QUDA  0.9.0
wilson_pack_twisted_face_core.h
Go to the documentation of this file.
1 // input spinor
2 #ifdef SPINOR_DOUBLE
3 #define spinorFloat double
4 #define i00_re I0.x
5 #define i00_im I0.y
6 #define i01_re I1.x
7 #define i01_im I1.y
8 #define i02_re I2.x
9 #define i02_im I2.y
10 #define i10_re I3.x
11 #define i10_im I3.y
12 #define i11_re I4.x
13 #define i11_im I4.y
14 #define i12_re I5.x
15 #define i12_im I5.y
16 #define i20_re I6.x
17 #define i20_im I6.y
18 #define i21_re I7.x
19 #define i21_im I7.y
20 #define i22_re I8.x
21 #define i22_im I8.y
22 #define i30_re I9.x
23 #define i30_im I9.y
24 #define i31_re I10.x
25 #define i31_im I10.y
26 #define i32_re I11.x
27 #define i32_im I11.y
28 #else
29 #define spinorFloat float
30 #define i00_re I0.x
31 #define i00_im I0.y
32 #define i01_re I0.z
33 #define i01_im I0.w
34 #define i02_re I1.x
35 #define i02_im I1.y
36 #define i10_re I1.z
37 #define i10_im I1.w
38 #define i11_re I2.x
39 #define i11_im I2.y
40 #define i12_re I2.z
41 #define i12_im I2.w
42 #define i20_re I3.x
43 #define i20_im I3.y
44 #define i21_re I3.z
45 #define i21_im I3.w
46 #define i22_re I4.x
47 #define i22_im I4.y
48 #define i30_re I4.z
49 #define i30_im I4.w
50 #define i31_re I5.x
51 #define i31_im I5.y
52 #define i32_re I5.z
53 #define i32_im I5.w
54 #endif // SPINOR_DOUBLE
55 
56 #include "io_spinor.h"
57 
58 if (face_num) {
59 
60  switch(dim) {
61  case 0:
62  {
63  // read spinor from device memory
64 #ifdef TWIST_INV_DSLASH
65 #ifdef SPINOR_DOUBLE
66  const spinorFloat a = param.a;
67  const spinorFloat b = param.b;
68 #else
69  const spinorFloat a = param.a_f;
70  const spinorFloat b = param.b_f;
71 #endif
72 #endif
73  READ_SPINOR(SPINORTEX, param.sp_stride, idx, idx);
74  APPLY_TWIST_INV( a, b, i);
75 
76  spinorFloat a0_re, a0_im;
77  spinorFloat a1_re, a1_im;
78  spinorFloat a2_re, a2_im;
79  spinorFloat b0_re, b0_im;
80  spinorFloat b1_re, b1_im;
81  spinorFloat b2_re, b2_im;
82 
83  // project spinor into half spinors
84  a0_re = +i00_re-i30_im;
85  a0_im = +i00_im+i30_re;
86  a1_re = +i01_re-i31_im;
87  a1_im = +i01_im+i31_re;
88  a2_re = +i02_re-i32_im;
89  a2_im = +i02_im+i32_re;
90  b0_re = +i10_re-i20_im;
91  b0_im = +i10_im+i20_re;
92  b1_re = +i11_re-i21_im;
93  b1_im = +i11_im+i21_re;
94  b2_re = +i12_re-i22_im;
95  b2_im = +i12_im+i22_re;
96 
97  // write half spinor back to device memory
98  WRITE_HALF_SPINOR(face_volume, face_idx);
99  }
100  break;
101  case 1:
102  {
103  // read spinor from device memory
104 #ifdef TWIST_INV_DSLASH
105 #ifdef SPINOR_DOUBLE
106  const spinorFloat a = param.a;
107  const spinorFloat b = param.b;
108 #else
109  const spinorFloat a = param.a_f;
110  const spinorFloat b = param.b_f;
111 #endif
112 #endif
113  READ_SPINOR(SPINORTEX, param.sp_stride, idx, idx);
114  APPLY_TWIST_INV( a, b, i);
115 
116  spinorFloat a0_re, a0_im;
117  spinorFloat a1_re, a1_im;
118  spinorFloat a2_re, a2_im;
119  spinorFloat b0_re, b0_im;
120  spinorFloat b1_re, b1_im;
121  spinorFloat b2_re, b2_im;
122 
123  // project spinor into half spinors
124  a0_re = +i00_re+i30_re;
125  a0_im = +i00_im+i30_im;
126  a1_re = +i01_re+i31_re;
127  a1_im = +i01_im+i31_im;
128  a2_re = +i02_re+i32_re;
129  a2_im = +i02_im+i32_im;
130  b0_re = +i10_re-i20_re;
131  b0_im = +i10_im-i20_im;
132  b1_re = +i11_re-i21_re;
133  b1_im = +i11_im-i21_im;
134  b2_re = +i12_re-i22_re;
135  b2_im = +i12_im-i22_im;
136 
137  // write half spinor back to device memory
138  WRITE_HALF_SPINOR(face_volume, face_idx);
139  }
140  break;
141  case 2:
142  {
143  // read spinor from device memory
144 #ifdef TWIST_INV_DSLASH
145 #ifdef SPINOR_DOUBLE
146  const spinorFloat a = param.a;
147  const spinorFloat b = param.b;
148 #else
149  const spinorFloat a = param.a_f;
150  const spinorFloat b = param.b_f;
151 #endif
152 #endif
153  READ_SPINOR(SPINORTEX, param.sp_stride, idx, idx);
154  APPLY_TWIST_INV( a, b, i);
155 
156  spinorFloat a0_re, a0_im;
157  spinorFloat a1_re, a1_im;
158  spinorFloat a2_re, a2_im;
159  spinorFloat b0_re, b0_im;
160  spinorFloat b1_re, b1_im;
161  spinorFloat b2_re, b2_im;
162 
163  // project spinor into half spinors
164  a0_re = +i00_re-i20_im;
165  a0_im = +i00_im+i20_re;
166  a1_re = +i01_re-i21_im;
167  a1_im = +i01_im+i21_re;
168  a2_re = +i02_re-i22_im;
169  a2_im = +i02_im+i22_re;
170  b0_re = +i10_re+i30_im;
171  b0_im = +i10_im-i30_re;
172  b1_re = +i11_re+i31_im;
173  b1_im = +i11_im-i31_re;
174  b2_re = +i12_re+i32_im;
175  b2_im = +i12_im-i32_re;
176 
177  // write half spinor back to device memory
178  WRITE_HALF_SPINOR(face_volume, face_idx);
179  }
180  break;
181  case 3:
182  {
183  // read spinor from device memory
184 #ifdef TWIST_INV_DSLASH
185 #ifdef SPINOR_DOUBLE
186  const spinorFloat a = param.a;
187  const spinorFloat b = param.b;
188 #else
189  const spinorFloat a = param.a_f;
190  const spinorFloat b = param.b_f;
191 #endif
192 #endif
193  READ_SPINOR(SPINORTEX, param.sp_stride, idx, idx);
194  APPLY_TWIST_INV( a, b, i);
195 
196  spinorFloat a0_re, a0_im;
197  spinorFloat a1_re, a1_im;
198  spinorFloat a2_re, a2_im;
199  spinorFloat b0_re, b0_im;
200  spinorFloat b1_re, b1_im;
201  spinorFloat b2_re, b2_im;
202 
203  // project spinor into half spinors
204  a0_re = +2*i00_re;
205  a0_im = +2*i00_im;
206  a1_re = +2*i01_re;
207  a1_im = +2*i01_im;
208  a2_re = +2*i02_re;
209  a2_im = +2*i02_im;
210  b0_re = +2*i10_re;
211  b0_im = +2*i10_im;
212  b1_re = +2*i11_re;
213  b1_im = +2*i11_im;
214  b2_re = +2*i12_re;
215  b2_im = +2*i12_im;
216 
217  // write half spinor back to device memory
218  WRITE_HALF_SPINOR(face_volume, face_idx);
219  }
220  break;
221  }
222 
223 } else {
224 
225  switch(dim) {
226  case 0:
227  {
228  // read spinor from device memory
229 #ifdef TWIST_INV_DSLASH
230 #ifdef SPINOR_DOUBLE
231  const spinorFloat a = param.a;
232  const spinorFloat b = param.b;
233 #else
234  const spinorFloat a = param.a_f;
235  const spinorFloat b = param.b_f;
236 #endif
237 #endif
238  READ_SPINOR(SPINORTEX, param.sp_stride, idx, idx);
239  APPLY_TWIST_INV( a, b, i);
240 
241  spinorFloat a0_re, a0_im;
242  spinorFloat a1_re, a1_im;
243  spinorFloat a2_re, a2_im;
244  spinorFloat b0_re, b0_im;
245  spinorFloat b1_re, b1_im;
246  spinorFloat b2_re, b2_im;
247 
248  // project spinor into half spinors
249  a0_re = +i00_re+i30_im;
250  a0_im = +i00_im-i30_re;
251  a1_re = +i01_re+i31_im;
252  a1_im = +i01_im-i31_re;
253  a2_re = +i02_re+i32_im;
254  a2_im = +i02_im-i32_re;
255  b0_re = +i10_re+i20_im;
256  b0_im = +i10_im-i20_re;
257  b1_re = +i11_re+i21_im;
258  b1_im = +i11_im-i21_re;
259  b2_re = +i12_re+i22_im;
260  b2_im = +i12_im-i22_re;
261 
262  // write half spinor back to device memory
263  WRITE_HALF_SPINOR(face_volume, face_idx);
264  }
265  break;
266  case 1:
267  {
268  // read spinor from device memory
269 #ifdef TWIST_INV_DSLASH
270 #ifdef SPINOR_DOUBLE
271  const spinorFloat a = param.a;
272  const spinorFloat b = param.b;
273 #else
274  const spinorFloat a = param.a_f;
275  const spinorFloat b = param.b_f;
276 #endif
277 #endif
278  READ_SPINOR(SPINORTEX, param.sp_stride, idx, idx);
279  APPLY_TWIST_INV( a, b, i);
280 
281  spinorFloat a0_re, a0_im;
282  spinorFloat a1_re, a1_im;
283  spinorFloat a2_re, a2_im;
284  spinorFloat b0_re, b0_im;
285  spinorFloat b1_re, b1_im;
286  spinorFloat b2_re, b2_im;
287 
288  // project spinor into half spinors
289  a0_re = +i00_re-i30_re;
290  a0_im = +i00_im-i30_im;
291  a1_re = +i01_re-i31_re;
292  a1_im = +i01_im-i31_im;
293  a2_re = +i02_re-i32_re;
294  a2_im = +i02_im-i32_im;
295  b0_re = +i10_re+i20_re;
296  b0_im = +i10_im+i20_im;
297  b1_re = +i11_re+i21_re;
298  b1_im = +i11_im+i21_im;
299  b2_re = +i12_re+i22_re;
300  b2_im = +i12_im+i22_im;
301 
302  // write half spinor back to device memory
303  WRITE_HALF_SPINOR(face_volume, face_idx);
304  }
305  break;
306  case 2:
307  {
308  // read spinor from device memory
309 #ifdef TWIST_INV_DSLASH
310 #ifdef SPINOR_DOUBLE
311  const spinorFloat a = param.a;
312  const spinorFloat b = param.b;
313 #else
314  const spinorFloat a = param.a_f;
315  const spinorFloat b = param.b_f;
316 #endif
317 #endif
318  READ_SPINOR(SPINORTEX, param.sp_stride, idx, idx);
319  APPLY_TWIST_INV( a, b, i);
320 
321  spinorFloat a0_re, a0_im;
322  spinorFloat a1_re, a1_im;
323  spinorFloat a2_re, a2_im;
324  spinorFloat b0_re, b0_im;
325  spinorFloat b1_re, b1_im;
326  spinorFloat b2_re, b2_im;
327 
328  // project spinor into half spinors
329  a0_re = +i00_re+i20_im;
330  a0_im = +i00_im-i20_re;
331  a1_re = +i01_re+i21_im;
332  a1_im = +i01_im-i21_re;
333  a2_re = +i02_re+i22_im;
334  a2_im = +i02_im-i22_re;
335  b0_re = +i10_re-i30_im;
336  b0_im = +i10_im+i30_re;
337  b1_re = +i11_re-i31_im;
338  b1_im = +i11_im+i31_re;
339  b2_re = +i12_re-i32_im;
340  b2_im = +i12_im+i32_re;
341 
342  // write half spinor back to device memory
343  WRITE_HALF_SPINOR(face_volume, face_idx);
344  }
345  break;
346  case 3:
347  {
348  // read spinor from device memory
349 #ifdef TWIST_INV_DSLASH
350 #ifdef SPINOR_DOUBLE
351  const spinorFloat a = param.a;
352  const spinorFloat b = param.b;
353 #else
354  const spinorFloat a = param.a_f;
355  const spinorFloat b = param.b_f;
356 #endif
357 #endif
358  READ_SPINOR(SPINORTEX, param.sp_stride, idx, idx);
359  APPLY_TWIST_INV( a, b, i);
360 
361  spinorFloat a0_re, a0_im;
362  spinorFloat a1_re, a1_im;
363  spinorFloat a2_re, a2_im;
364  spinorFloat b0_re, b0_im;
365  spinorFloat b1_re, b1_im;
366  spinorFloat b2_re, b2_im;
367 
368  // project spinor into half spinors
369  a0_re = +2*i20_re;
370  a0_im = +2*i20_im;
371  a1_re = +2*i21_re;
372  a1_im = +2*i21_im;
373  a2_re = +2*i22_re;
374  a2_im = +2*i22_im;
375  b0_re = +2*i30_re;
376  b0_im = +2*i30_im;
377  b1_re = +2*i31_re;
378  b1_im = +2*i31_im;
379  b2_re = +2*i32_re;
380  b2_im = +2*i32_im;
381 
382  // write half spinor back to device memory
383  WRITE_HALF_SPINOR(face_volume, face_idx);
384  }
385  break;
386  }
387 
388 }
389 
390 // undefine to prevent warning when precision is changed
391 #undef spinorFloat
392 #undef SHARED_STRIDE
393 
394 #undef i00_re
395 #undef i00_im
396 #undef i01_re
397 #undef i01_im
398 #undef i02_re
399 #undef i02_im
400 #undef i10_re
401 #undef i10_im
402 #undef i11_re
403 #undef i11_im
404 #undef i12_re
405 #undef i12_im
406 #undef i20_re
407 #undef i20_im
408 #undef i21_re
409 #undef i21_im
410 #undef i22_re
411 #undef i22_im
412 #undef i30_re
413 #undef i30_im
414 #undef i31_re
415 #undef i31_im
416 #undef i32_re
417 #undef i32_im
418 
i00_re
#define i00_re
Definition: wilson_pack_twisted_face_core.h:30
i31_re
#define i31_re
Definition: wilson_pack_twisted_face_core.h:50
a1_re
spinorFloat a1_re
Definition: asym_wilson_clover_dslash_dagger_fermi_core.h:929
i21_im
#define i21_im
Definition: wilson_pack_twisted_face_core.h:45
i31_im
#define i31_im
Definition: wilson_pack_twisted_face_core.h:51
i10_im
#define i10_im
Definition: wilson_pack_twisted_face_core.h:37
i00_im
#define i00_im
Definition: wilson_pack_twisted_face_core.h:31
i01_re
#define i01_re
Definition: wilson_pack_twisted_face_core.h:32
APPLY_TWIST_INV
#define APPLY_TWIST_INV(a, b, reg)
**************************only for deg tm:*******************************
Definition: io_spinor.h:1314
a2_im
spinorFloat a2_im
Definition: asym_wilson_clover_dslash_dagger_fermi_core.h:930
b1_im
spinorFloat b1_im
Definition: asym_wilson_clover_dslash_dagger_fermi_core.h:932
dim
static __inline__ dim3 dim3 void size_t cudaStream_t int dim
Definition: CMakeCUDACompilerId.cpp1.ii:15687
i30_im
#define i30_im
Definition: wilson_pack_twisted_face_core.h:49
i21_re
#define i21_re
Definition: wilson_pack_twisted_face_core.h:44
i02_im
#define i02_im
Definition: wilson_pack_twisted_face_core.h:35
i11_re
#define i11_re
Definition: wilson_pack_twisted_face_core.h:38
i12_re
#define i12_re
Definition: wilson_pack_twisted_face_core.h:40
b2_im
spinorFloat b2_im
Definition: asym_wilson_clover_dslash_dagger_fermi_core.h:933
a0_im
spinorFloat a0_im
Definition: asym_wilson_clover_dslash_dagger_fermi_core.h:928
param
QudaGaugeParam param
Definition: pack_test.cpp:17
b
#define b
Definition: dw_dslash4_core.h:83
i01_im
#define i01_im
Definition: wilson_pack_twisted_face_core.h:33
i20_re
#define i20_re
Definition: wilson_pack_twisted_face_core.h:42
b1_re
spinorFloat b1_re
Definition: asym_wilson_clover_dslash_dagger_fermi_core.h:932
i32_im
#define i32_im
Definition: wilson_pack_twisted_face_core.h:53
SPINORTEX
#define SPINORTEX
Definition: contract_core.h:891
fused_exterior_ndeg_tm_dslash_cuda_gen.i
int i
start here
Definition: fused_exterior_ndeg_tm_dslash_cuda_gen.py:816
READ_SPINOR
#define READ_SPINOR
Definition: contract_core.h:886
a1_im
spinorFloat a1_im
Definition: asym_wilson_clover_dslash_dagger_fermi_core.h:929
i32_re
#define i32_re
Definition: wilson_pack_twisted_face_core.h:52
io_spinor.h
i20_im
#define i20_im
Definition: wilson_pack_twisted_face_core.h:43
b0_im
spinorFloat b0_im
Definition: asym_wilson_clover_dslash_dagger_fermi_core.h:931
b0_re
spinorFloat b0_re
Definition: asym_wilson_clover_dslash_dagger_fermi_core.h:931
i12_im
#define i12_im
Definition: wilson_pack_twisted_face_core.h:41
a2_re
spinorFloat a2_re
Definition: asym_wilson_clover_dslash_dagger_fermi_core.h:930
idx
int idx
Definition: staggered_fused_exterior_dslash_core.h:355
face_volume
const int face_volume
Definition: tm_ndeg_fused_exterior_dslash_core.h:199
i22_re
#define i22_re
Definition: wilson_pack_twisted_face_core.h:46
i22_im
#define i22_im
Definition: wilson_pack_twisted_face_core.h:47
face_idx
int face_idx
Definition: dw_dslash4_core.h:198
face_num
const int face_num
Definition: dw_dslash4_core.h:223
i10_re
#define i10_re
Definition: wilson_pack_twisted_face_core.h:36
b2_re
spinorFloat b2_re
Definition: asym_wilson_clover_dslash_dagger_fermi_core.h:933
a
#define a
Definition: dw_dslash4_core.h:82
a0_re
spinorFloat a0_re
Definition: asym_wilson_clover_dslash_dagger_fermi_core.h:928
spinorFloat
#define spinorFloat
Definition: wilson_pack_twisted_face_core.h:29
i02_re
#define i02_re
Definition: wilson_pack_twisted_face_core.h:34
i30_re
#define i30_re
Definition: wilson_pack_twisted_face_core.h:48
i11_im
#define i11_im
Definition: wilson_pack_twisted_face_core.h:39
Generated by   doxygen 1.8.14