-
Notifications
You must be signed in to change notification settings - Fork 1
/
cuda_extmath.h
executable file
·298 lines (251 loc) · 7.63 KB
/
cuda_extmath.h
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
294
295
296
297
298
#ifndef __CUDA_EXTMATH
#define __CUDA_EXTMATH
#include <math.h>
#include "extmath.h"
#include <cuda.h>
//#include <cutil_inline.h>
//__device__ inline float_type cuda_abs2(float_type* z) {return (*z)*(*z)+(*(z+1))*(*(z+1));}
__device__ inline float_type cuda_abs2(float_type re, float_type im) {return re*re+im*im;}
//__device__ __host__ inline int floor_(float_type x) { int c=(int)x; return (c<=x)?(c):(c-1);}
#define floor_ floor
#define SWAP(a,b) tempr=(a);(a)=(b);(b)=tempr
__device__ inline void fft_device(float_type* data, unsigned int nn, int isign)
/*Replaces data[1..2*nn] by its discrete Fourier transform, if isign is input as 1; or replaces
data[1..2*nn] by nn times its inverse discrete Fourier transform, if isign is input as −1.
data is a complex array of length nn or, equivalently, a real array of length 2*nn. nn MUST
be an integer power of 2 (this is not checked for!).*/
{
unsigned int n,mmax,m,j,istep,i;
float_type wtemp,wr,wpr,wpi,wi,theta;
float_type tempr,tempi;
float_type rdi, rdj, idi, idj;
data--;
n=nn << 1;
j=1;
for (i=1;i<n;i+=2)
{
// This is the bit-reversal section of the routine.
if (j > i)
{
SWAP(data[j],data[i]);
//Exchange the two complex numbers.
SWAP(data[(j+1)],data[(i+1)]);
}
m=n >> 1;
while (m >= 2 && j > m)
{
j -= m;
m >>= 1;
}
j += m;
}
//Here begins the Danielson-Lanczos section of the routine.
mmax=2;
while (n > mmax)
{
//Outer loop executed log2 nn times.
istep=mmax << 1;
theta=(float_type)isign*(6.28318530717959/(float_type)mmax);
//Initialize the trigonometric recurrence.
wtemp=sin_p(0.5*theta);
wpr = -2.0*wtemp*wtemp;
wpi=sin_f(theta);
wr=1.0;
wi=0.0;
for (m=1;m<mmax;m+=2)
{
for (i=m;i<=n;i+=istep)
{
j=i+mmax;
rdj = data[j]; idj = data[j+1];
rdi = data[i]; idi = data[i+1];
tempr=wr*rdj-wi*idj;
tempi=wr*idj+wi*rdj;
data[j] =rdi-tempr;
data[(j+1)]=idi-tempi;
data[i] = rdi+tempr;
data[(i+1)] = idi+tempi;
}
wr=(wtemp=wr)*wpr-wi*wpi+wr;
wi=wi*wpr+wtemp*wpi+wi;
}
mmax=istep;
}
if (isign == 1) for (i=1;i<2*nn+1;i++) data[i]/=nn;
}
__device__ inline void fft_device_strided(float_type* data, unsigned int nn, int isign, int stride)
/*Replaces data[1..2*nn] by its discrete Fourier transform, if isign is input as 1; or replaces
data[1..2*nn] by nn times its inverse discrete Fourier transform, if isign is input as −1.
data is a complex array of length nn or, equivalently, a real array of length 2*nn. nn MUST
be an integer power of 2 (this is not checked for!).*/
{
unsigned int n,mmax,m,j,istep,i;
float_type wtemp,wr,wpr,wpi,wi,theta;
float_type tempr,tempi;
float_type rdi, idi, rdj, idj;
data-=stride;
n=nn << 1;
j=1;
for (i=1;i<n;i+=2)
{
// This is the bit-reversal section of the routine.
if (j > i)
{
SWAP(data[j*stride],data[i*stride]);
//Exchange the two complex numbers.
SWAP(data[(j+1)*stride],data[(i+1)*stride]);
}
m=n >> 1;
while (m >= 2 && j > m)
{
j -= m;
m >>= 1;
}
j += m;
}
//Here begins the Danielson-Lanczos section of the routine.
mmax=2;
while (n > mmax)
{
//Outer loop executed log2 nn times.
istep=mmax << 1;
theta=(float_type)isign*(6.28318530717959/(float_type)mmax);
//Initialize the trigonometric recurrence.
wtemp=sin_p(0.5*theta);
wpr = -2.0*wtemp*wtemp;
wpi=sin_p(theta);
wr=1.0;
wi=0.0;
for (m=1;m<mmax;m+=2)
{
for (i=m;i<=n;i+=istep)
{
j=i+mmax;
rdj = data[j*stride]; idj = data[(j+1)*stride];
rdi = data[i*stride]; idi = data[(i+1)*stride];
tempr=wr*rdj-wi*idj;
tempi=wr*idj+wi*rdj;
data[j*stride] =rdi-tempr;
data[(j+1)*stride]=idi-tempi;
data[i*stride] = rdi+tempr;
data[(i+1)*stride] = idi+tempi;
}
wr=(wtemp=wr)*wpr-wi*wpi+wr;
wi=wi*wpr+wtemp*wpi+wi;
}
mmax=istep;
}
if (isign == 1) for (i=1;i<2*nn+1;i++) data[i*stride]/=nn;
}
__device__ inline void getpolar(float_type reX, float_type imX, float_type* ro, float_type* phi)
{
float_type A = sqrt_p(reX*reX + imX*imX);
(*ro) = A; (*phi) = acos_p(reX/A)*sign(imX);
}
__device__ inline void getpolar(float_type* X, float_type* ro, float_type* phi)
{
float_type reX = X[0], imX = X[1];
getpolar(reX, imX, ro, phi);
}
__device__ inline void sqrtc(float_type reX, float_type imX, float_type* reY, float_type* imY)
{
float_type A,phi; getpolar(reX, imX, &A, &phi);
float_type s,c; sincos_f(phi/2.0, &s, &c);
float_type absy = sqrt_p(A);
(*reY) = absy*c; (*imY) = absy*s;
}
__device__ inline void sqrtc(float_type* X, float_type* reY, float_type* imY)
{
float_type reX = X[0], imX = X[1];
sqrtc(reX, imX, reY, imY);
}
__device__ inline float_type device_taylor_sum(float_type x, float_type* k, int N, int startn)
{
float_type r = 0;
float_type xn = 1; for (int j=0; j<startn; j++) xn*=x;
for (int j=0; j<N; j++) {r+=k[j]*xn; xn*=x;}
return r;
}
__device__ inline float_type device_taylor_sum1(float_type x, float_type* k, int N)
{
float_type r = 0;
float_type xn = x;
for (int j=0; j<N; j++) {r+=k[j]*xn; xn*=x;}
return r;
}
__device__ inline float_type device_taylor_sum0(float_type x, float_type* k, int N)
{
float_type r = 0;
float_type xn = 1;
for (int j=0; j<N; j++) {r+=k[j]*xn; xn*=x;}
return r;
}
__device__ inline float_type device_sqrtHO(float_type x)
{
float_type k[20] = {0.5, -0.125, 0.0625, -0.0390625, 0.02734375, -0.0205078125, 0.01611328125, -0.013092041015625, 0.010910034179688, \
-0.009273529052734, 0.008008956909180, -0.007007837295532, 0.006199240684509, -0.005535036325455, 0.004981532692909, -0.004514514002949, \
0.004116174532101, -0.003773159987759, 0.003475278936094, -0.003214633015887};
return device_taylor_sum1(x, k, 20);
}
__device__ inline void device_taylor_sum(float_type rex, float_type imx, float_type* rer, float_type* imr, float_type* k, int N, int startn)
{
float_type rexn = 1, imxn = 0;
float_type t = 0;
float_type rer_ = 0, imr_ = 0;
for (int j=0; j<startn; j++)
{
t=rexn*rex-imxn*imx;
imxn=rexn*imx+imxn*rex;
rexn=t;
}
for (int j=0; j<N; j++)
{
rer_+=k[j]*rexn;
imr_+=k[j]*imxn;
t=rexn*rex-imxn*imx;
imxn=rexn*imx+imxn*rex;
rexn=t;
}
(*rer)=rer_;
(*imr)=imr_;
}
__device__ inline void device_taylor_sum1(float_type rex, float_type imx, float_type* rer, float_type* imr, float_type* k, int N)
{
float_type rexn = rex, imxn = imx;
float_type t = 0;
float_type rer_ = 0, imr_ = 0;
for (int j=0; j<N; j++)
{
rer_+=k[j]*rexn;
imr_+=k[j]*imxn;
t=rexn*rex-imxn*imx;
imxn=rexn*imx+imxn*rex;
rexn=t;
}
(*rer)=rer_;
(*imr)=imr_;
}
__device__ inline void device_taylor_sum0(float_type rex, float_type imx, float_type* rer, float_type* imr, float_type* k, int N)
{
float_type rexn = 1, imxn = 0;
float_type t = 0;
float_type rer_ = 0, imr_ = 0;
for (int j=0; j<N; j++)
{
rer_+=k[j]*rexn;
imr_+=k[j]*imxn;
t =rexn*rex-imxn*imx;
imxn=rexn*imx+imxn*rex;
rexn=t;
}
(*rer)=rer_;
(*imr)=imr_;
}
__device__ inline void device_sqrtHO(float_type rex, float_type imx, float_type* rer, float_type* imr)
{
float_type k[20] = {0.5, -0.125, 0.0625, -0.0390625, 0.02734375, -0.0205078125, 0.01611328125, -0.013092041015625, 0.010910034179688, \
-0.009273529052734, 0.008008956909180, -0.007007837295532, 0.006199240684509, -0.005535036325455, 0.004981532692909, -0.004514514002949, \
0.004116174532101, -0.003773159987759, 0.003475278936094, -0.003214633015887};
device_taylor_sum1(rex, imx, rer, imr, k, 20);
}
#endif