fieldslice.cu
7.54 KB
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
290
291
292
293
#include "fieldslice.h"
#include "dataTypes.h"
#include "rts/cuda_handle_error.h"
__global__ void field_intensity(bsComplex* x, bsComplex* y, bsComplex* z, ptype* I, unsigned int N)
{
//compute the index for this thread
int i = blockIdx.x * blockDim.x + threadIdx.x;
if(i >= N) return;
ptype xm = x[i].abs();
if(y != NULL && z != NULL)
{
ptype ym = y[i].abs();
ptype zm = z[i].abs();
I[i] = xm*xm + ym*ym + zm*zm;
}
else
{
I[i] = xm*xm;
}
}
__global__ void resample_intensity(bsComplex* x, bsComplex* y, bsComplex* z, ptype* D, int uR, int vR, int ss)
{
//get the current coordinate in the plane slice
int iu = blockIdx.x * blockDim.x + threadIdx.x;
int iv = blockIdx.y * blockDim.y + threadIdx.y;
//make sure that the thread indices are in-bounds
if(iu >= uR || iv >= vR) return;
//compute the index into the detector
int i = iv*uR + iu;
//compute the index into the field
int fi;
//initialize the intensity for the pixel to zero
ptype I = 0;
ptype xm = 0;
ptype ym = 0;
ptype zm = 0;
int ix, iy;
for(ix = 0; ix<ss; ix++)
for(iy = 0; iy<ss; iy++)
{
//fi = iv*ss*ss*uR + iy*ss*uR + iu*ss + ix;
fi = (iv*ss + iy)*ss*uR + iu*ss + ix;
if(x !=NULL)
xm = x[fi].abs();
if(y != NULL)
ym = y[fi].abs();
if(z != NULL)
zm = z[fi].abs();
I += xm*xm + ym*ym + zm*zm;
}
D[i] += I/(ss*ss);
}
__global__ void field_real(bsComplex* field_component, ptype* V, unsigned int N)
{
//compute the index for this thread
int i = blockIdx.x * blockDim.x + threadIdx.x;
if(i >= N) return;
V[i] = field_component[i].real();
}
__global__ void field_imaginary(bsComplex* field_component, ptype* V, unsigned int N)
{
//compute the index for this thread
int i = blockIdx.x * blockDim.x + threadIdx.x;
if(i >= N) return;
V[i] = field_component[i].imag();
}
__global__ void field_sqrt(ptype* input, ptype* output, unsigned int N)
{
//compute the index for this thread
int i = blockIdx.x * blockDim.x + threadIdx.x;
if(i >= N) return;
output[i] = sqrt(input[i]);
}
__global__ void field_scale(bsComplex* x, bsComplex* y, bsComplex* z, unsigned int N, ptype v)
{
//compute the index for this thread
int i = blockIdx.x * blockDim.x + threadIdx.x;
if(i >= N) return;
if(x != NULL)
x[i] *= v;
if(y != NULL)
y[i] *= v;
if(z != NULL)
z[i] *= v;
}
scalarslice fieldslice::Mag()
{
//compute the magnitude of the field at each rtsPoint in the slice
scalarslice* result = new scalarslice(R[0], R[1]);
//compute the total number of values in the slice
unsigned int N = R[0] * R[1];
int gridDim = (N+BLOCK-1)/BLOCK;
field_intensity<<<gridDim, BLOCK>>>(x_hat, y_hat, z_hat, result->S, N);
field_sqrt<<<gridDim, BLOCK>>>(result->S, result->S, N);
return *result;
}
scalarslice fieldslice::Real()
{
//compute the magnitude of the field at each rtsPoint in the slice
//create a scalar slice at the same resolution as the field
scalarslice* result = new scalarslice(R[0], R[1]);
//compute the total number of values in the slice
unsigned int N = R[0] * R[1];
int gridDim = (N+BLOCK-1)/BLOCK;
field_real<<<gridDim, BLOCK>>>(x_hat, result->S, N);
return *result;
}
scalarslice fieldslice::Imag()
{
//compute the magnitude of the field at each rtsPoint in the slice
//create a scalar slice at the same resolution as the field
scalarslice* result = new scalarslice(R[0], R[1]);
//compute the total number of values in the slice
unsigned int N = R[0] * R[1];
int gridDim = (N+BLOCK-1)/BLOCK;
field_imaginary<<<gridDim, BLOCK>>>(x_hat, result->S, N);
return *result;
}
void fieldslice::IntegrateAndResample(scalarslice* detector, unsigned int supersample)
{
//compute the intensity and resample at the detector resolution
unsigned int D[2];
D[0] = detector->R[0];
D[1] = detector->R[1];
//create one thread for each detector pixel
dim3 dimBlock(SQRT_BLOCK, SQRT_BLOCK);
dim3 dimGrid((D[0] + SQRT_BLOCK -1)/SQRT_BLOCK, (D[1] + SQRT_BLOCK - 1)/SQRT_BLOCK);
resample_intensity<<<dimGrid, dimBlock>>>(x_hat, y_hat, z_hat, detector->S, D[0], D[1], supersample);
}
scalarslice fieldslice::Intensity()
{
//compute the magnitude of the field at each rtsPoint in the slice
//create a scalar slice at the same resolution as the field
scalarslice* result = new scalarslice(R[0], R[1]);
//compute the total number of values in the slice
unsigned int N = R[0] * R[1];
int gridDim = (N+BLOCK-1)/BLOCK;
field_intensity<<<gridDim, BLOCK>>>(x_hat, y_hat, z_hat, result->S, N);
return *result;
}
void fieldslice::ScaleField(ptype v)
{
//This function scales the field by some constant value v
//This is mostly used for the inverse FFT, which has to divide the field by R^2
//compute the total number of values in the slice
unsigned int N = R[0] * R[1];
//cout<<"Size of mag field: "<<N<<endl;
int gridDim = (N+BLOCK-1)/BLOCK;
field_scale<<<gridDim, BLOCK>>>(x_hat, y_hat, z_hat, N, v);
}
void fieldslice::init_gpu()
{
//allocate space on the GPU for the field slice
HANDLE_ERROR(cudaMalloc((void**)&x_hat, R[0] * R[1] * sizeof(bsComplex)));
//HANDLE_ERROR(cudaMemset(x_hat, 0, R[0] * R[1] * sizeof(bsComplex)));
//if the field is scalar, y_hat and z_hat are unused
if(scalarField)
{
y_hat = NULL;
z_hat = NULL;
}
else
{
HANDLE_ERROR(cudaMalloc((void**)&y_hat, R[0] * R[1] * sizeof(bsComplex)));
//HANDLE_ERROR(cudaMemset(y_hat, 0, R[0] * R[1] * sizeof(bsComplex)));
HANDLE_ERROR(cudaMalloc((void**)&z_hat, R[0] * R[1] * sizeof(bsComplex)));
//HANDLE_ERROR(cudaMemset(z_hat, 0, R[0] * R[1] * sizeof(bsComplex)));
}
clear_gpu();
}
void fieldslice::kill_gpu()
{
if(x_hat != NULL)
HANDLE_ERROR(cudaFree(x_hat));
if(y_hat != NULL)
HANDLE_ERROR(cudaFree(y_hat));
if(z_hat != NULL)
HANDLE_ERROR(cudaFree(z_hat));
}
void fieldslice::clear_gpu()
{
int memsize = R[0] * R[1] * sizeof(bsComplex);
if(x_hat != NULL)
HANDLE_ERROR(cudaMemset(x_hat, 0, memsize));
if(y_hat != NULL)
HANDLE_ERROR(cudaMemset(y_hat, 0, memsize));
if(z_hat != NULL)
HANDLE_ERROR(cudaMemset(z_hat, 0, memsize));
}
__global__ void copy_crop(bsComplex* source, bsComplex* dest, int u, int v, int su, int sv, int uR, int vR)
{
//get the current coordinate in the plane slice
int iu = blockIdx.x * blockDim.x + threadIdx.x;
int iv = blockIdx.y * blockDim.y + threadIdx.y;
//make sure that the thread indices are in-bounds
if(iu >= su || iv >= sv) return;
//compute the destination index
int i = iv*su + iu;
//compute the source index
int sourceV = v + iv;
int sourceU = u + iu;
int is = sourceV * uR + sourceU;
dest[i] = source[is];
}
fieldslice fieldslice::crop(int u, int v, int su, int sv)
{
//create a new field slice with the appropriate settings
fieldslice result(su, sv);
result.scalarField = scalarField;
//allocate space for the new field
result.init_gpu();
//create one thread for each pixel of the field slice
dim3 dimBlock(SQRT_BLOCK, SQRT_BLOCK);
dim3 dimGrid((su + SQRT_BLOCK -1)/SQRT_BLOCK, (sv + SQRT_BLOCK - 1)/SQRT_BLOCK);
//call a kernel to copy the cropped to the new field slice
if(x_hat != NULL)
copy_crop<<<dimGrid, dimBlock>>>(x_hat, result.x_hat, u, v, su, sv, R[0], R[1]);
if(y_hat != NULL)
copy_crop<<<dimGrid, dimBlock>>>(y_hat, result.y_hat, u, v, su, sv, R[0], R[1]);
if(z_hat != NULL)
copy_crop<<<dimGrid, dimBlock>>>(z_hat, result.z_hat, u, v, su, sv, R[0], R[1]);
return result;
}