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
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
|
/******************************************************************************
* SOFA, Simulation Open-Framework Architecture, version 1.0 beta 4 *
* (c) 2006-2009 MGH, INRIA, USTL, UJF, CNRS *
* *
* This library is free software; you can redistribute it and/or modify it *
* under the terms of the GNU Lesser General Public License as published by *
* the Free Software Foundation; either version 2.1 of the License, or (at *
* your option) any later version. *
* *
* This library 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 Lesser General Public License *
* for more details. *
* *
* You should have received a copy of the GNU Lesser General Public License *
* along with this library; if not, write to the Free Software Foundation, *
* Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. *
*******************************************************************************
* SOFA :: Modules *
* *
* Authors: The SOFA Team and external contributors (see Authors.txt) *
* *
* Contact information: contact@sofa-framework.org *
******************************************************************************/
#include "CudaCommon.h"
#include "CudaMath.h"
#include "CudaTexture.h"
#include "cuda.h"
//#define umul24(x,y) ((x)*(y))
#if defined(__cplusplus) && CUDA_VERSION < 2000
namespace sofa
{
namespace gpu
{
namespace cuda
{
#endif
extern "C"
{
void CudaVisualModelCuda3f_calcTNormals(unsigned int nbElem, unsigned int nbVertex, const void* elems, void* fnormals, const void* x);
void CudaVisualModelCuda3f_calcQNormals(unsigned int nbElem, unsigned int nbVertex, const void* elems, void* fnormals, const void* x);
void CudaVisualModelCuda3f_calcVNormals(unsigned int nbElem, unsigned int nbVertex, unsigned int nbElemPerVertex, const void* velems, void* vnormals, const void* fnormals, const void* x);
void CudaVisualModelCuda3f1_calcTNormals(unsigned int nbElem, unsigned int nbVertex, const void* elems, void* fnormals, const void* x);
void CudaVisualModelCuda3f1_calcQNormals(unsigned int nbElem, unsigned int nbVertex, const void* elems, void* fnormals, const void* x);
void CudaVisualModelCuda3f1_calcVNormals(unsigned int nbElem, unsigned int nbVertex, unsigned int nbElemPerVertex, const void* velems, void* vnormals, const void* fnormals, const void* x);
#ifdef SOFA_GPU_CUDA_DOUBLE
void CudaVisualModelCuda3d_calcTNormals(unsigned int nbElem, unsigned int nbVertex, const void* elems, void* fnormals, const void* x);
void CudaVisualModelCuda3d_calcQNormals(unsigned int nbElem, unsigned int nbVertex, const void* elems, void* fnormals, const void* x);
void CudaVisualModelCuda3d_calcVNormals(unsigned int nbElem, unsigned int nbVertex, unsigned int nbElemPerVertex, const void* velems, void* vnormals, const void* fnormals, const void* x);
void CudaVisualModelCuda3d1_calcTNormals(unsigned int nbElem, unsigned int nbVertex, const void* elems, void* fnormals, const void* x);
void CudaVisualModelCuda3d1_calcQNormals(unsigned int nbElem, unsigned int nbVertex, const void* elems, void* fnormals, const void* x);
void CudaVisualModelCuda3d1_calcVNormals(unsigned int nbElem, unsigned int nbVertex, unsigned int nbElemPerVertex, const void* velems, void* vnormals, const void* fnormals, const void* x);
#endif // SOFA_GPU_CUDA_DOUBLE
}
//////////////////////
// GPU-side methods //
//////////////////////
// #define USE_TEXTURE false
#ifdef USE_TEXTURE
#undef USE_TEXTURE
#endif
// no texture is used unless this template is specialized
template<typename real, class TIn>
class CudaCudaVisualModelTextures
{
public:
static __host__ void setX(const void* /*x*/)
{
}
static __inline__ __device__ CudaVec3<real> getX(int i, const TIn* x)
{
return CudaVec3<real>::make(x[i]);
}
static __host__ void setN(const void* /*x*/)
{
}
static __inline__ __device__ CudaVec3<real> getN(int i, const TIn* x)
{
return CudaVec3<real>::make(x[i]);
}
};
#ifdef USE_TEXTURE
static texture<float,1,cudaReadModeElementType> tex_3f_x;
static texture<float,1,cudaReadModeElementType> tex_3f_n;
template<>
class CudaCudaVisualModelTextures<float, CudaVec3<float> >
{
public:
typedef float real;
typedef CudaVec3<real> TIn;
static __host__ void setX(const void* x)
{
static const void* cur = NULL;
if (x!=cur)
{
cudaBindTexture((size_t*)NULL, tex_3f_x, x);
cur = x;
}
}
static __inline__ __device__ CudaVec3<real> getX(int i, const TIn* x)
{
int i3 = umul24(i,3);
float x1 = tex1Dfetch(tex_3f_x, i3);
float x2 = tex1Dfetch(tex_3f_x, i3+1);
float x3 = tex1Dfetch(tex_3f_x, i3+2);
return CudaVec3<real>::make(x1,x2,x3);
}
static __host__ void setN(const void* n)
{
static const void* cur = NULL;
if (n!=cur)
{
cudaBindTexture((size_t*)NULL, tex_3f_n, n);
cur = n;
}
}
static __inline__ __device__ CudaVec3<real> getN(int i, const TIn* n)
{
int i3 = umul24(i,3);
float x1 = tex1Dfetch(tex_3f_n, i3);
float x2 = tex1Dfetch(tex_3f_n, i3+1);
float x3 = tex1Dfetch(tex_3f_n, i3+2);
return CudaVec3<real>::make(x1,x2,x3);
}
};
static texture<float4,1,cudaReadModeElementType> tex_3f1_x;
static texture<float4,1,cudaReadModeElementType> tex_3f1_n;
template<>
class CudaCudaVisualModelTextures<float, CudaVec4<float> >
{
public:
typedef float real;
typedef CudaVec4<real> TIn;
static __host__ void setX(const void* x)
{
static const void* cur = NULL;
if (x!=cur)
{
cudaBindTexture((size_t*)NULL, tex_3f1_x, x);
cur = x;
}
}
static __inline__ __device__ CudaVec3<real> getX(int i, const TIn* x)
{
return CudaVec3<real>::make(tex1Dfetch(tex_3f1_x, i));
}
static __host__ void setN(const void* n)
{
static const void* cur = NULL;
if (n!=cur)
{
cudaBindTexture((size_t*)NULL, tex_3f1_n, n);
cur = n;
}
}
static __inline__ __device__ CudaVec3<real> getN(int i, const TIn* n)
{
return CudaVec3<real>::make(tex1Dfetch(tex_3f1_n, i));
}
};
#endif
template<typename real, class TIn>
__global__ void CudaVisualModelCuda3t_calcTNormals_kernel(int nbElem, const int* elems, real* fnormals, const TIn* x)
{
int index0 = umul24(blockIdx.x,BSIZE); //blockDim.x;
int index1 = threadIdx.x;
int index = index0+index1;
int index3 = umul24(index1,3);
int iext = umul24(blockIdx.x,BSIZE*3)+index1; //index0*3+index1;
__shared__ union {
int itemp[3*BSIZE];
real rtemp[3*BSIZE];
} s;
s.itemp[index1] = elems[iext];
s.itemp[index1+BSIZE] = elems[iext+BSIZE];
s.itemp[index1+2*BSIZE] = elems[iext+2*BSIZE];
__syncthreads();
CudaVec3<real> N = CudaVec3<real>::make(0,0,0);
if (index < nbElem)
{
CudaVec3<real> A = CudaCudaVisualModelTextures<real,TIn>::getX(s.itemp[index3+0], x);
CudaVec3<real> B = CudaCudaVisualModelTextures<real,TIn>::getX(s.itemp[index3+1], x);
CudaVec3<real> C = CudaCudaVisualModelTextures<real,TIn>::getX(s.itemp[index3+2], x);
B -= A;
C -= A;
N = cross(B,C);
N *= invnorm(N);
}
if (sizeof(real) != sizeof(int)) __syncthreads();
s.rtemp[index3+0] = N.x;
s.rtemp[index3+1] = N.y;
s.rtemp[index3+2] = N.z;
__syncthreads();
fnormals[iext] = s.rtemp[index1];
fnormals[iext+BSIZE] = s.rtemp[index1+BSIZE];
fnormals[iext+2*BSIZE] = s.rtemp[index1+2*BSIZE];
}
template<typename real, class TIn>
__global__ void CudaVisualModelCuda3t1_calcTNormals_kernel(int nbElem, const int* elems, CudaVec4<real>* fnormals, const TIn* x)
{
int index0 = umul24(blockIdx.x,BSIZE); //blockDim.x;
int index1 = threadIdx.x;
int index = index0+index1;
int index3 = umul24(index1,3);
int iext = umul24(blockIdx.x,BSIZE*3)+index1; //index0*3+index1;
__shared__ int itemp[3*BSIZE];
itemp[index1] = elems[iext];
itemp[index1+BSIZE] = elems[iext+BSIZE];
itemp[index1+2*BSIZE] = elems[iext+2*BSIZE];
__syncthreads();
CudaVec3<real> N = CudaVec3<real>::make(0,0,0);
if (index < nbElem)
{
CudaVec3<real> A = CudaCudaVisualModelTextures<real,TIn>::getX(itemp[index3+0], x);
CudaVec3<real> B = CudaCudaVisualModelTextures<real,TIn>::getX(itemp[index3+1], x);
CudaVec3<real> C = CudaCudaVisualModelTextures<real,TIn>::getX(itemp[index3+2], x);
B -= A;
C -= A;
N = cross(B,C);
N *= invnorm(N);
}
fnormals[index] = CudaVec4<real>::make(N,0.0f);
}
template<typename real, class TIn>
__global__ void CudaVisualModelCuda3t_calcQNormals_kernel(int nbElem, const int4* elems, real* fnormals, const TIn* x)
{
int index0 = umul24(blockIdx.x,BSIZE); //blockDim.x;
int index1 = threadIdx.x;
int index = index0+index1;
int index3 = umul24(index1,3);
int iext = umul24(blockIdx.x,BSIZE*3)+index1; //index0*3+index1;
__shared__ real rtemp[3*BSIZE];
CudaVec3<real> N = CudaVec3<real>::make(0,0,0);
if (index < nbElem)
{
int4 itemp = elems[index];
CudaVec3<real> A = CudaCudaVisualModelTextures<real,TIn>::getX(itemp.x, x);
CudaVec3<real> B = CudaCudaVisualModelTextures<real,TIn>::getX(itemp.y, x);
CudaVec3<real> C = CudaCudaVisualModelTextures<real,TIn>::getX(itemp.z, x);
CudaVec3<real> D = CudaCudaVisualModelTextures<real,TIn>::getX(itemp.w, x);
C -= A;
D -= B;
N = cross(C,D);
N *= invnorm(N);
}
rtemp[index3+0] = N.x;
rtemp[index3+1] = N.y;
rtemp[index3+2] = N.z;
__syncthreads();
fnormals[iext] = rtemp[index1];
fnormals[iext+BSIZE] = rtemp[index1+BSIZE];
fnormals[iext+2*BSIZE] = rtemp[index1+2*BSIZE];
}
template<typename real, class TIn>
__global__ void CudaVisualModelCuda3t1_calcQNormals_kernel(int nbElem, const int4* elems, CudaVec4<real>* fnormals, const TIn* x)
{
int index0 = umul24(blockIdx.x,BSIZE); //blockDim.x;
int index1 = threadIdx.x;
int index = index0+index1;
CudaVec3<real> N = CudaVec3<real>::make(0,0,0);
if (index < nbElem)
{
int4 itemp = elems[index];
CudaVec3<real> A = CudaCudaVisualModelTextures<real,TIn>::getX(itemp.x, x);
CudaVec3<real> B = CudaCudaVisualModelTextures<real,TIn>::getX(itemp.y, x);
CudaVec3<real> C = CudaCudaVisualModelTextures<real,TIn>::getX(itemp.z, x);
CudaVec3<real> D = CudaCudaVisualModelTextures<real,TIn>::getX(itemp.w, x);
C -= A;
D -= B;
N = cross(C,D);
N *= invnorm(N);
}
fnormals[index] = CudaVec4<real>::make(N,0.0f);
}
template<typename real, class TIn>
__global__ void CudaVisualModelCuda3t_calcVNormals_kernel(int nbVertex, unsigned int nbElemPerVertex, const int* velems, real* vnormals, const TIn* fnormals)
{
int index0 = umul24(blockIdx.x,BSIZE); //blockDim.x;
int index1 = threadIdx.x;
int index3 = umul24(index1,3); //3*index1;
__shared__ real temp[3*BSIZE];
int iext = umul24(blockIdx.x,BSIZE*3)+index1; //index0*3+index1;
CudaVec3<real> n = CudaVec3<real>::make(0.0f,0.0f,0.0f);
velems+=umul24(index0,nbElemPerVertex)+index1;
if (index0+index1 < nbVertex)
{
for (int s = 0;s < nbElemPerVertex; s++)
{
int i = *velems -1;
velems+=BSIZE;
if (i != -1)
{
n += CudaCudaVisualModelTextures<real,TIn>::getN(i,fnormals);
}
}
real invn = invnorm(n);
if (invn < 100000.0)
n *= invn;
}
temp[index3 ] = n.x;
temp[index3+1] = n.y;
temp[index3+2] = n.z;
__syncthreads();
vnormals[iext ] = temp[index1 ];
vnormals[iext+ BSIZE] = temp[index1+ BSIZE];
vnormals[iext+2*BSIZE] = temp[index1+2*BSIZE];
}
template<typename real, class TIn>
__global__ void CudaVisualModelCuda3t1_calcVNormals_kernel(int nbVertex, unsigned int nbElemPerVertex, const int* velems, CudaVec4<real>* vnormals, const TIn* fnormals)
{
const int index0 = umul24(blockIdx.x,BSIZE); //blockDim.x;
const int index1 = threadIdx.x;
const int index = index0 + index1;
CudaVec3<real> n = CudaVec3<real>::make(0.0f,0.0f,0.0f);
velems+=umul24(index0,nbElemPerVertex)+index1;
if (index < nbVertex)
{
for (int s = 0;s < nbElemPerVertex; s++)
{
int i = *velems -1;
velems+=BSIZE;
if (i != -1)
{
n += CudaCudaVisualModelTextures<real,TIn>::getN(i,fnormals);
}
}
real invn = invnorm(n);
if (invn < 100000.0)
n *= invn;
}
vnormals[index] = CudaVec4<real>::make(n,0.0f);
}
//////////////////////
// CPU-side methods //
//////////////////////
void CudaVisualModelCuda3f_calcTNormals(unsigned int nbElem, unsigned int nbVertex, const void* elems, void* fnormals, const void* x)
{
CudaCudaVisualModelTextures<float,CudaVec3<float> >::setX(x);
dim3 threads1(BSIZE,1);
dim3 grid1((nbElem+BSIZE-1)/BSIZE,1);
CudaVisualModelCuda3t_calcTNormals_kernel<float, CudaVec3<float> ><<< grid1, threads1 >>>(nbElem, (const int*)elems, (float*)fnormals, (const CudaVec3<float>*)x);
}
void CudaVisualModelCuda3f_calcQNormals(unsigned int nbElem, unsigned int nbVertex, const void* elems, void* fnormals, const void* x)
{
CudaCudaVisualModelTextures<float,CudaVec3<float> >::setX(x);
dim3 threads1(BSIZE,1);
dim3 grid1((nbElem+BSIZE-1)/BSIZE,1);
CudaVisualModelCuda3t_calcQNormals_kernel<float, CudaVec3<float> ><<< grid1, threads1 >>>(nbElem, (const int4*)elems, (float*)fnormals, (const CudaVec3<float>*)x);
}
void CudaVisualModelCuda3f_calcVNormals(unsigned int nbElem, unsigned int nbVertex, unsigned int nbElemPerVertex, const void* velems, void* vnormals, const void* fnormals, const void* x)
{
dim3 threads2(BSIZE,1);
dim3 grid2((nbVertex+BSIZE-1)/BSIZE,1);
CudaVisualModelCuda3t_calcVNormals_kernel<float, CudaVec3<float> ><<< grid2, threads2 >>>(nbVertex, nbElemPerVertex, (const int*)velems, (float*)vnormals, (const CudaVec3<float>*)fnormals);
}
void CudaVisualModelCuda3f1_calcTNormals(unsigned int nbElem, unsigned int nbVertex, const void* elems, void* fnormals, const void* x)
{
CudaCudaVisualModelTextures<float,CudaVec4<float> >::setX(x);
dim3 threads1(BSIZE,1);
dim3 grid1((nbElem+BSIZE-1)/BSIZE,1);
CudaVisualModelCuda3t1_calcTNormals_kernel<float, CudaVec4<float> ><<< grid1, threads1 >>>(nbElem, (const int*)elems, (CudaVec4<float>*)fnormals, (const CudaVec4<float>*)x);
}
void CudaVisualModelCuda3f1_calcQNormals(unsigned int nbElem, unsigned int nbVertex, const void* elems, void* fnormals, const void* x)
{
CudaCudaVisualModelTextures<float,CudaVec4<float> >::setX(x);
dim3 threads1(BSIZE,1);
dim3 grid1((nbElem+BSIZE-1)/BSIZE,1);
CudaVisualModelCuda3t1_calcQNormals_kernel<float, CudaVec4<float> ><<< grid1, threads1 >>>(nbElem, (const int4*)elems, (CudaVec4<float>*)fnormals, (const CudaVec4<float>*)x);
}
void CudaVisualModelCuda3f1_calcVNormals(unsigned int nbElem, unsigned int nbVertex, unsigned int nbElemPerVertex, const void* velems, void* vnormals, const void* fnormals, const void* x)
{
dim3 threads2(BSIZE,1);
dim3 grid2((nbVertex+BSIZE-1)/BSIZE,1);
CudaVisualModelCuda3t1_calcVNormals_kernel<float, CudaVec4<float> ><<< grid2, threads2 >>>(nbVertex, nbElemPerVertex, (const int*)velems, (CudaVec4<float>*)vnormals, (const CudaVec4<float>*)fnormals);
}
#ifdef SOFA_GPU_CUDA_DOUBLE
void CudaVisualModelCuda3d_calcTNormals(unsigned int nbElem, unsigned int nbVertex, const void* elems, void* fnormals, const void* x)
{
CudaCudaVisualModelTextures<double,CudaVec3<double> >::setX(x);
dim3 threads1(BSIZE,1);
dim3 grid1((nbElem+BSIZE-1)/BSIZE,1);
CudaVisualModelCuda3t_calcTNormals_kernel<double, CudaVec3<double> ><<< grid1, threads1 >>>(nbElem, (const int*)elems, (double*)fnormals, (const CudaVec3<double>*)x);
}
void CudaVisualModelCuda3d_calcQNormals(unsigned int nbElem, unsigned int nbVertex, const void* elems, void* fnormals, const void* x)
{
CudaCudaVisualModelTextures<double,CudaVec3<double> >::setX(x);
dim3 threads1(BSIZE,1);
dim3 grid1((nbElem+BSIZE-1)/BSIZE,1);
CudaVisualModelCuda3t_calcQNormals_kernel<double, CudaVec3<double> ><<< grid1, threads1 >>>(nbElem, (const int4*)elems, (double*)fnormals, (const CudaVec3<double>*)x);
}
void CudaVisualModelCuda3d_calcVNormals(unsigned int nbElem, unsigned int nbVertex, unsigned int nbElemPerVertex, const void* velems, void* vnormals, const void* fnormals, const void* x)
{
dim3 threads2(BSIZE,1);
dim3 grid2((nbVertex+BSIZE-1)/BSIZE,1);
CudaVisualModelCuda3t_calcVNormals_kernel<double, CudaVec3<double> ><<< grid2, threads2 >>>(nbVertex, nbElemPerVertex, (const int*)velems, (double*)vnormals, (const CudaVec3<double>*)fnormals);
}
void CudaVisualModelCuda3d1_calcTNormals(unsigned int nbElem, unsigned int nbVertex, const void* elems, void* fnormals, const void* x)
{
CudaCudaVisualModelTextures<double,CudaVec4<double> >::setX(x);
dim3 threads1(BSIZE,1);
dim3 grid1((nbElem+BSIZE-1)/BSIZE,1);
CudaVisualModelCuda3t1_calcTNormals_kernel<double, CudaVec4<double> ><<< grid1, threads1 >>>(nbElem, (const int*)elems, (CudaVec4<double>*)fnormals, (const CudaVec4<double>*)x);
}
void CudaVisualModelCuda3d1_calcQNormals(unsigned int nbElem, unsigned int nbVertex, const void* elems, void* fnormals, const void* x)
{
CudaCudaVisualModelTextures<double,CudaVec4<double> >::setX(x);
dim3 threads1(BSIZE,1);
dim3 grid1((nbElem+BSIZE-1)/BSIZE,1);
CudaVisualModelCuda3t1_calcQNormals_kernel<double, CudaVec4<double> ><<< grid1, threads1 >>>(nbElem, (const int4*)elems, (CudaVec4<double>*)fnormals, (const CudaVec4<double>*)x);
}
void CudaVisualModelCuda3d1_calcVNormals(unsigned int nbElem, unsigned int nbVertex, unsigned int nbElemPerVertex, const void* velems, void* vnormals, const void* fnormals, const void* x)
{
dim3 threads2(BSIZE,1);
dim3 grid2((nbVertex+BSIZE-1)/BSIZE,1);
CudaVisualModelCuda3t1_calcVNormals_kernel<double, CudaVec4<double> ><<< grid2, threads2 >>>(nbVertex, nbElemPerVertex, (const int*)velems, (CudaVec4<double>*)vnormals, (const CudaVec4<double>*)fnormals);
}
#endif // SOFA_GPU_CUDA_DOUBLE
#if defined(__cplusplus) && CUDA_VERSION < 2000
} // namespace cuda
} // namespace gpu
} // namespace sofa
#endif
|
