xmpyu %fr5L,%fr4L,%fr9
71: fstd %fr9,-104(%r30)
72: ldd -128(%r30),lo ; lo = low 64 bit of product
73: ldd -120(%r30),m0 ; m0 = mid0 64 bit of product
74: ldd -112(%r30),m1 ; m1 = mid1 64 bit of product
75: ldd -104(%r30),hi ; hi = high 64 bit of product
76: addib,= -1,%r24,L$end1
77: nop
78: fldd 0(sptr),%fr4
79: ldo 8(sptr),sptr
80: addib,= -1,%r24,L$end2
81: nop
82: L$loop
83: xmpyu %fr5R,%fr4R,%fr6
84: fstd %fr6,-128(%r30)
85: xmpyu %fr5R,%fr4L,%fr7
86: fstd %fr7,-120(%r30)
87: xmpyu %fr5L,%fr4R,%fr8
88: fstd %fr8,-112(%r30)
89: xmpyu %fr5L,%fr4L,%fr9
90: fstd %fr9,-104(%r30)
91: ldd 0(rptr),rlimb
92: extrd,u lo,31,32,t1 ; t1 = hi32(lo)
93: extrd,u lo,63,32,t4 ; t4 = lo32(lo)
94: add,l m0,t1,t1 ; t1 += m0
95: add,l,*nuv m1,t1,t1 ; t1 += m1
96: add,l %r5,hi,hi ; propagate carry
97: extrd,u t1,31,32,t2 ; t2 = hi32(t1)
98: depd,z t1,31,32,t5 ; t5 = lo32(t1)
99: add,l t5,t4,t4 ; t4 += lo32(t1)
100: ldd -128(%r30),lo ; lo = low 64 bit of product
101: add cylimb,rlimb,rlimb
102: ldd -120(%r30),m0 ; m0 = mid0 64 bit of product
103: add,dc t2,hi,cylimb
104: ldd -112(%r30),m1 ; m1 = mid1 64 bit of product
105: add t4,rlimb,t3
106: ldd -104(%r30),hi ; hi = high 64 bit of product
107: add,dc %r0,cylimb,cylimb
108: fldd 0(sptr),%fr4
109: ldo 8(sptr),sptr
110: std t3,0(rptr)
111: addib,<> -1,%r24,L$loop
112: ldo 8(rptr),rptr
113: L$end2
114: xmpyu %fr5R,%fr4R,%fr6
115: fstd %fr6,-128(%r30)
116: xmpyu %fr5R,%fr4L,%fr7
117: fstd %fr7,-120(%r30)
118: xmpyu %fr5L,%fr4R,%fr8
119: fstd %fr8,-112(%r30)
120: xmpyu %fr5L,%fr4L,%fr9
121: fstd %fr9,-104(%r30)
122: ldd 0(rptr),rlimb
123: extrd,u lo,31,32,t1 ; t1 = hi32(lo)
124: extrd,u lo,63,32,t4 ; t4 = lo32(lo)
125: add,l m0,t1,t1 ; t1 += m0
126: add,l,*nuv m1,t1,t1 ; t1 += m0
127: add,l %r5,hi,hi ; propagate carry
128: extrd,u t1,31,32,t2 ; t2 = hi32(t1)
129: depd,z t1,31,32,t5 ; t5 = lo32(t1)
130: add,l t5,t4,t4 ; t4 += lo32(t1)
131: ldd -128(%r30),lo ; lo = low 64 bit of product
132: add cylimb,rlimb,rlimb
133: ldd -120(%r30),m0 ; m0 = mid0 64 bit of product
134: add,dc t2,hi,cylimb
135: ldd -112(%r30),m1 ; m1 = mid1 64 bit of product
136: add t4,rlimb,t3
137: ldd -104(%r30),hi ; hi = high 64 bit of product
138: add,dc %r0,cylimb,cylimb
139: std t3,0(rptr)
140: ldo 8(rptr),rptr
141: L$end1
142: ldd 0(rptr),rlimb
143: extrd,u lo,31,32,t1 ; t1 = hi32(lo)
144: extrd,u lo,63,32,t4 ; t4 = lo32(lo)
145: add,l m0,t1,t1 ; t1 += m0
146: add,l,*nuv m1,t1,t1 ; t1 += m0
147: add,l %r5,hi,hi ; propagate carry
148: extrd,u t1,31,32,t2 ; t2 = hi32(t1)
149: depd,z t1,31,32,t5 ; t5 = lo32(t1)
150: add,l t5,t4,t4 ; t4 += lo32(t1)
151: add cylimb,rlimb,rlimb
152: add,dc t2,hi,cylimb
153: add t4,rlimb,t3
154: add,dc %r0,cylimb,cylimb
155: std t3,0(rptr)
156: ldo 8(rptr),rptr
157:
158: ldd -96(%r30),%r3
159: ldd -88(%r30),%r4
160: ldd -80(%r30),%r5
161: ldd -72(%r30),%r6
162:
163: extrd,u cylimb,31,32,%r28
164: bve (%r2)
165: .exit
166: ldo -128(%r30),%r30
167: .procend