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
|
/* ************************************************************************
* Copyright 2013 Advanced Micro Devices, Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
* ************************************************************************/
static const char *nrm2_hypot_kernel = "
#ifdef DOUBLE_PRECISION
#ifdef cl_khr_fp64
#pragma OPENCL EXTENSION cl_khr_fp64 : enable
#else
#pragma OPENCL EXTENSION cl_amd_fp64 : enable
#endif
#define MIN 0x1.0p-1022 // Min in case of d/z (values from khronos site)
#else
#define MIN 0x1.0p-126f // Min in case od s/c
#endif
__kernel void %PREFIXnrm2_hypot_kernel( __global %TYPE *_X, __global %PTYPE *scratchBuff,
uint N, uint offx, int incx )
{
__global %TYPE *X = _X + offx;
#ifdef RETURN_ON_INVALID
// Incase of incx<1, NRM2 will be zero
if( get_global_id(0) == 0 ) {
scratchBuff[0] = (%PTYPE)0.0;
}
return;
#endif
int gOffset;
%TYPE%V res = (%TYPE%V) 0.0;
for( gOffset=(get_global_id(0) * %V); (gOffset + %V - 1)<N; gOffset+=( get_global_size(0) * %V ) )
{
%TYPE%V vReg1;
#ifdef INCX_NONUNITY
%VLOADWITHINCX( vReg1, (X + (gOffset*incx)), incx);
#else
vReg1 = %VLOAD( 0, (X + gOffset) );
#endif
res = hypot( res, vReg1 );
}
%TYPE nrm2 = %REDUCE_HYPOT( res );
// Loop for the last thread to handle the tail part of the vector
// Using the same gOffset used above
for( ; gOffset<N; gOffset++ )
{
%TYPE sReg1;
sReg1 = X[gOffset * incx];
nrm2 = hypot( nrm2, sReg1 );
}
// Note: this has to be called outside any if-conditions- because REDUCTION uses barrier
// dotP of work-item 0 will have the final reduced item of the work-group
%REDUCTION_BY_HYPOT( nrm2 );
%PTYPE nrm2_ptype;
#ifdef COMPLEX
nrm2_ptype = hypot( nrm2.even, nrm2.odd );
#else
nrm2_ptype = nrm2;
#endif
if( (get_local_id(0)) == 0 ) {
scratchBuff[ get_group_id(0) ] = nrm2_ptype;
}
}
\n";
static const char *nrm2_ssq_kernel = "
#ifdef DOUBLE_PRECISION
#ifdef cl_khr_fp64
#pragma OPENCL EXTENSION cl_khr_fp64 : enable
#else
#pragma OPENCL EXTENSION cl_amd_fp64 : enable
#endif
#define MAX 0x1.fffffffffffffp1023 // Max in case of d/z (values from khronos site)
#else
#define MAX 0x1.fffffep127f // Max in case of s/c
#endif
#define PZERO (%PTYPE)0.0
#define ZERO (%TYPE)0.0
#define VZERO (%TYPE%V)0.0
//
// Same scratch buffer will be used both scale and ssq.
// So a scratch buffer of size 2*N is needed.
// scale will be stored in scratch-buffer from [0] to [get_num_groups(0) - 1]
// ssq will be stored from [get_num_groups(0)] to [2*get_num_groups(0) - 1]
//
__kernel void %PREFIXnrm2_ssq_kernel( __global %TYPE *_X, __global %PTYPE *scratchBuff,
uint N, uint offx, int incx )
{
__global %TYPE *X = _X + offx;
uint numWGs = get_num_groups(0);
#ifdef RETURN_ON_INVALID
// Incase of incx<1, NRM2 will be zero
if( get_global_id(0) == 0 ) {
scratchBuff[0] = PZERO;
scratchBuff[numWGs] = PZERO;
}
return;
#endif
// First we find the max element in the whole work-group
// i.e calculating scale
%TYPE maxFound = (%TYPE) -MAX;
int gOffset;
for( gOffset=(get_global_id(0) * %V); (gOffset + %V - 1)<N; gOffset+=( get_global_size(0) * %V ) )
{
%TYPE%V vReg1;
#ifdef INCX_NONUNITY
%VLOADWITHINCX( vReg1, (X + (gOffset*incx)), incx);
#else
vReg1 = %VLOAD( 0, (X + gOffset) );
#endif
vReg1 = fabs( vReg1 );
%TYPE regMax = %REDUCE_MAX( vReg1 );
maxFound = fmax( maxFound, regMax );
}
for( ; gOffset<N; gOffset++ )
{
%TYPE sReg1;
sReg1 = X[gOffset * incx];
sReg1 = fabs( sReg1 );
maxFound = fmax( maxFound, sReg1 );
}
%REDUCTION_BY_MAX( maxFound );
__local %PTYPE _scale;
if( (get_local_id(0)) == 0 ) {
#ifdef COMPLEX
_scale = fmax( maxFound.even, maxFound.odd );
#else
_scale = maxFound;
#endif
}
barrier(CLK_LOCAL_MEM_FENCE);
// At this point we have scale.
// Now we calculate ssq by loading the array again and dividing the
// elements by scale and squaring it.
%TYPE ssq = ZERO;
%PTYPE scaleOfWG = _scale;
// If scaleOfWG was zero, that means the whole array encountered before was filled with zeroes
// Note: _scale is a local variable, either all enter or none
if(isnotequal(scaleOfWG, PZERO))
{
for( gOffset=(get_global_id(0) * %V); (gOffset + %V - 1)<N; gOffset+=( get_global_size(0) * %V ) )
{
%TYPE%V vReg1;
#ifdef INCX_NONUNITY
%VLOADWITHINCX( vReg1, (X + (gOffset*incx)), incx);
#else
vReg1 = %VLOAD( 0, (X + gOffset) );
#endif
vReg1 = fabs( vReg1 );
%TYPE%V tempSsq = (vReg1 / scaleOfWG) * (vReg1 / scaleOfWG);
ssq += %REDUCE_SUM( tempSsq );
}
for( ; gOffset<N; gOffset++ )
{
%TYPE sReg1;
sReg1 = X[gOffset * incx];
sReg1 = fabs( sReg1 );
ssq += (sReg1 / scaleOfWG) * (sReg1 / scaleOfWG);
}
%REDUCTION_BY_SUM( ssq );
}
if( (get_local_id(0)) == 0 ) {
scratchBuff[ get_group_id(0) ] = scaleOfWG;
#ifdef COMPLEX
scratchBuff[ numWGs + get_group_id(0) ] = ssq.even + ssq.odd;
#else
scratchBuff[ numWGs + get_group_id(0) ] = ssq;
#endif
}
}
\n";
|
