1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
|
/*========================== begin_copyright_notice ============================
Copyright (C) 2017-2021 Intel Corporation
SPDX-License-Identifier: MIT
============================= end_copyright_notice ===========================*/
#ifndef __BIF_64BIT_DIV_EMULATION__
#define __BIF_64BIT_DIV_EMULATION__
static INLINE ulong OVERLOADABLE __builtin_IB_emulate_udiv( ulong N, ulong D, private ulong* Rem )
{
ulong Q = 0;
ulong R = 0;
int I = 63;
while( I != -1 )
{
R = R << 1;
R = (R & ~1UL) | ( ( N >> I ) & 1UL );
if( R >= D )
{
R = R - D;
Q |= ( 1UL << I );
}
I = I - 1;
}
if (Rem)
*Rem = R;
return Q;
}
INLINE ulong OVERLOADABLE __builtin_IB_udiv( ulong N, ulong D )
{
return __builtin_IB_emulate_udiv(N, D, 0);
}
// vector versions so we can call these functions before
// scalarization occurs
INLINE ulong2 OVERLOADABLE __builtin_IB_udiv( ulong2 N, ulong2 D )
{
ulong2 result;
result.s0 = __builtin_IB_udiv( N.s0, D.s0 );
result.s1 = __builtin_IB_udiv( N.s1, D.s1 );
return result;
}
INLINE ulong3 OVERLOADABLE __builtin_IB_udiv( ulong3 N, ulong3 D )
{
ulong3 result;
result.s0 = __builtin_IB_udiv( N.s0, D.s0 );
result.s1 = __builtin_IB_udiv( N.s1, D.s1 );
result.s2 = __builtin_IB_udiv( N.s2, D.s2 );
return result;
}
INLINE ulong4 OVERLOADABLE __builtin_IB_udiv( ulong4 N, ulong4 D )
{
ulong4 result;
result.s01 = __builtin_IB_udiv( N.s01, D.s01 );
result.s23 = __builtin_IB_udiv( N.s23, D.s23 );
return result;
}
INLINE ulong8 OVERLOADABLE __builtin_IB_udiv( ulong8 N, ulong8 D )
{
ulong8 result;
result.s0123 = __builtin_IB_udiv( N.s0123, D.s0123 );
result.s4567 = __builtin_IB_udiv( N.s4567, D.s4567 );
return result;
}
INLINE ulong16 OVERLOADABLE __builtin_IB_udiv( ulong16 N, ulong16 D )
{
ulong16 result;
result.s01234567 = __builtin_IB_udiv( N.s01234567, D.s01234567 );
result.s89abcdef = __builtin_IB_udiv( N.s89abcdef, D.s89abcdef );
return result;
}
// ----------------------------------------------------------------------------
INLINE ulong OVERLOADABLE __builtin_IB_umod( ulong N, ulong D )
{
ulong R = 0;
__builtin_IB_emulate_udiv(N, D, &R);
return R;
}
// vector versions so we can call these functions before
// scalarization occurs
INLINE ulong2 OVERLOADABLE __builtin_IB_umod( ulong2 N, ulong2 D )
{
ulong2 result;
result.s0 = __builtin_IB_umod( N.s0, D.s0 );
result.s1 = __builtin_IB_umod( N.s1, D.s1 );
return result;
}
INLINE ulong3 OVERLOADABLE __builtin_IB_umod( ulong3 N, ulong3 D )
{
ulong3 result;
result.s0 = __builtin_IB_umod( N.s0, D.s0 );
result.s1 = __builtin_IB_umod( N.s1, D.s1 );
result.s2 = __builtin_IB_umod( N.s2, D.s2 );
return result;
}
INLINE ulong4 OVERLOADABLE __builtin_IB_umod( ulong4 N, ulong4 D )
{
ulong4 result;
result.s01 = __builtin_IB_umod( N.s01, D.s01 );
result.s23 = __builtin_IB_umod( N.s23, D.s23 );
return result;
}
INLINE ulong8 OVERLOADABLE __builtin_IB_umod( ulong8 N, ulong8 D )
{
ulong8 result;
result.s0123 = __builtin_IB_umod( N.s0123, D.s0123 );
result.s4567 = __builtin_IB_umod( N.s4567, D.s4567 );
return result;
}
INLINE ulong16 OVERLOADABLE __builtin_IB_umod( ulong16 N, ulong16 D )
{
ulong16 result;
result.s01234567 = __builtin_IB_umod( N.s01234567, D.s01234567 );
result.s89abcdef = __builtin_IB_umod( N.s89abcdef, D.s89abcdef );
return result;
}
// ----------------------------------------------------------------------------
INLINE long OVERLOADABLE __builtin_IB_sdiv( long N, long D )
{
ulong Q = __builtin_IB_emulate_udiv( SPIRV_OCL_BUILTIN(s_abs, _i64, )(N), SPIRV_OCL_BUILTIN(s_abs, _i64, )(D), 0 );
ulong SIGN = ( N ^ D ) & 0x8000000000000000;
if( SIGN )
{
return -Q;
}
return as_long( Q );
}
// vector versions so we can call these functions before
// scalarization occurs
INLINE long2 OVERLOADABLE __builtin_IB_sdiv( long2 N, long2 D )
{
long2 result;
result.s0 = __builtin_IB_sdiv( N.s0, D.s0 );
result.s1 = __builtin_IB_sdiv( N.s1, D.s1 );
return result;
}
INLINE long3 OVERLOADABLE __builtin_IB_sdiv( long3 N, long3 D )
{
long3 result;
result.s0 = __builtin_IB_sdiv( N.s0, D.s0 );
result.s1 = __builtin_IB_sdiv( N.s1, D.s1 );
result.s2 = __builtin_IB_sdiv( N.s2, D.s2 );
return result;
}
INLINE long4 OVERLOADABLE __builtin_IB_sdiv( long4 N, long4 D )
{
long4 result;
result.s01 = __builtin_IB_sdiv( N.s01, D.s01 );
result.s23 = __builtin_IB_sdiv( N.s23, D.s23 );
return result;
}
INLINE long8 OVERLOADABLE __builtin_IB_sdiv( long8 N, long8 D )
{
long8 result;
result.s0123 = __builtin_IB_sdiv( N.s0123, D.s0123 );
result.s4567 = __builtin_IB_sdiv( N.s4567, D.s4567 );
return result;
}
INLINE long16 OVERLOADABLE __builtin_IB_sdiv( long16 N, long16 D )
{
long16 result;
result.s01234567 = __builtin_IB_sdiv( N.s01234567, D.s01234567 );
result.s89abcdef = __builtin_IB_sdiv( N.s89abcdef, D.s89abcdef );
return result;
}
// ----------------------------------------------------------------------------
INLINE long OVERLOADABLE __builtin_IB_smod( long N, long D )
{
ulong R = 0;
__builtin_IB_emulate_udiv(SPIRV_OCL_BUILTIN(s_abs, _i64, )(N), SPIRV_OCL_BUILTIN(s_abs, _i64, )(D), &R);
uchar SIGN = (N < 0);
return SIGN ? -R : R;
}
// vector versions so we can call these functions before
// scalarization occurs
INLINE long2 OVERLOADABLE __builtin_IB_smod( long2 N, long2 D )
{
long2 result;
result.s0 = __builtin_IB_smod( N.s0, D.s0 );
result.s1 = __builtin_IB_smod( N.s1, D.s1 );
return result;
}
INLINE long3 OVERLOADABLE __builtin_IB_smod( long3 N, long3 D )
{
long3 result;
result.s0 = __builtin_IB_smod( N.s0, D.s0 );
result.s1 = __builtin_IB_smod( N.s1, D.s1 );
result.s2 = __builtin_IB_smod( N.s2, D.s2 );
return result;
}
INLINE long4 OVERLOADABLE __builtin_IB_smod( long4 N, long4 D )
{
long4 result;
result.s01 = __builtin_IB_smod( N.s01, D.s01 );
result.s23 = __builtin_IB_smod( N.s23, D.s23 );
return result;
}
INLINE long8 OVERLOADABLE __builtin_IB_smod( long8 N, long8 D )
{
long8 result;
result.s0123 = __builtin_IB_smod( N.s0123, D.s0123 );
result.s4567 = __builtin_IB_smod( N.s4567, D.s4567 );
return result;
}
INLINE long16 OVERLOADABLE __builtin_IB_smod( long16 N, long16 D )
{
long16 result;
result.s01234567 = __builtin_IB_smod( N.s01234567, D.s01234567 );
result.s89abcdef = __builtin_IB_smod( N.s89abcdef, D.s89abcdef );
return result;
}
#endif
|
