1 | // GENERATED CODE --- DO NOT EDIT --- |
---|
2 | // Code is produced by sasmodels.gen from sasmodels/models/MODEL.c |
---|
3 | |
---|
4 | #ifdef __OPENCL_VERSION__ |
---|
5 | # define USE_OPENCL |
---|
6 | #endif |
---|
7 | |
---|
8 | #define USE_KAHAN_SUMMATION 0 |
---|
9 | |
---|
10 | // If opencl is not available, then we are compiling a C function |
---|
11 | // Note: if using a C++ compiler, then define kernel as extern "C" |
---|
12 | #ifndef USE_OPENCL |
---|
13 | # ifdef __cplusplus |
---|
14 | #include <cstdio> |
---|
15 | #include <cmath> |
---|
16 | using namespace std; |
---|
17 | #if defined(_MSC_VER) |
---|
18 | #include <limits> |
---|
19 | #include <float.h> |
---|
20 | #define kernel extern "C" __declspec( dllexport ) |
---|
21 | inline double trunc(double x) { return x>=0?floor(x):-floor(-x); } |
---|
22 | inline double fmin(double x, double y) { return x>y ? y : x; } |
---|
23 | inline double fmax(double x, double y) { return x<y ? y : x; } |
---|
24 | inline double isnan(double x) { return _isnan(x); } |
---|
25 | #define NAN (std::numeric_limits<double>::quiet_NaN()) // non-signalling NaN |
---|
26 | static double cephes_expm1(double x) { |
---|
27 | // Adapted from the cephes math library. |
---|
28 | // Copyright 1984 - 1992 by Stephen L. Moshier |
---|
29 | if (x != x || x == 0.0) { |
---|
30 | return x; // NaN and +/- 0 |
---|
31 | } else if (x < -0.5 || x > 0.5) { |
---|
32 | return exp(x) - 1.0; |
---|
33 | } else { |
---|
34 | const double xsq = x*x; |
---|
35 | const double p = ((( |
---|
36 | +1.2617719307481059087798E-4)*xsq |
---|
37 | +3.0299440770744196129956E-2)*xsq |
---|
38 | +9.9999999999999999991025E-1); |
---|
39 | const double q = (((( |
---|
40 | +3.0019850513866445504159E-6)*xsq |
---|
41 | +2.5244834034968410419224E-3)*xsq |
---|
42 | +2.2726554820815502876593E-1)*xsq |
---|
43 | +2.0000000000000000000897E0); |
---|
44 | double r = x * p; |
---|
45 | r = r / (q - r); |
---|
46 | return r+r; |
---|
47 | } |
---|
48 | } |
---|
49 | #define expm1 cephes_expm1 |
---|
50 | #else |
---|
51 | #define kernel extern "C" |
---|
52 | #endif |
---|
53 | inline void SINCOS(double angle, double &svar, double &cvar) { svar=sin(angle); cvar=cos(angle); } |
---|
54 | # else |
---|
55 | #include <stdio.h> |
---|
56 | #include <tgmath.h> // C99 type-generic math, so sin(float) => sinf |
---|
57 | // MSVC doesn't support C99, so no need for dllexport on C99 branch |
---|
58 | #define kernel |
---|
59 | #define SINCOS(angle,svar,cvar) do {const double _t_=angle; svar=sin(_t_);cvar=cos(_t_);} while (0) |
---|
60 | # endif |
---|
61 | # define global |
---|
62 | # define local |
---|
63 | # define constant const |
---|
64 | // OpenCL powr(a,b) = C99 pow(a,b), b >= 0 |
---|
65 | // OpenCL pown(a,b) = C99 pow(a,b), b integer |
---|
66 | # define powr(a,b) pow(a,b) |
---|
67 | # define pown(a,b) pow(a,b) |
---|
68 | #else |
---|
69 | # if defined(USE_SINCOS) |
---|
70 | # define SINCOS(angle,svar,cvar) svar=sincos(angle,&cvar) |
---|
71 | # else |
---|
72 | # define SINCOS(angle,svar,cvar) do {const double _t_=angle; svar=sin(_t_);cvar=cos(_t_);} while (0) |
---|
73 | # endif |
---|
74 | #endif |
---|
75 | |
---|
76 | // Standard mathematical constants: |
---|
77 | // M_E, M_LOG2E, M_LOG10E, M_LN2, M_LN10, M_PI, M_PI_2=pi/2, M_PI_4=pi/4, |
---|
78 | // M_1_PI=1/pi, M_2_PI=2/pi, M_2_SQRTPI=2/sqrt(pi), SQRT2, SQRT1_2=sqrt(1/2) |
---|
79 | // OpenCL defines M_constant_F for float constants, and nothing if double |
---|
80 | // is not enabled on the card, which is why these constants may be missing |
---|
81 | #ifndef M_PI |
---|
82 | # define M_PI 3.141592653589793 |
---|
83 | #endif |
---|
84 | #ifndef M_PI_2 |
---|
85 | # define M_PI_2 1.570796326794897 |
---|
86 | #endif |
---|
87 | #ifndef M_PI_4 |
---|
88 | # define M_PI_4 0.7853981633974483 |
---|
89 | #endif |
---|
90 | #ifndef M_E |
---|
91 | # define M_E 2.718281828459045091 |
---|
92 | #endif |
---|
93 | |
---|
94 | // Non-standard function library |
---|
95 | // pi/180, used for converting between degrees and radians |
---|
96 | // 4/3 pi for computing sphere volumes |
---|
97 | // square and cube for computing squares and cubes |
---|
98 | #ifndef M_PI_180 |
---|
99 | # define M_PI_180 0.017453292519943295 |
---|
100 | #endif |
---|
101 | #ifndef M_4PI_3 |
---|
102 | # define M_4PI_3 4.18879020478639 |
---|
103 | #endif |
---|
104 | //inline double square(double x) { return pow(x,2.0); } |
---|
105 | //inline double square(double x) { return pown(x,2); } |
---|
106 | inline double square(double x) { return x*x; } |
---|
107 | inline double cube(double x) { return x*x*x; } |
---|
108 | inline double sinc(double x) { return x==0 ? 1.0 : sin(x)/x; } |
---|
109 | |
---|
110 | |
---|
111 | %(DEFINES)s |
---|
112 | |
---|
113 | %(SOURCES)s |
---|
114 | |
---|
115 | /* |
---|
116 | ########################################################## |
---|
117 | # # |
---|
118 | # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! # |
---|
119 | # !! !! # |
---|
120 | # !! KEEP THIS CODE CONSISTENT WITH KERNELPY.PY !! # |
---|
121 | # !! !! # |
---|
122 | # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! # |
---|
123 | # # |
---|
124 | ########################################################## |
---|
125 | */ |
---|
126 | |
---|
127 | #ifdef IQ_KERNEL_NAME |
---|
128 | kernel void IQ_KERNEL_NAME( |
---|
129 | global const double *q, |
---|
130 | global double *result, |
---|
131 | const int Nq, |
---|
132 | #ifdef IQ_OPEN_LOOPS |
---|
133 | #ifdef USE_OPENCL |
---|
134 | global double *loops_g, |
---|
135 | #endif |
---|
136 | local double *loops, |
---|
137 | const double cutoff, |
---|
138 | IQ_DISPERSION_LENGTH_DECLARATIONS, |
---|
139 | #endif |
---|
140 | IQ_FIXED_PARAMETER_DECLARATIONS |
---|
141 | ) |
---|
142 | { |
---|
143 | #ifdef USE_OPENCL |
---|
144 | #ifdef IQ_OPEN_LOOPS |
---|
145 | // copy loops info to local memory |
---|
146 | event_t e = async_work_group_copy(loops, loops_g, (IQ_DISPERSION_LENGTH_SUM)*2, 0); |
---|
147 | wait_group_events(1, &e); |
---|
148 | #endif |
---|
149 | |
---|
150 | int i = get_global_id(0); |
---|
151 | if (i < Nq) |
---|
152 | #else |
---|
153 | #pragma omp parallel for |
---|
154 | for (int i=0; i < Nq; i++) |
---|
155 | #endif |
---|
156 | { |
---|
157 | const double qi = q[i]; |
---|
158 | #ifdef IQ_OPEN_LOOPS |
---|
159 | double ret=0.0, norm=0.0; |
---|
160 | IQ_OPEN_LOOPS |
---|
161 | //for (int radius_i=0; radius_i < Nradius; radius_i++) { |
---|
162 | // const double radius = loops[2*(radius_i)]; |
---|
163 | // const double radius_w = loops[2*(radius_i)+1]; |
---|
164 | |
---|
165 | const double weight = IQ_WEIGHT_PRODUCT; |
---|
166 | if (weight > cutoff) { |
---|
167 | const double scattering = Iq(qi, IQ_PARAMETERS); |
---|
168 | // allow kernels to exclude invalid regions by returning NaN |
---|
169 | if (!isnan(scattering)) { |
---|
170 | ret += weight*scattering; |
---|
171 | #ifdef VOLUME_PARAMETERS |
---|
172 | norm += weight * form_volume(VOLUME_PARAMETERS); |
---|
173 | #else |
---|
174 | norm += weight; |
---|
175 | #endif |
---|
176 | } |
---|
177 | //else { printf("exclude qx,qy,I:%%g,%%g,%%g\n",qi,scattering); } |
---|
178 | } |
---|
179 | IQ_CLOSE_LOOPS |
---|
180 | // norm can only be zero if volume is zero, so no scattering |
---|
181 | result[i] = (norm > 0. ? scale*ret/norm + background : background); |
---|
182 | #else |
---|
183 | result[i] = scale*Iq(qi, IQ_PARAMETERS) + background; |
---|
184 | #endif |
---|
185 | } |
---|
186 | } |
---|
187 | #endif |
---|
188 | |
---|
189 | |
---|
190 | #ifdef IQXY_KERNEL_NAME |
---|
191 | kernel void IQXY_KERNEL_NAME( |
---|
192 | global const double *qx, |
---|
193 | global const double *qy, |
---|
194 | global double *result, |
---|
195 | const int Nq, |
---|
196 | #ifdef IQXY_OPEN_LOOPS |
---|
197 | #ifdef USE_OPENCL |
---|
198 | global double *loops_g, |
---|
199 | #endif |
---|
200 | local double *loops, |
---|
201 | const double cutoff, |
---|
202 | IQXY_DISPERSION_LENGTH_DECLARATIONS, |
---|
203 | #endif |
---|
204 | IQXY_FIXED_PARAMETER_DECLARATIONS |
---|
205 | ) |
---|
206 | { |
---|
207 | #ifdef USE_OPENCL |
---|
208 | #ifdef IQXY_OPEN_LOOPS |
---|
209 | // copy loops info to local memory |
---|
210 | event_t e = async_work_group_copy(loops, loops_g, (IQXY_DISPERSION_LENGTH_SUM)*2, 0); |
---|
211 | wait_group_events(1, &e); |
---|
212 | #endif |
---|
213 | |
---|
214 | int i = get_global_id(0); |
---|
215 | if (i < Nq) |
---|
216 | #else |
---|
217 | #pragma omp parallel for |
---|
218 | for (int i=0; i < Nq; i++) |
---|
219 | #endif |
---|
220 | { |
---|
221 | const double qxi = qx[i]; |
---|
222 | const double qyi = qy[i]; |
---|
223 | #if USE_KAHAN_SUMMATION |
---|
224 | double accumulated_error = 0.0; |
---|
225 | #endif |
---|
226 | #ifdef IQXY_OPEN_LOOPS |
---|
227 | double ret=0.0, norm=0.0; |
---|
228 | IQXY_OPEN_LOOPS |
---|
229 | //for (int radius_i=0; radius_i < Nradius; radius_i++) { |
---|
230 | // const double radius = loops[2*(radius_i)]; |
---|
231 | // const double radius_w = loops[2*(radius_i)+1]; |
---|
232 | double weight = IQXY_WEIGHT_PRODUCT; |
---|
233 | if (weight > cutoff) { |
---|
234 | |
---|
235 | const double scattering = Iqxy(qxi, qyi, IQXY_PARAMETERS); |
---|
236 | if (!isnan(scattering)) { // if scattering is bad, exclude it from sum |
---|
237 | #if defined(IQXY_HAS_THETA) |
---|
238 | // Force a nominal value for the spherical correction even when |
---|
239 | // theta is +90/-90 so that there are no divide by zero problems. |
---|
240 | // For cos(theta) fixed at 90, we effectively multiply top and bottom |
---|
241 | // by 1e-6, so the effect cancels. |
---|
242 | const double spherical_correction = fmax(fabs(cos(M_PI_180*theta)), 1.e-6); |
---|
243 | weight *= spherical_correction; |
---|
244 | #endif |
---|
245 | const double next = weight * scattering; |
---|
246 | #if USE_KAHAN_SUMMATION |
---|
247 | const double y = next - accumulated_error; |
---|
248 | const double t = ret + y; |
---|
249 | accumulated_error = (t - ret) - y; |
---|
250 | ret = t; |
---|
251 | #else |
---|
252 | ret += next; |
---|
253 | #endif |
---|
254 | #ifdef VOLUME_PARAMETERS |
---|
255 | norm += weight*form_volume(VOLUME_PARAMETERS); |
---|
256 | #else |
---|
257 | norm += weight; |
---|
258 | #endif |
---|
259 | } |
---|
260 | //else { printf("exclude qx,qy,I:%%g,%%g,%%g\n",qi,scattering); } |
---|
261 | } |
---|
262 | IQXY_CLOSE_LOOPS |
---|
263 | // norm can only be zero if volume is zero, so no scattering |
---|
264 | result[i] = (norm>0. ? scale*ret/norm + background : background); |
---|
265 | #else |
---|
266 | result[i] = scale*Iqxy(qxi, qyi, IQXY_PARAMETERS) + background; |
---|
267 | #endif |
---|
268 | } |
---|
269 | } |
---|
270 | #endif |
---|