forked from fmihpc/vlasiator
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathcpu_acc_load_blocks.cpp
265 lines (253 loc) · 16.1 KB
/
cpu_acc_load_blocks.cpp
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
/*
* This file is part of Vlasiator.
* Copyright 2010-2016 Finnish Meteorological Institute,
* 2017 CSC - IT center for Science
*
* For details of usage, see the COPYING file and read the "Rules of the Road"
* at http://www.physics.helsinki.fi/vlasiator/
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License along
* with this program; if not, write to the Free Software Foundation, Inc.,
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
*/
#include "vec.h"
#include "cpu_acc_load_blocks.hpp"
void loadColumnBlockData(
const vmesh::VelocityMesh<vmesh::GlobalID,vmesh::LocalID>& vmesh,
vmesh::VelocityBlockContainer<vmesh::LocalID>& blockContainer,
vmesh::GlobalID* blocks,
vmesh::LocalID n_blocks,
const int dimension,
Vec* __restrict__ values) {
// first set the 0 values for the two empty blocks
// we store above and below the existing blocks
for (uint k=0; k<WID; ++k) {
for (uint j = 0; j < WID; j += VECL/WID){
values[i_pcolumnv(j, k, -1, n_blocks)] = Vec(0);
values[i_pcolumnv(j, k, n_blocks, n_blocks)] = Vec(0);
}
}
/*[[[cog
import cog
WID = 4
WID2 = WID * WID
for dimension in range(0, 2):
if dimension == 0:
cell_indices_to_id = [ WID2, WID, 1]
if dimension == 1:
cell_indices_to_id = [ 1, WID2, WID]
if dimension == 2:
cell_indices_to_id = [ 1, WID, WID2]
cellid_transpose=[]
for k in range(0,WID):
for j in range(0,WID):
for i in range(0,WID):
cellid_transpose.append(i * cell_indices_to_id[0] + j * cell_indices_to_id[1] + k * cell_indices_to_id[2])
cog.outl("if(dimension == %s ) {" % dimension)
cog.outl(" for (vmesh::LocalID block_k=0; block_k<n_blocks; ++block_k) {")
cog.outl(" Realf* __restrict__ data = blockContainer.getData(vmesh.getLocalID(blocks[block_k]));")
for vecl in [4, 8, 16]:
for accuracy in ["f", "d"]:
cog.outl("#ifdef VEC%d%s_AGNER" % (vecl, accuracy.upper()))
cell = 0
for k in range(0, WID):
for planeVector in range(0, WID2/vecl):
cog.out(" values[i_pcolumnv_b(%d, %d, block_k, n_blocks)] = gather%d%s<" % (planeVector, k, vecl, accuracy) )
for vi in range(0,vecl):
cog.out("%d" %cellid_transpose[cell])
if vi < vecl-1:
cog.out(" ,")
cell = cell + 1
cog.outl(">(data);")
cog.outl("#endif //VEC%d%s_AGNER" % (vecl, accuracy.upper()))
cog.outl(" //zero old output data")
cog.outl(" for (uint i=0; i<WID3; ++i) {")
cog.outl(" data[i]=0;")
cog.outl(" }")
cog.outl(" }")
cog.outl("}")
]]]*/
if(dimension == 0 ) {
for (vmesh::LocalID block_k=0; block_k<n_blocks; ++block_k) {
Realf* __restrict__ data = blockContainer.getData(vmesh.getLocalID(blocks[block_k]));
#ifdef VEC4F_AGNER
values[i_pcolumnv_b(0, 0, block_k, n_blocks)] = gather4f<0 ,16 ,32 ,48>(data);
values[i_pcolumnv_b(1, 0, block_k, n_blocks)] = gather4f<4 ,20 ,36 ,52>(data);
values[i_pcolumnv_b(2, 0, block_k, n_blocks)] = gather4f<8 ,24 ,40 ,56>(data);
values[i_pcolumnv_b(3, 0, block_k, n_blocks)] = gather4f<12 ,28 ,44 ,60>(data);
values[i_pcolumnv_b(0, 1, block_k, n_blocks)] = gather4f<1 ,17 ,33 ,49>(data);
values[i_pcolumnv_b(1, 1, block_k, n_blocks)] = gather4f<5 ,21 ,37 ,53>(data);
values[i_pcolumnv_b(2, 1, block_k, n_blocks)] = gather4f<9 ,25 ,41 ,57>(data);
values[i_pcolumnv_b(3, 1, block_k, n_blocks)] = gather4f<13 ,29 ,45 ,61>(data);
values[i_pcolumnv_b(0, 2, block_k, n_blocks)] = gather4f<2 ,18 ,34 ,50>(data);
values[i_pcolumnv_b(1, 2, block_k, n_blocks)] = gather4f<6 ,22 ,38 ,54>(data);
values[i_pcolumnv_b(2, 2, block_k, n_blocks)] = gather4f<10 ,26 ,42 ,58>(data);
values[i_pcolumnv_b(3, 2, block_k, n_blocks)] = gather4f<14 ,30 ,46 ,62>(data);
values[i_pcolumnv_b(0, 3, block_k, n_blocks)] = gather4f<3 ,19 ,35 ,51>(data);
values[i_pcolumnv_b(1, 3, block_k, n_blocks)] = gather4f<7 ,23 ,39 ,55>(data);
values[i_pcolumnv_b(2, 3, block_k, n_blocks)] = gather4f<11 ,27 ,43 ,59>(data);
values[i_pcolumnv_b(3, 3, block_k, n_blocks)] = gather4f<15 ,31 ,47 ,63>(data);
#endif //VEC4F_AGNER
#ifdef VEC4D_AGNER
values[i_pcolumnv_b(0, 0, block_k, n_blocks)] = gather4d<0 ,16 ,32 ,48>(data);
values[i_pcolumnv_b(1, 0, block_k, n_blocks)] = gather4d<4 ,20 ,36 ,52>(data);
values[i_pcolumnv_b(2, 0, block_k, n_blocks)] = gather4d<8 ,24 ,40 ,56>(data);
values[i_pcolumnv_b(3, 0, block_k, n_blocks)] = gather4d<12 ,28 ,44 ,60>(data);
values[i_pcolumnv_b(0, 1, block_k, n_blocks)] = gather4d<1 ,17 ,33 ,49>(data);
values[i_pcolumnv_b(1, 1, block_k, n_blocks)] = gather4d<5 ,21 ,37 ,53>(data);
values[i_pcolumnv_b(2, 1, block_k, n_blocks)] = gather4d<9 ,25 ,41 ,57>(data);
values[i_pcolumnv_b(3, 1, block_k, n_blocks)] = gather4d<13 ,29 ,45 ,61>(data);
values[i_pcolumnv_b(0, 2, block_k, n_blocks)] = gather4d<2 ,18 ,34 ,50>(data);
values[i_pcolumnv_b(1, 2, block_k, n_blocks)] = gather4d<6 ,22 ,38 ,54>(data);
values[i_pcolumnv_b(2, 2, block_k, n_blocks)] = gather4d<10 ,26 ,42 ,58>(data);
values[i_pcolumnv_b(3, 2, block_k, n_blocks)] = gather4d<14 ,30 ,46 ,62>(data);
values[i_pcolumnv_b(0, 3, block_k, n_blocks)] = gather4d<3 ,19 ,35 ,51>(data);
values[i_pcolumnv_b(1, 3, block_k, n_blocks)] = gather4d<7 ,23 ,39 ,55>(data);
values[i_pcolumnv_b(2, 3, block_k, n_blocks)] = gather4d<11 ,27 ,43 ,59>(data);
values[i_pcolumnv_b(3, 3, block_k, n_blocks)] = gather4d<15 ,31 ,47 ,63>(data);
#endif //VEC4D_AGNER
#ifdef VEC8F_AGNER
values[i_pcolumnv_b(0, 0, block_k, n_blocks)] = gather8f<0 ,16 ,32 ,48 ,4 ,20 ,36 ,52>(data);
values[i_pcolumnv_b(1, 0, block_k, n_blocks)] = gather8f<8 ,24 ,40 ,56 ,12 ,28 ,44 ,60>(data);
values[i_pcolumnv_b(0, 1, block_k, n_blocks)] = gather8f<1 ,17 ,33 ,49 ,5 ,21 ,37 ,53>(data);
values[i_pcolumnv_b(1, 1, block_k, n_blocks)] = gather8f<9 ,25 ,41 ,57 ,13 ,29 ,45 ,61>(data);
values[i_pcolumnv_b(0, 2, block_k, n_blocks)] = gather8f<2 ,18 ,34 ,50 ,6 ,22 ,38 ,54>(data);
values[i_pcolumnv_b(1, 2, block_k, n_blocks)] = gather8f<10 ,26 ,42 ,58 ,14 ,30 ,46 ,62>(data);
values[i_pcolumnv_b(0, 3, block_k, n_blocks)] = gather8f<3 ,19 ,35 ,51 ,7 ,23 ,39 ,55>(data);
values[i_pcolumnv_b(1, 3, block_k, n_blocks)] = gather8f<11 ,27 ,43 ,59 ,15 ,31 ,47 ,63>(data);
#endif //VEC8F_AGNER
#ifdef VEC8D_AGNER
values[i_pcolumnv_b(0, 0, block_k, n_blocks)] = gather8d<0 ,16 ,32 ,48 ,4 ,20 ,36 ,52>(data);
values[i_pcolumnv_b(1, 0, block_k, n_blocks)] = gather8d<8 ,24 ,40 ,56 ,12 ,28 ,44 ,60>(data);
values[i_pcolumnv_b(0, 1, block_k, n_blocks)] = gather8d<1 ,17 ,33 ,49 ,5 ,21 ,37 ,53>(data);
values[i_pcolumnv_b(1, 1, block_k, n_blocks)] = gather8d<9 ,25 ,41 ,57 ,13 ,29 ,45 ,61>(data);
values[i_pcolumnv_b(0, 2, block_k, n_blocks)] = gather8d<2 ,18 ,34 ,50 ,6 ,22 ,38 ,54>(data);
values[i_pcolumnv_b(1, 2, block_k, n_blocks)] = gather8d<10 ,26 ,42 ,58 ,14 ,30 ,46 ,62>(data);
values[i_pcolumnv_b(0, 3, block_k, n_blocks)] = gather8d<3 ,19 ,35 ,51 ,7 ,23 ,39 ,55>(data);
values[i_pcolumnv_b(1, 3, block_k, n_blocks)] = gather8d<11 ,27 ,43 ,59 ,15 ,31 ,47 ,63>(data);
#endif //VEC8D_AGNER
#ifdef VEC16F_AGNER
values[i_pcolumnv_b(0, 0, block_k, n_blocks)] = gather16f<0 ,16 ,32 ,48 ,4 ,20 ,36 ,52 ,8 ,24 ,40 ,56 ,12 ,28 ,44 ,60>(data);
values[i_pcolumnv_b(0, 1, block_k, n_blocks)] = gather16f<1 ,17 ,33 ,49 ,5 ,21 ,37 ,53 ,9 ,25 ,41 ,57 ,13 ,29 ,45 ,61>(data);
values[i_pcolumnv_b(0, 2, block_k, n_blocks)] = gather16f<2 ,18 ,34 ,50 ,6 ,22 ,38 ,54 ,10 ,26 ,42 ,58 ,14 ,30 ,46 ,62>(data);
values[i_pcolumnv_b(0, 3, block_k, n_blocks)] = gather16f<3 ,19 ,35 ,51 ,7 ,23 ,39 ,55 ,11 ,27 ,43 ,59 ,15 ,31 ,47 ,63>(data);
#endif //VEC16F_AGNER
#ifdef VEC16D_AGNER
values[i_pcolumnv_b(0, 0, block_k, n_blocks)] = gather16d<0 ,16 ,32 ,48 ,4 ,20 ,36 ,52 ,8 ,24 ,40 ,56 ,12 ,28 ,44 ,60>(data);
values[i_pcolumnv_b(0, 1, block_k, n_blocks)] = gather16d<1 ,17 ,33 ,49 ,5 ,21 ,37 ,53 ,9 ,25 ,41 ,57 ,13 ,29 ,45 ,61>(data);
values[i_pcolumnv_b(0, 2, block_k, n_blocks)] = gather16d<2 ,18 ,34 ,50 ,6 ,22 ,38 ,54 ,10 ,26 ,42 ,58 ,14 ,30 ,46 ,62>(data);
values[i_pcolumnv_b(0, 3, block_k, n_blocks)] = gather16d<3 ,19 ,35 ,51 ,7 ,23 ,39 ,55 ,11 ,27 ,43 ,59 ,15 ,31 ,47 ,63>(data);
#endif //VEC16D_AGNER
//zero old output data
for (uint i=0; i<WID3; ++i) {
data[i]=0;
}
}
}
if(dimension == 1 ) {
for (vmesh::LocalID block_k=0; block_k<n_blocks; ++block_k) {
Realf* __restrict__ data = blockContainer.getData(vmesh.getLocalID(blocks[block_k]));
#ifdef VEC4F_AGNER
values[i_pcolumnv_b(0, 0, block_k, n_blocks)] = gather4f<0 ,1 ,2 ,3>(data);
values[i_pcolumnv_b(1, 0, block_k, n_blocks)] = gather4f<16 ,17 ,18 ,19>(data);
values[i_pcolumnv_b(2, 0, block_k, n_blocks)] = gather4f<32 ,33 ,34 ,35>(data);
values[i_pcolumnv_b(3, 0, block_k, n_blocks)] = gather4f<48 ,49 ,50 ,51>(data);
values[i_pcolumnv_b(0, 1, block_k, n_blocks)] = gather4f<4 ,5 ,6 ,7>(data);
values[i_pcolumnv_b(1, 1, block_k, n_blocks)] = gather4f<20 ,21 ,22 ,23>(data);
values[i_pcolumnv_b(2, 1, block_k, n_blocks)] = gather4f<36 ,37 ,38 ,39>(data);
values[i_pcolumnv_b(3, 1, block_k, n_blocks)] = gather4f<52 ,53 ,54 ,55>(data);
values[i_pcolumnv_b(0, 2, block_k, n_blocks)] = gather4f<8 ,9 ,10 ,11>(data);
values[i_pcolumnv_b(1, 2, block_k, n_blocks)] = gather4f<24 ,25 ,26 ,27>(data);
values[i_pcolumnv_b(2, 2, block_k, n_blocks)] = gather4f<40 ,41 ,42 ,43>(data);
values[i_pcolumnv_b(3, 2, block_k, n_blocks)] = gather4f<56 ,57 ,58 ,59>(data);
values[i_pcolumnv_b(0, 3, block_k, n_blocks)] = gather4f<12 ,13 ,14 ,15>(data);
values[i_pcolumnv_b(1, 3, block_k, n_blocks)] = gather4f<28 ,29 ,30 ,31>(data);
values[i_pcolumnv_b(2, 3, block_k, n_blocks)] = gather4f<44 ,45 ,46 ,47>(data);
values[i_pcolumnv_b(3, 3, block_k, n_blocks)] = gather4f<60 ,61 ,62 ,63>(data);
#endif //VEC4F_AGNER
#ifdef VEC4D_AGNER
values[i_pcolumnv_b(0, 0, block_k, n_blocks)] = gather4d<0 ,1 ,2 ,3>(data);
values[i_pcolumnv_b(1, 0, block_k, n_blocks)] = gather4d<16 ,17 ,18 ,19>(data);
values[i_pcolumnv_b(2, 0, block_k, n_blocks)] = gather4d<32 ,33 ,34 ,35>(data);
values[i_pcolumnv_b(3, 0, block_k, n_blocks)] = gather4d<48 ,49 ,50 ,51>(data);
values[i_pcolumnv_b(0, 1, block_k, n_blocks)] = gather4d<4 ,5 ,6 ,7>(data);
values[i_pcolumnv_b(1, 1, block_k, n_blocks)] = gather4d<20 ,21 ,22 ,23>(data);
values[i_pcolumnv_b(2, 1, block_k, n_blocks)] = gather4d<36 ,37 ,38 ,39>(data);
values[i_pcolumnv_b(3, 1, block_k, n_blocks)] = gather4d<52 ,53 ,54 ,55>(data);
values[i_pcolumnv_b(0, 2, block_k, n_blocks)] = gather4d<8 ,9 ,10 ,11>(data);
values[i_pcolumnv_b(1, 2, block_k, n_blocks)] = gather4d<24 ,25 ,26 ,27>(data);
values[i_pcolumnv_b(2, 2, block_k, n_blocks)] = gather4d<40 ,41 ,42 ,43>(data);
values[i_pcolumnv_b(3, 2, block_k, n_blocks)] = gather4d<56 ,57 ,58 ,59>(data);
values[i_pcolumnv_b(0, 3, block_k, n_blocks)] = gather4d<12 ,13 ,14 ,15>(data);
values[i_pcolumnv_b(1, 3, block_k, n_blocks)] = gather4d<28 ,29 ,30 ,31>(data);
values[i_pcolumnv_b(2, 3, block_k, n_blocks)] = gather4d<44 ,45 ,46 ,47>(data);
values[i_pcolumnv_b(3, 3, block_k, n_blocks)] = gather4d<60 ,61 ,62 ,63>(data);
#endif //VEC4D_AGNER
#ifdef VEC8F_AGNER
values[i_pcolumnv_b(0, 0, block_k, n_blocks)] = gather8f<0 ,1 ,2 ,3 ,16 ,17 ,18 ,19>(data);
values[i_pcolumnv_b(1, 0, block_k, n_blocks)] = gather8f<32 ,33 ,34 ,35 ,48 ,49 ,50 ,51>(data);
values[i_pcolumnv_b(0, 1, block_k, n_blocks)] = gather8f<4 ,5 ,6 ,7 ,20 ,21 ,22 ,23>(data);
values[i_pcolumnv_b(1, 1, block_k, n_blocks)] = gather8f<36 ,37 ,38 ,39 ,52 ,53 ,54 ,55>(data);
values[i_pcolumnv_b(0, 2, block_k, n_blocks)] = gather8f<8 ,9 ,10 ,11 ,24 ,25 ,26 ,27>(data);
values[i_pcolumnv_b(1, 2, block_k, n_blocks)] = gather8f<40 ,41 ,42 ,43 ,56 ,57 ,58 ,59>(data);
values[i_pcolumnv_b(0, 3, block_k, n_blocks)] = gather8f<12 ,13 ,14 ,15 ,28 ,29 ,30 ,31>(data);
values[i_pcolumnv_b(1, 3, block_k, n_blocks)] = gather8f<44 ,45 ,46 ,47 ,60 ,61 ,62 ,63>(data);
#endif //VEC8F_AGNER
#ifdef VEC8D_AGNER
values[i_pcolumnv_b(0, 0, block_k, n_blocks)] = gather8d<0 ,1 ,2 ,3 ,16 ,17 ,18 ,19>(data);
values[i_pcolumnv_b(1, 0, block_k, n_blocks)] = gather8d<32 ,33 ,34 ,35 ,48 ,49 ,50 ,51>(data);
values[i_pcolumnv_b(0, 1, block_k, n_blocks)] = gather8d<4 ,5 ,6 ,7 ,20 ,21 ,22 ,23>(data);
values[i_pcolumnv_b(1, 1, block_k, n_blocks)] = gather8d<36 ,37 ,38 ,39 ,52 ,53 ,54 ,55>(data);
values[i_pcolumnv_b(0, 2, block_k, n_blocks)] = gather8d<8 ,9 ,10 ,11 ,24 ,25 ,26 ,27>(data);
values[i_pcolumnv_b(1, 2, block_k, n_blocks)] = gather8d<40 ,41 ,42 ,43 ,56 ,57 ,58 ,59>(data);
values[i_pcolumnv_b(0, 3, block_k, n_blocks)] = gather8d<12 ,13 ,14 ,15 ,28 ,29 ,30 ,31>(data);
values[i_pcolumnv_b(1, 3, block_k, n_blocks)] = gather8d<44 ,45 ,46 ,47 ,60 ,61 ,62 ,63>(data);
#endif //VEC8D_AGNER
#ifdef VEC16F_AGNER
values[i_pcolumnv_b(0, 0, block_k, n_blocks)] = gather16f<0 ,1 ,2 ,3 ,16 ,17 ,18 ,19 ,32 ,33 ,34 ,35 ,48 ,49 ,50 ,51>(data);
values[i_pcolumnv_b(0, 1, block_k, n_blocks)] = gather16f<4 ,5 ,6 ,7 ,20 ,21 ,22 ,23 ,36 ,37 ,38 ,39 ,52 ,53 ,54 ,55>(data);
values[i_pcolumnv_b(0, 2, block_k, n_blocks)] = gather16f<8 ,9 ,10 ,11 ,24 ,25 ,26 ,27 ,40 ,41 ,42 ,43 ,56 ,57 ,58 ,59>(data);
values[i_pcolumnv_b(0, 3, block_k, n_blocks)] = gather16f<12 ,13 ,14 ,15 ,28 ,29 ,30 ,31 ,44 ,45 ,46 ,47 ,60 ,61 ,62 ,63>(data);
#endif //VEC16F_AGNER
#ifdef VEC16D_AGNER
values[i_pcolumnv_b(0, 0, block_k, n_blocks)] = gather16d<0 ,1 ,2 ,3 ,16 ,17 ,18 ,19 ,32 ,33 ,34 ,35 ,48 ,49 ,50 ,51>(data);
values[i_pcolumnv_b(0, 1, block_k, n_blocks)] = gather16d<4 ,5 ,6 ,7 ,20 ,21 ,22 ,23 ,36 ,37 ,38 ,39 ,52 ,53 ,54 ,55>(data);
values[i_pcolumnv_b(0, 2, block_k, n_blocks)] = gather16d<8 ,9 ,10 ,11 ,24 ,25 ,26 ,27 ,40 ,41 ,42 ,43 ,56 ,57 ,58 ,59>(data);
values[i_pcolumnv_b(0, 3, block_k, n_blocks)] = gather16d<12 ,13 ,14 ,15 ,28 ,29 ,30 ,31 ,44 ,45 ,46 ,47 ,60 ,61 ,62 ,63>(data);
#endif //VEC16D_AGNER
//zero old output data
for (uint i=0; i<WID3; ++i) {
data[i]=0;
}
}
}
//[[[end]]]
if (dimension == 2) {
// copy block data for all blocks. Dimension 2 is easy, here
// data is in the right order
for (vmesh::LocalID block_k=0; block_k<n_blocks; ++block_k) {
Realf* __restrict__ data = blockContainer.getData(vmesh.getLocalID(blocks[block_k]));
uint offset = 0;
for (uint k=0; k<WID; ++k) {
for(uint planeVector = 0; planeVector < VEC_PER_PLANE; planeVector++){
values[i_pcolumnv_b(planeVector, k, block_k, n_blocks)].load(data + offset);
offset += VECL;
}
}
//zero old output data
for (uint i=0; i<WID3; ++i) {
data[i]=0;
}
}
}
}