質問編集履歴
6
修正
test
CHANGED
|
File without changes
|
test
CHANGED
|
@@ -8,37 +8,37 @@
|
|
|
8
8
|
|
|
9
9
|
正直自分ではもうどこが間違っているのかわからないので、質問させていただきました。おそらく吸収境界条件が間違っていると考えているのですが。
|
|
10
10
|
|
|
11
|
-
動画ではなく、画像ですがいくつか掲載させていただきます。
|
|
11
|
+
動画ではなく、画像ですがいくつか掲載させていただきます。（プログラムのdt(電界計算＋磁界計算時間)は2e-12s）
|
|
12
12
|
|
|
13
13
|
|
|
14
14
|
|
|
15
15
|
|
|
16
16
|
|
|
17
|
-
___________________t=0.8e-10
|
|
17
|
+
___________________t=0.8e-10s
|
|
18
18
|
|
|
19
19
|
|
|
20
20
|
|
|
21
|
-
___________________t=3.0e-10
|
|
21
|
+
___________________t=3.0e-10s
|
|
22
22
|
|
|
23
23
|
|
|
24
24
|
|
|
25
|
-
___________________t=3.2e-10
|
|
25
|
+
___________________t=3.2e-10s
|
|
26
26
|
|
|
27
27
|
|
|
28
28
|
|
|
29
|
-
___________________t=3.4e-10
|
|
29
|
+
___________________t=3.4e-10s
|
|
30
30
|
|
|
31
31
|
|
|
32
32
|
|
|
33
|
-
___________________t=3.6e-10
|
|
33
|
+
___________________t=3.6e-10s
|
|
34
34
|
|
|
35
35
|
|
|
36
36
|
|
|
37
|
-
___________________t=4.0e-10
|
|
37
|
+
___________________t=4.0e-10s
|
|
38
38
|
|
|
39
39
|
|
|
40
40
|
|
|
41
|
-
___________________t=2.0e-9
|
|
41
|
+
___________________t=2.0e-9s
|
|
42
42
|
|
|
43
43
|
|
|
44
44
|
|
5
画像追加
test
CHANGED
|
File without changes
|
test
CHANGED
|
@@ -32,10 +32,16 @@
|
|
|
32
32
|
|
|
33
33
|
___________________t=3.6e-10
|
|
34
34
|
|
|
35
|
-
|
|
35
|
+
|
|
36
36
|
|
|
37
37
|
___________________t=4.0e-10
|
|
38
38
|
|
|
39
|
+
|
|
40
|
+
|
|
41
|
+
___________________t=2.0e-9
|
|
42
|
+
|
|
43
|
+
|
|
44
|
+
|
|
39
45
|
### 該当のソースコード
|
|
40
46
|
|
|
41
47
|
```
|
4
修正
test
CHANGED
|
File without changes
|
test
CHANGED
|
@@ -22,19 +22,19 @@
|
|
|
22
22
|
|
|
23
23
|
|
|
24
24
|
|
|
25
|
-
___________________t=3.2e-1
|
|
25
|
+
___________________t=3.2e-10
|
|
26
26
|
|
|
27
27
|
|
|
28
28
|
|
|
29
|
-
___________________t=3.4e-1
|
|
29
|
+
___________________t=3.4e-10
|
|
30
30
|
|
|
31
31
|
|
|
32
32
|
|
|
33
|
-
___________________t=3.6e-1
|
|
33
|
+
___________________t=3.6e-10
|
|
34
34
|
|
|
35
35
|
](ca236bbb256437b08d877e028fca9a4f.png)
|
|
36
36
|
|
|
37
|
-
___________________t=40e-1
|
|
37
|
+
___________________t=4.0e-10
|
|
38
38
|
|
|
39
39
|
### 該当のソースコード
|
|
40
40
|
|
3
画像の追加
test
CHANGED
|
File without changes
|
test
CHANGED
|
@@ -8,9 +8,33 @@
|
|
|
8
8
|
|
|
9
9
|
正直自分ではもうどこが間違っているのかわからないので、質問させていただきました。おそらく吸収境界条件が間違っていると考えているのですが。
|
|
10
10
|
|
|
11
|
-
|
|
11
|
+
動画ではなく、画像ですがいくつか掲載させていただきます。
|
|
12
|
+
|
|
13
|
+
|
|
14
|
+
|
|
12
|
-
|
|
15
|
+
|
|
16
|
+
|
|
13
|
-
|
|
17
|
+
___________________t=0.8e-10
|
|
18
|
+
|
|
19
|
+
|
|
20
|
+
|
|
21
|
+
___________________t=3.0e-10
|
|
22
|
+
|
|
23
|
+
|
|
24
|
+
|
|
25
|
+
___________________t=3.2e-11
|
|
26
|
+
|
|
27
|
+
|
|
28
|
+
|
|
29
|
+
___________________t=3.4e-11
|
|
30
|
+
|
|
31
|
+
|
|
32
|
+
|
|
33
|
+
___________________t=3.6e-11
|
|
34
|
+
|
|
35
|
+
](ca236bbb256437b08d877e028fca9a4f.png)
|
|
36
|
+
|
|
37
|
+
___________________t=40e-11
|
|
14
38
|
|
|
15
39
|
### 該当のソースコード
|
|
16
40
|
|
2
修正
test
CHANGED
|
File without changes
|
test
CHANGED
|
@@ -2,434 +2,424 @@
|
|
|
2
2
|
|
|
3
3
|
|
|
4
4
|
|
|
5
|
-
|
|
5
|
+
|
|
6
6
|
|
|
7
7
|
NVIDIA社のcudaをvisualsudio2015用いて3次元FDTD法を練習しています。今回、サイズ41×41×41の立方体中のx=20,y=20の位置における0<z<41のEzに電圧を常時印加した場合に、z=20におけるxy平面のEzの値を見ているのですが、普通にCでこのプログラムを作った場合はきれいに最後まで円形の画像が出力され続けるのですが、cudaでは出力画像でx=0とy=0付近においてmurの1次吸収境界条件の効きが悪く、それが原因で発散しているような状況となっています。（吸収境界条件を入れずに完全導体で囲んだ場合はきれいに反射された画像が続く）
|
|
8
8
|
|
|
9
9
|
正直自分ではもうどこが間違っているのかわからないので、質問させていただきました。おそらく吸収境界条件が間違っていると考えているのですが。
|
|
10
10
|
|
|
11
|
-
|
|
11
|
+
|
|
12
|
-
|
|
13
|
-
|
|
14
|
-
|
|
12
|
+
|
|
13
|
+
|
|
14
|
+
|
|
15
|
-
###
|
|
15
|
+
### 該当のソースコード
|
|
16
|
-
|
|
17
|
-
|
|
18
16
|
|
|
19
17
|
```
|
|
20
18
|
|
|
21
|
-
|
|
19
|
+
/*--------------------------------------------メイン計算部分------------------------------------------------------------*/
|
|
20
|
+
|
|
21
|
+
while (t[0]<2e-9) {
|
|
22
|
+
|
|
23
|
+
|
|
24
|
+
|
|
25
|
+
culcEx << <grid, block >> > (Ex_d, oldEx_d, oldHy_d, oldHz_d, A_d, d_d);
|
|
26
|
+
|
|
27
|
+
culcEy << <grid, block >> > (Ey_d, oldEy_d, oldHx_d, oldHz_d, A_d, d_d);
|
|
28
|
+
|
|
29
|
+
culcEz << <grid, block >> > (Ez_d, oldEz_d, oldHx_d, oldHy_d, A_d, d_d);
|
|
30
|
+
|
|
31
|
+
cudaDeviceSynchronize();
|
|
32
|
+
|
|
33
|
+
|
|
34
|
+
|
|
35
|
+
|
|
36
|
+
|
|
37
|
+
vin[0] = 1.0*sin(2.0e10*M_PI*t[0]);
|
|
38
|
+
|
|
39
|
+
cudaMemcpy(vin_d, vin, sizeof(double), cudaMemcpyHostToDevice);
|
|
40
|
+
|
|
41
|
+
Init << <grid, block >> > ( Ez_d, t_d, vin_d);
|
|
42
|
+
|
|
43
|
+
cudaDeviceSynchronize();
|
|
44
|
+
|
|
45
|
+
Init2 << <grid, block >> > (Ex_d, Ey_d, Ez_d);
|
|
46
|
+
|
|
47
|
+
cudaDeviceSynchronize();
|
|
48
|
+
|
|
49
|
+
MurEx << <grid, block >> > (Ex_d, oldEx_d, A_d);
|
|
50
|
+
|
|
51
|
+
MurEy << <grid, block >> > (Ey_d, oldEy_d, A_d);
|
|
52
|
+
|
|
53
|
+
MurEz << <grid, block >> > (Ez_d, oldEz_d, A_d);
|
|
54
|
+
|
|
55
|
+
cudaDeviceSynchronize();
|
|
56
|
+
|
|
57
|
+
|
|
58
|
+
|
|
59
|
+
|
|
60
|
+
|
|
61
|
+
culct << <2,2 >> > (t_d);
|
|
62
|
+
|
|
63
|
+
t[0] += (dt / 2);
|
|
64
|
+
|
|
65
|
+
cudaDeviceSynchronize();
|
|
66
|
+
|
|
67
|
+
|
|
68
|
+
|
|
69
|
+
culcHx << <grid, block >> > (Ey_d, Ez_d, Hx_d, oldHx_d, A_d, d_d);
|
|
70
|
+
|
|
71
|
+
culcHy << <grid, block >> > (Ex_d, Ez_d, Hy_d, oldHy_d, A_d, d_d);
|
|
72
|
+
|
|
73
|
+
culcHz << <grid, block >> > (Ex_d, Ey_d, Hz_d, oldHz_d, A_d, d_d);
|
|
74
|
+
|
|
75
|
+
cudaDeviceSynchronize();
|
|
76
|
+
|
|
77
|
+
|
|
78
|
+
|
|
79
|
+
change << <grid, block >> > (Ex_d, Ey_d, Ez_d, oldEx_d, oldEy_d, oldEz_d, Hx_d, Hy_d, Hz_d, oldHx_d, oldHy_d, oldHz_d);
|
|
80
|
+
|
|
81
|
+
cudaDeviceSynchronize();
|
|
82
|
+
|
|
83
|
+
|
|
84
|
+
|
|
85
|
+
culct << <2, 2 >> > (t_d);
|
|
86
|
+
|
|
87
|
+
t[0] += (dt / 2);
|
|
88
|
+
|
|
89
|
+
cudaDeviceSynchronize();
|
|
90
|
+
|
|
91
|
+
output << <grid, block >> > (out_d, Ez_d);
|
|
92
|
+
|
|
93
|
+
cudaDeviceSynchronize();
|
|
94
|
+
|
|
95
|
+
|
|
96
|
+
|
|
97
|
+
}
|
|
98
|
+
|
|
99
|
+
/*---------------------------------------------------------------------------------------------------------------------*/
|
|
100
|
+
|
|
101
|
+
|
|
102
|
+
|
|
103
|
+
/*------------------吸収境界条件部分----------------------------*/
|
|
104
|
+
|
|
105
|
+
__global__ void MurEx(double *Ex, double *oldEx, double *A){
|
|
106
|
+
|
|
107
|
+
int i = blockIdx.x*blockDim.x + threadIdx.x;
|
|
108
|
+
|
|
109
|
+
int j = blockIdx.y*blockDim.y + threadIdx.y;
|
|
110
|
+
|
|
111
|
+
int k = blockIdx.z*blockDim.z + threadIdx.z;
|
|
112
|
+
|
|
113
|
+
|
|
114
|
+
|
|
115
|
+
if (i < nx && j < ny && k < nz) {
|
|
116
|
+
|
|
117
|
+
Ex[(k * ny * nx + j * nx + 0)] =
|
|
118
|
+
|
|
119
|
+
oldEx[(k * ny * nx + j * nx + 1)] + A[3] * (Ex[(k * ny * nx + j * nx + 1)] - oldEx[(k * ny * nx + j * nx + 0)]);
|
|
120
|
+
|
|
121
|
+
Ex[(k * ny * nx + j * nx + nx - 1)] = oldEx[(k * ny * nx + j * nx + nx - 2)] + A[3] * (Ex[(k * ny * nx + j * nx + nx - 2)] - oldEx[(k * ny * nx + j * nx + nx - 1)]);
|
|
122
|
+
|
|
123
|
+
}
|
|
124
|
+
|
|
125
|
+
__syncthreads();
|
|
126
|
+
|
|
127
|
+
}
|
|
128
|
+
|
|
129
|
+
|
|
130
|
+
|
|
131
|
+
__global__ void MurEy( double *Ey, double *oldEy, double *A){
|
|
132
|
+
|
|
133
|
+
int i = blockIdx.x*blockDim.x + threadIdx.x;
|
|
134
|
+
|
|
135
|
+
int j = blockIdx.y*blockDim.y + threadIdx.y;
|
|
136
|
+
|
|
137
|
+
int k = blockIdx.z*blockDim.z + threadIdx.z;
|
|
138
|
+
|
|
139
|
+
|
|
140
|
+
|
|
141
|
+
if (i < nx && j < ny && k < nz) {
|
|
142
|
+
|
|
143
|
+
Ey[(k * ny * nx + 0 * nx + i)] =
|
|
144
|
+
|
|
145
|
+
oldEy[(k * ny * nx + 1 * nx + i)] + A[3] * (Ey[(k * ny * nx + 1 * nx + i)] - oldEy[(k * ny * nx + 0 * nx + i)]);
|
|
146
|
+
|
|
147
|
+
Ey[(k * ny * nx + (ny - 1) * nx + i)] =
|
|
148
|
+
|
|
149
|
+
oldEy[(k * ny * nx + (ny - 2) * nx + i)] + A[3] * (Ey[(k * ny * nx + (ny - 2) * nx + i)] - oldEy[(k * ny * nx + (ny - 1) * nx + i)]);
|
|
150
|
+
|
|
151
|
+
}
|
|
152
|
+
|
|
153
|
+
__syncthreads();
|
|
154
|
+
|
|
155
|
+
}
|
|
156
|
+
|
|
157
|
+
|
|
158
|
+
|
|
159
|
+
__global__ void MurEz( double *Ez, double *oldEz, double *A){
|
|
160
|
+
|
|
161
|
+
int i = blockIdx.x*blockDim.x + threadIdx.x;
|
|
162
|
+
|
|
163
|
+
int j = blockIdx.y*blockDim.y + threadIdx.y;
|
|
164
|
+
|
|
165
|
+
int k = blockIdx.z*blockDim.z + threadIdx.z;
|
|
166
|
+
|
|
167
|
+
|
|
168
|
+
|
|
169
|
+
if (i < nx && j < ny && k < nz) {
|
|
170
|
+
|
|
171
|
+
Ez[(k * ny * nx + j * nx + 0)] =
|
|
172
|
+
|
|
173
|
+
oldEz[(k * ny * nx + j * nx + 1)] + A[3] * (Ez[(k * ny * nx + j * nx + 1)] - oldEz[(k * ny * nx + j * nx + 0)]);
|
|
174
|
+
|
|
175
|
+
Ez[(k * ny * nx + j * nx + nx - 1)] =
|
|
176
|
+
|
|
177
|
+
oldEz[(k * ny * nx + j * nx + nx - 2)] + A[3] * (Ez[(k * ny * nx + j * nx + nx - 2)] - oldEz[(k * ny * nx + j * nx + nx - 1)]);
|
|
178
|
+
|
|
179
|
+
|
|
180
|
+
|
|
181
|
+
Ez[(k * ny * nx + 0 * nx + i)] =
|
|
182
|
+
|
|
183
|
+
oldEz[(k * ny * nx + 1 * nx + i)] + A[3] * (Ez[(k * ny * nx + 1 * nx + i)] - oldEz[(k * ny * nx + 0 * nx + i)]);
|
|
184
|
+
|
|
185
|
+
Ez[(k * ny * nx + (ny - 1) * nx + i)]
|
|
186
|
+
|
|
187
|
+
= oldEz[(k * ny * nx + (ny - 2) * nx + i)] + A[3] * (Ez[(k * ny * nx + (ny - 2) * nx + i)] - oldEz[(k * ny * nx + (ny - 1) * nx + i)]);
|
|
188
|
+
|
|
189
|
+
}
|
|
190
|
+
|
|
191
|
+
__syncthreads();
|
|
192
|
+
|
|
193
|
+
}
|
|
194
|
+
|
|
195
|
+
/*------------------------------------------------------------------------------*/
|
|
196
|
+
|
|
197
|
+
//その他補足（プログラムより抜粋)
|
|
198
|
+
|
|
199
|
+
♯define nx 41
|
|
200
|
+
|
|
201
|
+
♯define ny 41
|
|
202
|
+
|
|
203
|
+
♯define nz 41
|
|
204
|
+
|
|
205
|
+
|
|
206
|
+
|
|
207
|
+
dim3 grid;
|
|
208
|
+
|
|
209
|
+
grid.x = 3;
|
|
210
|
+
|
|
211
|
+
grid.y = 3;
|
|
212
|
+
|
|
213
|
+
grid.z = 24;
|
|
214
|
+
|
|
215
|
+
dim3 block;
|
|
216
|
+
|
|
217
|
+
block.x = 16;
|
|
218
|
+
|
|
219
|
+
block.y = 16;
|
|
220
|
+
|
|
221
|
+
block.z = 2;
|
|
222
|
+
|
|
223
|
+
/*--------------------------電磁界計算用係数設定------------------------*/
|
|
224
|
+
|
|
225
|
+
A[0] = ((1 - (sig*dt) / (2 * ep)) / (1 + (sig*dt) / (2 * ep)));
|
|
226
|
+
|
|
227
|
+
A[1] = (dt / ep) / (1 + (sig*dt / (2 * ep)));
|
|
228
|
+
|
|
229
|
+
A[2] = dt / mu;
|
|
230
|
+
|
|
231
|
+
A[3] = (c*dt - dx) / (c*dt + dx);
|
|
232
|
+
|
|
233
|
+
/*-----------------------------------------------
|
|
234
|
+
|
|
235
|
+
0:電界計算,1:電界計算,2:磁界計算,3:Mur吸収境界条件
|
|
236
|
+
|
|
237
|
+
-------------------------------------------------*/
|
|
238
|
+
|
|
239
|
+
d[0] = 2.0e-12;
|
|
240
|
+
|
|
241
|
+
d[1] = 1.5e-3;
|
|
242
|
+
|
|
243
|
+
d[2] = 1.5e-3;
|
|
244
|
+
|
|
245
|
+
d[3] = 1.5e-3;
|
|
246
|
+
|
|
247
|
+
/*-----------------------------------------------
|
|
248
|
+
|
|
249
|
+
0:dt,1:dx,2:dy,3:dz
|
|
250
|
+
|
|
251
|
+
-------------------------------------------------*/
|
|
252
|
+
|
|
253
|
+
/*----------------------------電界計算----------------------*/
|
|
254
|
+
|
|
255
|
+
__global__ void culcEx(double *Ex, double *oldEx, double *oldHy, double *oldHz, double *A, double *d)
|
|
256
|
+
|
|
257
|
+
{
|
|
258
|
+
|
|
259
|
+
|
|
260
|
+
|
|
261
|
+
int i = blockIdx.x*blockDim.x + threadIdx.x;
|
|
262
|
+
|
|
263
|
+
int j = blockIdx.y*blockDim.y + threadIdx.y+1;
|
|
264
|
+
|
|
265
|
+
int k = blockIdx.z*blockDim.z + threadIdx.z+1;
|
|
266
|
+
|
|
267
|
+
|
|
268
|
+
|
|
269
|
+
if ((0 < j) && (0 < k) && (i < nx) && (j < ny) && (k < nz)) {
|
|
270
|
+
|
|
271
|
+
Ex[(k * ny * nx + j * nx + i)] = A[0] * oldEx[(k * ny * nx + j * nx + i)] + A[1] * (((oldHz[(k * ny * nx + j * nx + i)] - oldHz[(k * ny * nx + (j - 1) * nx + i)]) / d[2]) - ((oldHy[(k * ny * nx + j * nx + i)] - oldHy[((k - 1) * ny * nx + j * nx + i)]) / d[3])); //4
|
|
272
|
+
|
|
273
|
+
}
|
|
274
|
+
|
|
275
|
+
__syncthreads();
|
|
276
|
+
|
|
277
|
+
|
|
278
|
+
|
|
279
|
+
}
|
|
280
|
+
|
|
281
|
+
|
|
282
|
+
|
|
283
|
+
__global__ void culcEy(double *Ey, double *oldEy, double *oldHx, double *oldHz, double *A, double *d)
|
|
284
|
+
|
|
285
|
+
{
|
|
286
|
+
|
|
287
|
+
int i = blockIdx.x*blockDim.x + threadIdx.x+1;
|
|
288
|
+
|
|
289
|
+
int j = blockIdx.y*blockDim.y + threadIdx.y;
|
|
290
|
+
|
|
291
|
+
int k = blockIdx.z*blockDim.z + threadIdx.z+1;
|
|
292
|
+
|
|
293
|
+
|
|
294
|
+
|
|
295
|
+
if ((0<i) && (0<k) && (i < nx) && (j < ny) && (k < nz)) {
|
|
296
|
+
|
|
297
|
+
Ey[(k * ny * nx + j * nx + i)] = A[0] * oldEy[(k * ny * nx + j * nx + i)] + A[1] * (((oldHx[(k * ny * nx + j * nx + i)] - oldHx[((k - 1) * ny * nx + j * nx + i)]) / d[3]) - ((oldHz[(k * ny * nx + j * nx + i)] - oldHz[(k * ny * nx + j * nx + (i - 1))]) / d[3])); //4
|
|
298
|
+
|
|
299
|
+
|
|
300
|
+
|
|
301
|
+
}
|
|
302
|
+
|
|
303
|
+
__syncthreads();
|
|
304
|
+
|
|
305
|
+
|
|
306
|
+
|
|
307
|
+
}
|
|
308
|
+
|
|
309
|
+
|
|
310
|
+
|
|
311
|
+
__global__ void culcEz(double *Ez, double *oldEz, double *oldHx, double *oldHy, double *A, double *d)
|
|
312
|
+
|
|
313
|
+
{
|
|
314
|
+
|
|
315
|
+
int i = blockIdx.x*blockDim.x + threadIdx.x+1;
|
|
316
|
+
|
|
317
|
+
int j = blockIdx.y*blockDim.y + threadIdx.y+1;
|
|
318
|
+
|
|
319
|
+
int k = blockIdx.z*blockDim.z + threadIdx.z;
|
|
320
|
+
|
|
321
|
+
|
|
322
|
+
|
|
323
|
+
if ((0<i) && (0<j) && (i < nx) && (j < ny) && (k < nz)) {
|
|
324
|
+
|
|
325
|
+
Ez[(k * ny * nx + j * nx + i)] = A[0] * oldEz[(k * ny * nx + j * nx + i)] + A[1] * (((oldHy[(k * ny * nx + j * nx + i)] - oldHy[(k * ny * nx + j * nx + (i - 1))]) / d[1]) - ((oldHx[(k * ny * nx + j * nx + i)] - oldHx[(k * ny * nx + (j - 1) * nx + i)]) / d[2])); //4
|
|
326
|
+
|
|
327
|
+
}
|
|
328
|
+
|
|
329
|
+
__syncthreads();
|
|
330
|
+
|
|
331
|
+
|
|
332
|
+
|
|
333
|
+
}
|
|
334
|
+
|
|
335
|
+
/*----------------------------磁界計算----------------------*/
|
|
336
|
+
|
|
337
|
+
__global__ void culcHx(double *Ey, double *Ez, double *Hx, double *oldHx, double *A, double *d)
|
|
338
|
+
|
|
339
|
+
{
|
|
340
|
+
|
|
341
|
+
|
|
342
|
+
|
|
343
|
+
int i = blockIdx.x*blockDim.x + threadIdx.x;
|
|
344
|
+
|
|
345
|
+
int j = blockIdx.y*blockDim.y + threadIdx.y;
|
|
346
|
+
|
|
347
|
+
int k = blockIdx.z*blockDim.z + threadIdx.z;
|
|
348
|
+
|
|
349
|
+
|
|
350
|
+
|
|
351
|
+
if ((0<i)&&(i<nx) && (j<ny - 1) && (k<nz - 1)) {
|
|
352
|
+
|
|
353
|
+
Hx[k * ny * nx + j * nx + i] = oldHx[k * ny * nx + j * nx + i] - (A[2] * (((Ez[k * ny * nx + (j + 1) * nx + i] - Ez[k * ny * nx + j * nx + i]) / d[2]) - ((Ey[(k + 1) * ny * nx + j * nx + i] - Ey[k * ny * nx + j * nx + i]) / d[3]))); //5
|
|
354
|
+
|
|
355
|
+
}
|
|
356
|
+
|
|
357
|
+
__syncthreads();
|
|
358
|
+
|
|
359
|
+
|
|
360
|
+
|
|
361
|
+
}
|
|
362
|
+
|
|
363
|
+
|
|
364
|
+
|
|
365
|
+
|
|
366
|
+
|
|
367
|
+
__global__ void culcHy(double *Ex, double *Ez, double *Hy, double *oldHy, double *A, double *d)
|
|
368
|
+
|
|
369
|
+
{
|
|
370
|
+
|
|
371
|
+
int i = blockIdx.x*blockDim.x + threadIdx.x;
|
|
372
|
+
|
|
373
|
+
int j = blockIdx.y*blockDim.y + threadIdx.y;
|
|
374
|
+
|
|
375
|
+
int k = blockIdx.z*blockDim.z + threadIdx.z;
|
|
376
|
+
|
|
377
|
+
|
|
378
|
+
|
|
379
|
+
if ((0<j) && (i<nx - 1) && (j<ny) && (k<nz - 1)) {
|
|
380
|
+
|
|
381
|
+
Hy[k * ny * nx + j * nx + i] = oldHy[k * ny * nx + j * nx + i]- (A[2] * (((Ex[(k + 1) * ny * nx + j * nx + i] - Ex[k * ny * nx + j * nx + i]) / d[3]) - ((Ez[k * ny * nx + j * nx + (i + 1)] - Ez[k * ny * nx + j * nx + i]) / d[1]))); //6
|
|
382
|
+
|
|
383
|
+
}
|
|
384
|
+
|
|
385
|
+
__syncthreads();
|
|
386
|
+
|
|
387
|
+
|
|
388
|
+
|
|
389
|
+
}
|
|
390
|
+
|
|
391
|
+
|
|
392
|
+
|
|
393
|
+
|
|
394
|
+
|
|
395
|
+
__global__ void culcHz(double *Ex, double *Ey, double *Hz, double *oldHz, double *A, double *d)
|
|
396
|
+
|
|
397
|
+
{
|
|
398
|
+
|
|
399
|
+
int i = blockIdx.x*blockDim.x + threadIdx.x;
|
|
400
|
+
|
|
401
|
+
int j = blockIdx.y*blockDim.y + threadIdx.y;
|
|
402
|
+
|
|
403
|
+
int k = blockIdx.z*blockDim.z + threadIdx.z;
|
|
404
|
+
|
|
405
|
+
|
|
406
|
+
|
|
407
|
+
if ((0<j) && (i<nx - 1) && j<(ny - 1) && k<nz) {
|
|
408
|
+
|
|
409
|
+
Hz[k * ny * nx + j * nx + i] = oldHz[k * ny * nx + j * nx + i]- (A[2] * (((Ey[k * ny * nx + j * nx + (i + 1)] - Ey[k * ny * nx + j * nx + i]) / d[1]) - ((Ex[k * ny * nx + (j + 1) * nx + i] - Ex[k * ny * nx + j * nx + i]) / d[2]))); //6
|
|
410
|
+
|
|
411
|
+
}
|
|
412
|
+
|
|
413
|
+
__syncthreads();
|
|
414
|
+
|
|
415
|
+
|
|
416
|
+
|
|
417
|
+
}
|
|
22
418
|
|
|
23
419
|
```
|
|
24
420
|
|
|
25
421
|
|
|
26
422
|
|
|
27
|
-
### 該当のソースコード
|
|
28
|
-
|
|
29
|
-
/*--------------------------------------------メイン計算部分------------------------------------------------------------*/
|
|
30
|
-
|
|
31
|
-
while (t[0]<2e-9) {
|
|
32
|
-
|
|
33
|
-
|
|
34
|
-
|
|
35
|
-
culcEx << <grid, block >> > (Ex_d, oldEx_d, oldHy_d, oldHz_d, A_d, d_d);
|
|
36
|
-
|
|
37
|
-
culcEy << <grid, block >> > (Ey_d, oldEy_d, oldHx_d, oldHz_d, A_d, d_d);
|
|
38
|
-
|
|
39
|
-
culcEz << <grid, block >> > (Ez_d, oldEz_d, oldHx_d, oldHy_d, A_d, d_d);
|
|
40
|
-
|
|
41
|
-
cudaDeviceSynchronize();
|
|
42
|
-
|
|
43
|
-
|
|
44
|
-
|
|
45
|
-
|
|
46
|
-
|
|
47
|
-
vin[0] = 1.0*sin(2.0e10*M_PI*t[0]);
|
|
48
|
-
|
|
49
|
-
cudaMemcpy(vin_d, vin, sizeof(double), cudaMemcpyHostToDevice);
|
|
50
|
-
|
|
51
|
-
Init << <grid, block >> > ( Ez_d, t_d, vin_d);
|
|
52
|
-
|
|
53
|
-
cudaDeviceSynchronize();
|
|
54
|
-
|
|
55
|
-
Init2 << <grid, block >> > (Ex_d, Ey_d, Ez_d);
|
|
56
|
-
|
|
57
|
-
cudaDeviceSynchronize();
|
|
58
|
-
|
|
59
|
-
MurEx << <grid, block >> > (Ex_d, oldEx_d, A_d);
|
|
60
|
-
|
|
61
|
-
MurEy << <grid, block >> > (Ey_d, oldEy_d, A_d);
|
|
62
|
-
|
|
63
|
-
MurEz << <grid, block >> > (Ez_d, oldEz_d, A_d);
|
|
64
|
-
|
|
65
|
-
cudaDeviceSynchronize();
|
|
66
|
-
|
|
67
|
-
|
|
68
|
-
|
|
69
|
-
|
|
70
|
-
|
|
71
|
-
culct << <2,2 >> > (t_d);
|
|
72
|
-
|
|
73
|
-
t[0] += (dt / 2);
|
|
74
|
-
|
|
75
|
-
cudaDeviceSynchronize();
|
|
76
|
-
|
|
77
|
-
|
|
78
|
-
|
|
79
|
-
culcHx << <grid, block >> > (Ey_d, Ez_d, Hx_d, oldHx_d, A_d, d_d);
|
|
80
|
-
|
|
81
|
-
culcHy << <grid, block >> > (Ex_d, Ez_d, Hy_d, oldHy_d, A_d, d_d);
|
|
82
|
-
|
|
83
|
-
culcHz << <grid, block >> > (Ex_d, Ey_d, Hz_d, oldHz_d, A_d, d_d);
|
|
84
|
-
|
|
85
|
-
cudaDeviceSynchronize();
|
|
86
|
-
|
|
87
|
-
|
|
88
|
-
|
|
89
|
-
change << <grid, block >> > (Ex_d, Ey_d, Ez_d, oldEx_d, oldEy_d, oldEz_d, Hx_d, Hy_d, Hz_d, oldHx_d, oldHy_d, oldHz_d);
|
|
90
|
-
|
|
91
|
-
cudaDeviceSynchronize();
|
|
92
|
-
|
|
93
|
-
|
|
94
|
-
|
|
95
|
-
culct << <2, 2 >> > (t_d);
|
|
96
|
-
|
|
97
|
-
t[0] += (dt / 2);
|
|
98
|
-
|
|
99
|
-
cudaDeviceSynchronize();
|
|
100
|
-
|
|
101
|
-
output << <grid, block >> > (out_d, Ez_d);
|
|
102
|
-
|
|
103
|
-
cudaDeviceSynchronize();
|
|
104
|
-
|
|
105
|
-
|
|
106
|
-
|
|
107
|
-
}
|
|
108
|
-
|
|
109
|
-
/*---------------------------------------------------------------------------------------------------------------------*/
|
|
110
|
-
|
|
111
|
-
|
|
112
|
-
|
|
113
|
-
/*------------------吸収境界条件部分----------------------------*/
|
|
114
|
-
|
|
115
|
-
__global__ void MurEx(double *Ex, double *oldEx, double *A){
|
|
116
|
-
|
|
117
|
-
int i = blockIdx.x*blockDim.x + threadIdx.x;
|
|
118
|
-
|
|
119
|
-
int j = blockIdx.y*blockDim.y + threadIdx.y;
|
|
120
|
-
|
|
121
|
-
int k = blockIdx.z*blockDim.z + threadIdx.z;
|
|
122
|
-
|
|
123
|
-
|
|
124
|
-
|
|
125
|
-
if (i < nx && j < ny && k < nz) {
|
|
126
|
-
|
|
127
|
-
Ex[(k * ny * nx + j * nx + 0)] =
|
|
128
|
-
|
|
129
|
-
oldEx[(k * ny * nx + j * nx + 1)] + A[3] * (Ex[(k * ny * nx + j * nx + 1)] - oldEx[(k * ny * nx + j * nx + 0)]);
|
|
130
|
-
|
|
131
|
-
Ex[(k * ny * nx + j * nx + nx - 1)] = oldEx[(k * ny * nx + j * nx + nx - 2)] + A[3] * (Ex[(k * ny * nx + j * nx + nx - 2)] - oldEx[(k * ny * nx + j * nx + nx - 1)]);
|
|
132
|
-
|
|
133
|
-
}
|
|
134
|
-
|
|
135
|
-
__syncthreads();
|
|
136
|
-
|
|
137
|
-
}
|
|
138
|
-
|
|
139
|
-
|
|
140
|
-
|
|
141
|
-
__global__ void MurEy( double *Ey, double *oldEy, double *A){
|
|
142
|
-
|
|
143
|
-
int i = blockIdx.x*blockDim.x + threadIdx.x;
|
|
144
|
-
|
|
145
|
-
int j = blockIdx.y*blockDim.y + threadIdx.y;
|
|
146
|
-
|
|
147
|
-
int k = blockIdx.z*blockDim.z + threadIdx.z;
|
|
148
|
-
|
|
149
|
-
|
|
150
|
-
|
|
151
|
-
if (i < nx && j < ny && k < nz) {
|
|
152
|
-
|
|
153
|
-
Ey[(k * ny * nx + 0 * nx + i)] =
|
|
154
|
-
|
|
155
|
-
oldEy[(k * ny * nx + 1 * nx + i)] + A[3] * (Ey[(k * ny * nx + 1 * nx + i)] - oldEy[(k * ny * nx + 0 * nx + i)]);
|
|
156
|
-
|
|
157
|
-
Ey[(k * ny * nx + (ny - 1) * nx + i)] =
|
|
158
|
-
|
|
159
|
-
oldEy[(k * ny * nx + (ny - 2) * nx + i)] + A[3] * (Ey[(k * ny * nx + (ny - 2) * nx + i)] - oldEy[(k * ny * nx + (ny - 1) * nx + i)]);
|
|
160
|
-
|
|
161
|
-
}
|
|
162
|
-
|
|
163
|
-
__syncthreads();
|
|
164
|
-
|
|
165
|
-
}
|
|
166
|
-
|
|
167
|
-
|
|
168
|
-
|
|
169
|
-
__global__ void MurEz( double *Ez, double *oldEz, double *A){
|
|
170
|
-
|
|
171
|
-
int i = blockIdx.x*blockDim.x + threadIdx.x;
|
|
172
|
-
|
|
173
|
-
int j = blockIdx.y*blockDim.y + threadIdx.y;
|
|
174
|
-
|
|
175
|
-
int k = blockIdx.z*blockDim.z + threadIdx.z;
|
|
176
|
-
|
|
177
|
-
|
|
178
|
-
|
|
179
|
-
if (i < nx && j < ny && k < nz) {
|
|
180
|
-
|
|
181
|
-
Ez[(k * ny * nx + j * nx + 0)] =
|
|
182
|
-
|
|
183
|
-
oldEz[(k * ny * nx + j * nx + 1)] + A[3] * (Ez[(k * ny * nx + j * nx + 1)] - oldEz[(k * ny * nx + j * nx + 0)]);
|
|
184
|
-
|
|
185
|
-
Ez[(k * ny * nx + j * nx + nx - 1)] =
|
|
186
|
-
|
|
187
|
-
oldEz[(k * ny * nx + j * nx + nx - 2)] + A[3] * (Ez[(k * ny * nx + j * nx + nx - 2)] - oldEz[(k * ny * nx + j * nx + nx - 1)]);
|
|
188
|
-
|
|
189
|
-
|
|
190
|
-
|
|
191
|
-
Ez[(k * ny * nx + 0 * nx + i)] =
|
|
192
|
-
|
|
193
|
-
oldEz[(k * ny * nx + 1 * nx + i)] + A[3] * (Ez[(k * ny * nx + 1 * nx + i)] - oldEz[(k * ny * nx + 0 * nx + i)]);
|
|
194
|
-
|
|
195
|
-
Ez[(k * ny * nx + (ny - 1) * nx + i)]
|
|
196
|
-
|
|
197
|
-
= oldEz[(k * ny * nx + (ny - 2) * nx + i)] + A[3] * (Ez[(k * ny * nx + (ny - 2) * nx + i)] - oldEz[(k * ny * nx + (ny - 1) * nx + i)]);
|
|
198
|
-
|
|
199
|
-
}
|
|
200
|
-
|
|
201
|
-
__syncthreads();
|
|
202
|
-
|
|
203
|
-
}
|
|
204
|
-
|
|
205
|
-
/*------------------------------------------------------------------------------*/
|
|
206
|
-
|
|
207
|
-
//その他補足（プログラムより抜粋)
|
|
208
|
-
|
|
209
|
-
♯define nx 41
|
|
210
|
-
|
|
211
|
-
♯define ny 41
|
|
212
|
-
|
|
213
|
-
♯define nz 41
|
|
214
|
-
|
|
215
|
-
|
|
216
|
-
|
|
217
|
-
dim3 grid;
|
|
218
|
-
|
|
219
|
-
grid.x = 3;
|
|
220
|
-
|
|
221
|
-
grid.y = 3;
|
|
222
|
-
|
|
223
|
-
grid.z = 24;
|
|
224
|
-
|
|
225
|
-
dim3 block;
|
|
226
|
-
|
|
227
|
-
block.x = 16;
|
|
228
|
-
|
|
229
|
-
block.y = 16;
|
|
230
|
-
|
|
231
|
-
block.z = 2;
|
|
232
|
-
|
|
233
|
-
/*--------------------------電磁界計算用係数設定------------------------*/
|
|
234
|
-
|
|
235
|
-
A[0] = ((1 - (sig*dt) / (2 * ep)) / (1 + (sig*dt) / (2 * ep)));
|
|
236
|
-
|
|
237
|
-
A[1] = (dt / ep) / (1 + (sig*dt / (2 * ep)));
|
|
238
|
-
|
|
239
|
-
A[2] = dt / mu;
|
|
240
|
-
|
|
241
|
-
A[3] = (c*dt - dx) / (c*dt + dx);
|
|
242
|
-
|
|
243
|
-
/*-----------------------------------------------
|
|
244
|
-
|
|
245
|
-
0:電界計算,1:電界計算,2:磁界計算,3:Mur吸収境界条件
|
|
246
|
-
|
|
247
|
-
-------------------------------------------------*/
|
|
248
|
-
|
|
249
|
-
d[0] = 2.0e-12;
|
|
250
|
-
|
|
251
|
-
d[1] = 1.5e-3;
|
|
252
|
-
|
|
253
|
-
d[2] = 1.5e-3;
|
|
254
|
-
|
|
255
|
-
d[3] = 1.5e-3;
|
|
256
|
-
|
|
257
|
-
/*-----------------------------------------------
|
|
258
|
-
|
|
259
|
-
0:dt,1:dx,2:dy,3:dz
|
|
260
|
-
|
|
261
|
-
-------------------------------------------------*/
|
|
262
|
-
|
|
263
|
-
/*----------------------------電界計算----------------------*/
|
|
264
|
-
|
|
265
|
-
__global__ void culcEx(double *Ex, double *oldEx, double *oldHy, double *oldHz, double *A, double *d)
|
|
266
|
-
|
|
267
|
-
{
|
|
268
|
-
|
|
269
|
-
|
|
270
|
-
|
|
271
|
-
int i = blockIdx.x*blockDim.x + threadIdx.x;
|
|
272
|
-
|
|
273
|
-
int j = blockIdx.y*blockDim.y + threadIdx.y+1;
|
|
274
|
-
|
|
275
|
-
int k = blockIdx.z*blockDim.z + threadIdx.z+1;
|
|
276
|
-
|
|
277
|
-
|
|
278
|
-
|
|
279
|
-
if ((0 < j) && (0 < k) && (i < nx) && (j < ny) && (k < nz)) {
|
|
280
|
-
|
|
281
|
-
Ex[(k * ny * nx + j * nx + i)] = A[0] * oldEx[(k * ny * nx + j * nx + i)] + A[1] * (((oldHz[(k * ny * nx + j * nx + i)] - oldHz[(k * ny * nx + (j - 1) * nx + i)]) / d[2]) - ((oldHy[(k * ny * nx + j * nx + i)] - oldHy[((k - 1) * ny * nx + j * nx + i)]) / d[3])); //4
|
|
282
|
-
|
|
283
|
-
}
|
|
284
|
-
|
|
285
|
-
__syncthreads();
|
|
286
|
-
|
|
287
|
-
|
|
288
|
-
|
|
289
|
-
}
|
|
290
|
-
|
|
291
|
-
|
|
292
|
-
|
|
293
|
-
__global__ void culcEy(double *Ey, double *oldEy, double *oldHx, double *oldHz, double *A, double *d)
|
|
294
|
-
|
|
295
|
-
{
|
|
296
|
-
|
|
297
|
-
int i = blockIdx.x*blockDim.x + threadIdx.x+1;
|
|
298
|
-
|
|
299
|
-
int j = blockIdx.y*blockDim.y + threadIdx.y;
|
|
300
|
-
|
|
301
|
-
int k = blockIdx.z*blockDim.z + threadIdx.z+1;
|
|
302
|
-
|
|
303
|
-
|
|
304
|
-
|
|
305
|
-
if ((0<i) && (0<k) && (i < nx) && (j < ny) && (k < nz)) {
|
|
306
|
-
|
|
307
|
-
Ey[(k * ny * nx + j * nx + i)] = A[0] * oldEy[(k * ny * nx + j * nx + i)] + A[1] * (((oldHx[(k * ny * nx + j * nx + i)] - oldHx[((k - 1) * ny * nx + j * nx + i)]) / d[3]) - ((oldHz[(k * ny * nx + j * nx + i)] - oldHz[(k * ny * nx + j * nx + (i - 1))]) / d[3])); //4
|
|
308
|
-
|
|
309
|
-
|
|
310
|
-
|
|
311
|
-
}
|
|
312
|
-
|
|
313
|
-
__syncthreads();
|
|
314
|
-
|
|
315
|
-
|
|
316
|
-
|
|
317
|
-
}
|
|
318
|
-
|
|
319
|
-
|
|
320
|
-
|
|
321
|
-
__global__ void culcEz(double *Ez, double *oldEz, double *oldHx, double *oldHy, double *A, double *d)
|
|
322
|
-
|
|
323
|
-
{
|
|
324
|
-
|
|
325
|
-
int i = blockIdx.x*blockDim.x + threadIdx.x+1;
|
|
326
|
-
|
|
327
|
-
int j = blockIdx.y*blockDim.y + threadIdx.y+1;
|
|
328
|
-
|
|
329
|
-
int k = blockIdx.z*blockDim.z + threadIdx.z;
|
|
330
|
-
|
|
331
|
-
|
|
332
|
-
|
|
333
|
-
if ((0<i) && (0<j) && (i < nx) && (j < ny) && (k < nz)) {
|
|
334
|
-
|
|
335
|
-
Ez[(k * ny * nx + j * nx + i)] = A[0] * oldEz[(k * ny * nx + j * nx + i)] + A[1] * (((oldHy[(k * ny * nx + j * nx + i)] - oldHy[(k * ny * nx + j * nx + (i - 1))]) / d[1]) - ((oldHx[(k * ny * nx + j * nx + i)] - oldHx[(k * ny * nx + (j - 1) * nx + i)]) / d[2])); //4
|
|
336
|
-
|
|
337
|
-
}
|
|
338
|
-
|
|
339
|
-
__syncthreads();
|
|
340
|
-
|
|
341
|
-
|
|
342
|
-
|
|
343
|
-
}
|
|
344
|
-
|
|
345
|
-
/*----------------------------磁界計算----------------------*/
|
|
346
|
-
|
|
347
|
-
__global__ void culcHx(double *Ey, double *Ez, double *Hx, double *oldHx, double *A, double *d)
|
|
348
|
-
|
|
349
|
-
{
|
|
350
|
-
|
|
351
|
-
|
|
352
|
-
|
|
353
|
-
int i = blockIdx.x*blockDim.x + threadIdx.x;
|
|
354
|
-
|
|
355
|
-
int j = blockIdx.y*blockDim.y + threadIdx.y;
|
|
356
|
-
|
|
357
|
-
int k = blockIdx.z*blockDim.z + threadIdx.z;
|
|
358
|
-
|
|
359
|
-
|
|
360
|
-
|
|
361
|
-
if ((0<i)&&(i<nx) && (j<ny - 1) && (k<nz - 1)) {
|
|
362
|
-
|
|
363
|
-
Hx[k * ny * nx + j * nx + i] = oldHx[k * ny * nx + j * nx + i] - (A[2] * (((Ez[k * ny * nx + (j + 1) * nx + i] - Ez[k * ny * nx + j * nx + i]) / d[2]) - ((Ey[(k + 1) * ny * nx + j * nx + i] - Ey[k * ny * nx + j * nx + i]) / d[3]))); //5
|
|
364
|
-
|
|
365
|
-
}
|
|
366
|
-
|
|
367
|
-
__syncthreads();
|
|
368
|
-
|
|
369
|
-
|
|
370
|
-
|
|
371
|
-
}
|
|
372
|
-
|
|
373
|
-
|
|
374
|
-
|
|
375
|
-
|
|
376
|
-
|
|
377
|
-
__global__ void culcHy(double *Ex, double *Ez, double *Hy, double *oldHy, double *A, double *d)
|
|
378
|
-
|
|
379
|
-
{
|
|
380
|
-
|
|
381
|
-
int i = blockIdx.x*blockDim.x + threadIdx.x;
|
|
382
|
-
|
|
383
|
-
int j = blockIdx.y*blockDim.y + threadIdx.y;
|
|
384
|
-
|
|
385
|
-
int k = blockIdx.z*blockDim.z + threadIdx.z;
|
|
386
|
-
|
|
387
|
-
|
|
388
|
-
|
|
389
|
-
if ((0<j) && (i<nx - 1) && (j<ny) && (k<nz - 1)) {
|
|
390
|
-
|
|
391
|
-
Hy[k * ny * nx + j * nx + i] = oldHy[k * ny * nx + j * nx + i]- (A[2] * (((Ex[(k + 1) * ny * nx + j * nx + i] - Ex[k * ny * nx + j * nx + i]) / d[3]) - ((Ez[k * ny * nx + j * nx + (i + 1)] - Ez[k * ny * nx + j * nx + i]) / d[1]))); //6
|
|
392
|
-
|
|
393
|
-
}
|
|
394
|
-
|
|
395
|
-
__syncthreads();
|
|
396
|
-
|
|
397
|
-
|
|
398
|
-
|
|
399
|
-
}
|
|
400
|
-
|
|
401
|
-
|
|
402
|
-
|
|
403
|
-
|
|
404
|
-
|
|
405
|
-
__global__ void culcHz(double *Ex, double *Ey, double *Hz, double *oldHz, double *A, double *d)
|
|
406
|
-
|
|
407
|
-
{
|
|
408
|
-
|
|
409
|
-
int i = blockIdx.x*blockDim.x + threadIdx.x;
|
|
410
|
-
|
|
411
|
-
int j = blockIdx.y*blockDim.y + threadIdx.y;
|
|
412
|
-
|
|
413
|
-
int k = blockIdx.z*blockDim.z + threadIdx.z;
|
|
414
|
-
|
|
415
|
-
|
|
416
|
-
|
|
417
|
-
if ((0<j) && (i<nx - 1) && j<(ny - 1) && k<nz) {
|
|
418
|
-
|
|
419
|
-
Hz[k * ny * nx + j * nx + i] = oldHz[k * ny * nx + j * nx + i]- (A[2] * (((Ey[k * ny * nx + j * nx + (i + 1)] - Ey[k * ny * nx + j * nx + i]) / d[1]) - ((Ex[k * ny * nx + (j + 1) * nx + i] - Ex[k * ny * nx + j * nx + i]) / d[2]))); //6
|
|
420
|
-
|
|
421
|
-
}
|
|
422
|
-
|
|
423
|
-
__syncthreads();
|
|
424
|
-
|
|
425
|
-
|
|
426
|
-
|
|
427
|
-
}
|
|
428
|
-
|
|
429
|
-
|
|
430
|
-
|
|
431
|
-
|
|
432
|
-
|
|
433
423
|
|
|
434
424
|
|
|
435
425
|
|
1
誤字
test
CHANGED
|
File without changes
|
test
CHANGED
|
@@ -448,11 +448,7 @@
|
|
|
448
448
|
|
|
449
449
|
i = blockIdx.x*blockDim.x + threadIdx.x+1;のようにしてみたり
|
|
450
450
|
|
|
451
|
-
メイン関数中のMurE関数がくるところを
|
|
452
|
-
|
|
453
|
-
|
|
454
|
-
|
|
455
|
-
の
|
|
451
|
+
メイン関数中のMurE関数がくるところを変更してCPUで計算させてもみましたがダメでした。
|
|
456
452
|
|
|
457
453
|
### 補足情報（FW/ツールのバージョンなど）
|
|
458
454
|
|