Line data Source code
1 : #ifndef AMREX_GPU_LAUNCH_FUNCTS_C_H_
2 : #define AMREX_GPU_LAUNCH_FUNCTS_C_H_
3 : #include <AMReX_Config.H>
4 :
5 : namespace amrex {
6 :
7 : namespace detail {
8 : template <typename F, typename N>
9 : AMREX_GPU_DEVICE
10 : auto call_f (F const& f, N i)
11 : noexcept -> decltype(f(0))
12 : {
13 : f(i);
14 : }
15 :
16 : template <typename F, typename N>
17 : AMREX_GPU_DEVICE
18 : auto call_f (F const& f, N i)
19 : noexcept -> decltype(f(0,Gpu::Handler{}))
20 : {
21 : f(i,Gpu::Handler{});
22 : }
23 :
24 : template <typename F>
25 : AMREX_GPU_DEVICE
26 37891502 : auto call_f (F const& f, int i, int j, int k)
27 : noexcept -> decltype(f(0,0,0))
28 : {
29 37891502 : f(i,j,k);
30 37891502 : }
31 :
32 : template <typename F>
33 : AMREX_GPU_DEVICE
34 : auto call_f (F const& f, int i, int j, int k)
35 : noexcept -> decltype(f(0,0,0,Gpu::Handler{}))
36 : {
37 : f(i,j,k,Gpu::Handler{});
38 : }
39 :
40 : template <typename F, typename T>
41 : AMREX_GPU_DEVICE
42 : auto call_f (F const& f, int i, int j, int k, T n)
43 : noexcept -> decltype(f(0,0,0,0))
44 : {
45 : f(i,j,k,n);
46 : }
47 :
48 : template <typename F, typename T>
49 : AMREX_GPU_DEVICE
50 : auto call_f (F const& f, int i, int j, int k, T n)
51 : noexcept -> decltype(f(0,0,0,0,Gpu::Handler{}))
52 : {
53 : f(i,j,k,n,Gpu::Handler{});
54 : }
55 : }
56 :
57 : template<typename T, typename L>
58 : void launch (T const& n, L&& f) noexcept
59 : {
60 : std::forward<L>(f)(n);
61 : }
62 :
63 : template<int MT, typename T, typename L>
64 : void launch (T const& n, L&& f) noexcept
65 : {
66 : amrex::ignore_unused(MT);
67 : std::forward<L>(f)(n);
68 : }
69 :
70 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
71 : AMREX_ATTRIBUTE_FLATTEN_FOR
72 : void For (T n, L const& f) noexcept
73 : {
74 : for (T i = 0; i < n; ++i) {
75 : detail::call_f(f,i);
76 : }
77 : }
78 :
79 : template <int MT, typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
80 : void For (T n, L&& f) noexcept
81 : {
82 : amrex::ignore_unused(MT);
83 : For(n, std::forward<L>(f));
84 : }
85 :
86 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
87 : void For (Gpu::KernelInfo const&, T n, L&& f) noexcept
88 : {
89 : For(n, std::forward<L>(f));
90 : }
91 :
92 : template <int MT, typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
93 : void For (Gpu::KernelInfo const&, T n, L&& f) noexcept
94 : {
95 : amrex::ignore_unused(MT);
96 : For(n, std::forward<L>(f));
97 : }
98 :
99 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
100 : AMREX_ATTRIBUTE_FLATTEN_FOR
101 : void ParallelFor (T n, L const& f) noexcept
102 : {
103 : AMREX_PRAGMA_SIMD
104 : for (T i = 0; i < n; ++i) {
105 : detail::call_f(f,i);
106 : }
107 : }
108 :
109 : template <int MT, typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
110 : void ParallelFor (T n, L&& f) noexcept
111 : {
112 : amrex::ignore_unused(MT);
113 : ParallelFor(n, std::forward<L>(f));
114 : }
115 :
116 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
117 : void ParallelFor (Gpu::KernelInfo const&, T n, L&& f) noexcept
118 : {
119 : ParallelFor(n, std::forward<L>(f));
120 : }
121 :
122 : template <int MT, typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
123 : void ParallelFor (Gpu::KernelInfo const&, T n, L&& f) noexcept
124 : {
125 : amrex::ignore_unused(MT);
126 : ParallelFor(n, std::forward<L>(f));
127 : }
128 :
129 : template <typename L>
130 : AMREX_ATTRIBUTE_FLATTEN_FOR
131 : void For (Box const& box, L const& f) noexcept
132 : {
133 : const auto lo = amrex::lbound(box);
134 : const auto hi = amrex::ubound(box);
135 : for (int k = lo.z; k <= hi.z; ++k) {
136 : for (int j = lo.y; j <= hi.y; ++j) {
137 : for (int i = lo.x; i <= hi.x; ++i) {
138 : detail::call_f(f,i,j,k);
139 : }}}
140 : }
141 :
142 : template <int MT, typename L>
143 : void For (Box const& box, L&& f) noexcept
144 : {
145 : amrex::ignore_unused(MT);
146 : For(box, std::forward<L>(f));
147 : }
148 :
149 : template <typename L>
150 : void For (Gpu::KernelInfo const&, Box const& box, L&& f) noexcept
151 : {
152 : For(box, std::forward<L>(f));
153 : }
154 :
155 : template <int MT, typename L>
156 : void For (Gpu::KernelInfo const&, Box const& box, L&& f) noexcept
157 : {
158 : amrex::ignore_unused(MT);
159 : For(box, std::forward<L>(f));
160 : }
161 :
162 : template <typename L>
163 : AMREX_ATTRIBUTE_FLATTEN_FOR
164 705801 : void ParallelFor (Box const& box, L const& f) noexcept
165 : {
166 : const auto lo = amrex::lbound(box);
167 : const auto hi = amrex::ubound(box);
168 3034923 : for (int k = lo.z; k <= hi.z; ++k) {
169 10698132 : for (int j = lo.y; j <= hi.y; ++j) {
170 : AMREX_PRAGMA_SIMD
171 46260470 : for (int i = lo.x; i <= hi.x; ++i) {
172 37891502 : detail::call_f(f,i,j,k);
173 : }}}
174 705801 : }
175 :
176 : template <int MT, typename L>
177 : void ParallelFor (Box const& box, L&& f) noexcept
178 : {
179 : amrex::ignore_unused(MT);
180 : ParallelFor(box, std::forward<L>(f));
181 : }
182 :
183 : template <typename L>
184 : void ParallelFor (Gpu::KernelInfo const&, Box const& box, L&& f) noexcept
185 : {
186 : ParallelFor(box, std::forward<L>(f));
187 : }
188 :
189 : template <int MT, typename L>
190 : void ParallelFor (Gpu::KernelInfo const&, Box const& box, L&& f) noexcept
191 : {
192 : amrex::ignore_unused(MT);
193 : ParallelFor(box, std::forward<L>(f));
194 : }
195 :
196 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
197 : AMREX_ATTRIBUTE_FLATTEN_FOR
198 : void For (Box const& box, T ncomp, L const& f) noexcept
199 : {
200 : const auto lo = amrex::lbound(box);
201 : const auto hi = amrex::ubound(box);
202 : for (T n = 0; n < ncomp; ++n) {
203 : for (int k = lo.z; k <= hi.z; ++k) {
204 : for (int j = lo.y; j <= hi.y; ++j) {
205 : for (int i = lo.x; i <= hi.x; ++i) {
206 : detail::call_f(f,i,j,k,n);
207 : }}}
208 : }
209 : }
210 :
211 : template <int MT, typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
212 : void For (Box const& box, T ncomp, L&& f) noexcept
213 : {
214 : amrex::ignore_unused(MT);
215 : For(box, ncomp, std::forward<L>(f));
216 : }
217 :
218 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
219 : void For (Gpu::KernelInfo const&, Box const& box, T ncomp, L&& f) noexcept
220 : {
221 : For(box, ncomp, std::forward<L>(f));
222 : }
223 :
224 : template <int MT, typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
225 : void For (Gpu::KernelInfo const&, Box const& box, T ncomp, L&& f) noexcept
226 : {
227 : amrex::ignore_unused(MT);
228 : For(box, ncomp, std::forward<L>(f));
229 : }
230 :
231 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
232 : AMREX_ATTRIBUTE_FLATTEN_FOR
233 : void ParallelFor (Box const& box, T ncomp, L const& f) noexcept
234 : {
235 : const auto lo = amrex::lbound(box);
236 : const auto hi = amrex::ubound(box);
237 : for (T n = 0; n < ncomp; ++n) {
238 : for (int k = lo.z; k <= hi.z; ++k) {
239 : for (int j = lo.y; j <= hi.y; ++j) {
240 : AMREX_PRAGMA_SIMD
241 : for (int i = lo.x; i <= hi.x; ++i) {
242 : detail::call_f(f,i,j,k,n);
243 : }}}
244 : }
245 : }
246 :
247 : template <int MT, typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
248 : void ParallelFor (Box const& box, T ncomp, L&& f) noexcept
249 : {
250 : amrex::ignore_unused(MT);
251 : ParallelFor(box, ncomp, std::forward<L>(f));
252 : }
253 :
254 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
255 : void ParallelFor (Gpu::KernelInfo const&, Box const& box, T ncomp, L&& f) noexcept
256 : {
257 : ParallelFor(box, ncomp, std::forward<L>(f));
258 : }
259 :
260 : template <int MT, typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
261 : void ParallelFor (Gpu::KernelInfo const&, Box const& box, T ncomp, L&& f) noexcept
262 : {
263 : amrex::ignore_unused(MT);
264 : ParallelFor(box, ncomp, std::forward<L>(f));
265 : }
266 :
267 : template <typename L1, typename L2>
268 : void For (Box const& box1, Box const& box2, L1&& f1, L2&& f2) noexcept
269 : {
270 : For(box1, std::forward<L1>(f1));
271 : For(box2, std::forward<L2>(f2));
272 : }
273 :
274 : template <int MT, typename L1, typename L2>
275 : void For (Box const& box1, Box const& box2, L1&& f1, L2&& f2) noexcept
276 : {
277 : amrex::ignore_unused(MT);
278 : For(box1, std::forward<L1>(f1));
279 : For(box2, std::forward<L2>(f2));
280 : }
281 :
282 : template <typename L1, typename L2>
283 : void For (Gpu::KernelInfo const&, Box const& box1, Box const& box2, L1&& f1, L2&& f2) noexcept
284 : {
285 : For (box1, box2, std::forward<L1>(f1), std::forward<L2>(f2));
286 : }
287 :
288 : template <int MT, typename L1, typename L2>
289 : void For (Gpu::KernelInfo const&, Box const& box1, Box const& box2, L1&& f1, L2&& f2) noexcept
290 : {
291 : amrex::ignore_unused(MT);
292 : For (box1, box2, std::forward<L1>(f1), std::forward<L2>(f2));
293 : }
294 :
295 : template <typename L1, typename L2, typename L3>
296 : void For (Box const& box1, Box const& box2, Box const& box3, L1&& f1, L2&& f2, L3&& f3) noexcept
297 : {
298 : For(box1, std::forward<L1>(f1));
299 : For(box2, std::forward<L2>(f2));
300 : For(box3, std::forward<L3>(f3));
301 : }
302 :
303 : template <int MT, typename L1, typename L2, typename L3>
304 : void For (Box const& box1, Box const& box2, Box const& box3, L1&& f1, L2&& f2, L3&& f3) noexcept
305 : {
306 : amrex::ignore_unused(MT);
307 : For(box1, std::forward<L1>(f1));
308 : For(box2, std::forward<L2>(f2));
309 : For(box3, std::forward<L3>(f3));
310 : }
311 :
312 : template <typename L1, typename L2, typename L3>
313 : void For (Gpu::KernelInfo const&, Box const& box1, Box const& box2, Box const& box3, L1&& f1, L2&& f2, L3&& f3) noexcept
314 : {
315 : For(box1, box2, box3, std::forward<L1>(f1), std::forward<L2>(f2), std::forward<L3>(f3));
316 : }
317 :
318 : template <int MT, typename L1, typename L2, typename L3>
319 : void For (Gpu::KernelInfo const&, Box const& box1, Box const& box2, Box const& box3, L1&& f1, L2&& f2, L3&& f3) noexcept
320 : {
321 : amrex::ignore_unused(MT);
322 : For(box1, box2, box3, std::forward<L1>(f1), std::forward<L2>(f2), std::forward<L3>(f3));
323 : }
324 :
325 : template <typename T1, typename T2, typename L1, typename L2,
326 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
327 : typename M2=std::enable_if_t<std::is_integral_v<T2>> >
328 : void For (Box const& box1, T1 ncomp1, L1&& f1,
329 : Box const& box2, T2 ncomp2, L2&& f2) noexcept
330 : {
331 : For(box1, ncomp1, std::forward<L1>(f1));
332 : For(box2, ncomp2, std::forward<L2>(f2));
333 : }
334 :
335 : template <int MT, typename T1, typename T2, typename L1, typename L2,
336 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
337 : typename M2=std::enable_if_t<std::is_integral_v<T2>> >
338 : void For (Box const& box1, T1 ncomp1, L1&& f1,
339 : Box const& box2, T2 ncomp2, L2&& f2) noexcept
340 : {
341 : amrex::ignore_unused(MT);
342 : For(box1, ncomp1, std::forward<L1>(f1));
343 : For(box2, ncomp2, std::forward<L2>(f2));
344 : }
345 :
346 : template <typename T1, typename T2, typename L1, typename L2,
347 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
348 : typename M2=std::enable_if_t<std::is_integral_v<T2>> >
349 : void For (Gpu::KernelInfo const&,
350 : Box const& box1, T1 ncomp1, L1&& f1,
351 : Box const& box2, T2 ncomp2, L2&& f2) noexcept
352 : {
353 : For(box1,ncomp1,std::forward<L1>(f1),box2,ncomp2,std::forward<L2>(f2));
354 : }
355 :
356 : template <int MT, typename T1, typename T2, typename L1, typename L2,
357 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
358 : typename M2=std::enable_if_t<std::is_integral_v<T2>> >
359 : void For (Gpu::KernelInfo const&,
360 : Box const& box1, T1 ncomp1, L1&& f1,
361 : Box const& box2, T2 ncomp2, L2&& f2) noexcept
362 : {
363 : amrex::ignore_unused(MT);
364 : For(box1,ncomp1,std::forward<L1>(f1),box2,ncomp2,std::forward<L2>(f2));
365 : }
366 :
367 : template <typename T1, typename T2, typename T3, typename L1, typename L2, typename L3,
368 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
369 : typename M2=std::enable_if_t<std::is_integral_v<T2>>,
370 : typename M3=std::enable_if_t<std::is_integral_v<T3>> >
371 : void For (Box const& box1, T1 ncomp1, L1&& f1,
372 : Box const& box2, T2 ncomp2, L2&& f2,
373 : Box const& box3, T3 ncomp3, L3&& f3) noexcept
374 : {
375 : For(box1, ncomp1, std::forward<L1>(f1));
376 : For(box2, ncomp2, std::forward<L2>(f2));
377 : For(box3, ncomp3, std::forward<L3>(f3));
378 : }
379 :
380 : template <int MT, typename T1, typename T2, typename T3, typename L1, typename L2, typename L3,
381 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
382 : typename M2=std::enable_if_t<std::is_integral_v<T2>>,
383 : typename M3=std::enable_if_t<std::is_integral_v<T3>> >
384 : void For (Box const& box1, T1 ncomp1, L1&& f1,
385 : Box const& box2, T2 ncomp2, L2&& f2,
386 : Box const& box3, T3 ncomp3, L3&& f3) noexcept
387 : {
388 : amrex::ignore_unused(MT);
389 : For(box1, ncomp1, std::forward<L1>(f1));
390 : For(box2, ncomp2, std::forward<L2>(f2));
391 : For(box3, ncomp3, std::forward<L3>(f3));
392 : }
393 :
394 : template <typename T1, typename T2, typename T3, typename L1, typename L2, typename L3,
395 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
396 : typename M2=std::enable_if_t<std::is_integral_v<T2>>,
397 : typename M3=std::enable_if_t<std::is_integral_v<T3>> >
398 : void For (Gpu::KernelInfo const&,
399 : Box const& box1, T1 ncomp1, L1&& f1,
400 : Box const& box2, T2 ncomp2, L2&& f2,
401 : Box const& box3, T3 ncomp3, L3&& f3) noexcept
402 : {
403 : For(box1,ncomp1,std::forward<L1>(f1),
404 : box2,ncomp2,std::forward<L2>(f2),
405 : box3,ncomp3,std::forward<L3>(f3));
406 : }
407 :
408 : template <int MT, typename T1, typename T2, typename T3, typename L1, typename L2, typename L3,
409 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
410 : typename M2=std::enable_if_t<std::is_integral_v<T2>>,
411 : typename M3=std::enable_if_t<std::is_integral_v<T3>> >
412 : void For (Gpu::KernelInfo const&,
413 : Box const& box1, T1 ncomp1, L1&& f1,
414 : Box const& box2, T2 ncomp2, L2&& f2,
415 : Box const& box3, T3 ncomp3, L3&& f3) noexcept
416 : {
417 : amrex::ignore_unused(MT);
418 : For(box1,ncomp1,std::forward<L1>(f1),
419 : box2,ncomp2,std::forward<L2>(f2),
420 : box3,ncomp3,std::forward<L3>(f3));
421 : }
422 :
423 : template <typename L1, typename L2>
424 : void ParallelFor (Box const& box1, Box const& box2, L1&& f1, L2&& f2) noexcept
425 : {
426 : ParallelFor(box1, std::forward<L1>(f1));
427 : ParallelFor(box2, std::forward<L2>(f2));
428 : }
429 :
430 : template <int MT, typename L1, typename L2>
431 : void ParallelFor (Box const& box1, Box const& box2, L1&& f1, L2&& f2) noexcept
432 : {
433 : amrex::ignore_unused(MT);
434 : ParallelFor(box1, std::forward<L1>(f1));
435 : ParallelFor(box2, std::forward<L2>(f2));
436 : }
437 :
438 : template <typename L1, typename L2>
439 : void ParallelFor (Gpu::KernelInfo const&, Box const& box1, Box const& box2, L1&& f1, L2&& f2) noexcept
440 : {
441 : ParallelFor(box1,box2,std::forward<L1>(f1),std::forward<L2>(f2));
442 : }
443 :
444 : template <int MT, typename L1, typename L2>
445 : void ParallelFor (Gpu::KernelInfo const&, Box const& box1, Box const& box2, L1&& f1, L2&& f2) noexcept
446 : {
447 : amrex::ignore_unused(MT);
448 : ParallelFor(box1,box2,std::forward<L1>(f1),std::forward<L2>(f2));
449 : }
450 :
451 : template <typename L1, typename L2, typename L3>
452 : void ParallelFor (Box const& box1, Box const& box2, Box const& box3, L1&& f1, L2&& f2, L3&& f3) noexcept
453 : {
454 : ParallelFor(box1, std::forward<L1>(f1));
455 : ParallelFor(box2, std::forward<L2>(f2));
456 : ParallelFor(box3, std::forward<L3>(f3));
457 : }
458 :
459 : template <int MT, typename L1, typename L2, typename L3>
460 : void ParallelFor (Box const& box1, Box const& box2, Box const& box3, L1&& f1, L2&& f2, L3&& f3) noexcept
461 : {
462 : amrex::ignore_unused(MT);
463 : ParallelFor(box1, std::forward<L1>(f1));
464 : ParallelFor(box2, std::forward<L2>(f2));
465 : ParallelFor(box3, std::forward<L3>(f3));
466 : }
467 :
468 : template <typename L1, typename L2, typename L3>
469 : void ParallelFor (Gpu::KernelInfo const&, Box const& box1, Box const& box2, Box const& box3, L1&& f1, L2&& f2, L3&& f3) noexcept
470 : {
471 : ParallelFor(box1,box2,box3,std::forward<L1>(f1),std::forward<L2>(f2),std::forward<L3>(f3));
472 : }
473 :
474 : template <int MT, typename L1, typename L2, typename L3>
475 : void ParallelFor (Gpu::KernelInfo const&, Box const& box1, Box const& box2, Box const& box3, L1&& f1, L2&& f2, L3&& f3) noexcept
476 : {
477 : amrex::ignore_unused(MT);
478 : ParallelFor(box1,box2,box3,std::forward<L1>(f1),std::forward<L2>(f2),std::forward<L3>(f3));
479 : }
480 :
481 : template <typename T1, typename T2, typename L1, typename L2,
482 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
483 : typename M2=std::enable_if_t<std::is_integral_v<T2>> >
484 : void ParallelFor (Box const& box1, T1 ncomp1, L1&& f1,
485 : Box const& box2, T2 ncomp2, L2&& f2) noexcept
486 : {
487 : ParallelFor(box1, ncomp1, std::forward<L1>(f1));
488 : ParallelFor(box2, ncomp2, std::forward<L2>(f2));
489 : }
490 :
491 : template <int MT, typename T1, typename T2, typename L1, typename L2,
492 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
493 : typename M2=std::enable_if_t<std::is_integral_v<T2>> >
494 : void ParallelFor (Box const& box1, T1 ncomp1, L1&& f1,
495 : Box const& box2, T2 ncomp2, L2&& f2) noexcept
496 : {
497 : amrex::ignore_unused(MT);
498 : ParallelFor(box1, ncomp1, std::forward<L1>(f1));
499 : ParallelFor(box2, ncomp2, std::forward<L2>(f2));
500 : }
501 :
502 : template <typename T1, typename T2, typename L1, typename L2,
503 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
504 : typename M2=std::enable_if_t<std::is_integral_v<T2>> >
505 : void ParallelFor (Gpu::KernelInfo const&,
506 : Box const& box1, T1 ncomp1, L1&& f1,
507 : Box const& box2, T2 ncomp2, L2&& f2) noexcept
508 : {
509 : ParallelFor(box1,ncomp1,std::forward<L1>(f1),
510 : box2,ncomp2,std::forward<L2>(f2));
511 : }
512 :
513 : template <int MT, typename T1, typename T2, typename L1, typename L2,
514 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
515 : typename M2=std::enable_if_t<std::is_integral_v<T2>> >
516 : void ParallelFor (Gpu::KernelInfo const&,
517 : Box const& box1, T1 ncomp1, L1&& f1,
518 : Box const& box2, T2 ncomp2, L2&& f2) noexcept
519 : {
520 : amrex::ignore_unused(MT);
521 : ParallelFor(box1,ncomp1,std::forward<L1>(f1),
522 : box2,ncomp2,std::forward<L2>(f2));
523 : }
524 :
525 : template <typename T1, typename T2, typename T3, typename L1, typename L2, typename L3,
526 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
527 : typename M2=std::enable_if_t<std::is_integral_v<T2>>,
528 : typename M3=std::enable_if_t<std::is_integral_v<T3>> >
529 : void ParallelFor (Box const& box1, T1 ncomp1, L1&& f1,
530 : Box const& box2, T2 ncomp2, L2&& f2,
531 : Box const& box3, T3 ncomp3, L3&& f3) noexcept
532 : {
533 : ParallelFor(box1, ncomp1, std::forward<L1>(f1));
534 : ParallelFor(box2, ncomp2, std::forward<L2>(f2));
535 : ParallelFor(box3, ncomp3, std::forward<L3>(f3));
536 : }
537 :
538 : template <int MT, typename T1, typename T2, typename T3, typename L1, typename L2, typename L3,
539 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
540 : typename M2=std::enable_if_t<std::is_integral_v<T2>>,
541 : typename M3=std::enable_if_t<std::is_integral_v<T3>> >
542 : void ParallelFor (Box const& box1, T1 ncomp1, L1&& f1,
543 : Box const& box2, T2 ncomp2, L2&& f2,
544 : Box const& box3, T3 ncomp3, L3&& f3) noexcept
545 : {
546 : amrex::ignore_unused(MT);
547 : ParallelFor(box1, ncomp1, std::forward<L1>(f1));
548 : ParallelFor(box2, ncomp2, std::forward<L2>(f2));
549 : ParallelFor(box3, ncomp3, std::forward<L3>(f3));
550 : }
551 :
552 : template <typename T1, typename T2, typename T3, typename L1, typename L2, typename L3,
553 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
554 : typename M2=std::enable_if_t<std::is_integral_v<T2>>,
555 : typename M3=std::enable_if_t<std::is_integral_v<T3>> >
556 : void ParallelFor (Gpu::KernelInfo const&,
557 : Box const& box1, T1 ncomp1, L1&& f1,
558 : Box const& box2, T2 ncomp2, L2&& f2,
559 : Box const& box3, T3 ncomp3, L3&& f3) noexcept
560 : {
561 : ParallelFor(box1, ncomp1, std::forward<L1>(f1),
562 : box2, ncomp2, std::forward<L2>(f2),
563 : box3, ncomp3, std::forward<L3>(f3));
564 : }
565 :
566 : template <int MT, typename T1, typename T2, typename T3, typename L1, typename L2, typename L3,
567 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
568 : typename M2=std::enable_if_t<std::is_integral_v<T2>>,
569 : typename M3=std::enable_if_t<std::is_integral_v<T3>> >
570 : void ParallelFor (Gpu::KernelInfo const&,
571 : Box const& box1, T1 ncomp1, L1&& f1,
572 : Box const& box2, T2 ncomp2, L2&& f2,
573 : Box const& box3, T3 ncomp3, L3&& f3) noexcept
574 : {
575 : amrex::ignore_unused(MT);
576 : ParallelFor(box1, ncomp1, std::forward<L1>(f1),
577 : box2, ncomp2, std::forward<L2>(f2),
578 : box3, ncomp3, std::forward<L3>(f3));
579 : }
580 :
581 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
582 : void HostDeviceParallelFor (T n, L&& f) noexcept
583 : {
584 : ParallelFor(n,std::forward<L>(f));
585 : }
586 :
587 : template <int MT, typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
588 : void HostDeviceParallelFor (T n, L&& f) noexcept
589 : {
590 : amrex::ignore_unused(MT);
591 : ParallelFor(n,std::forward<L>(f));
592 : }
593 :
594 : template <typename L>
595 : void HostDeviceParallelFor (Box const& box, L&& f) noexcept
596 : {
597 : ParallelFor(box,std::forward<L>(f));
598 : }
599 :
600 : template <int MT, typename L>
601 : void HostDeviceParallelFor (Box const& box, L&& f) noexcept
602 : {
603 : amrex::ignore_unused(MT);
604 : ParallelFor(box,std::forward<L>(f));
605 : }
606 :
607 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
608 : void HostDeviceParallelFor (Box const& box, T ncomp, L&& f) noexcept
609 : {
610 : ParallelFor(box,ncomp,std::forward<L>(f));
611 : }
612 :
613 : template <int MT, typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
614 : void HostDeviceParallelFor (Box const& box, T ncomp, L&& f) noexcept
615 : {
616 : amrex::ignore_unused(MT);
617 : ParallelFor(box,ncomp,std::forward<L>(f));
618 : }
619 :
620 : template <typename L1, typename L2>
621 : void HostDeviceParallelFor (Box const& box1, Box const& box2, L1&& f1, L2&& f2) noexcept
622 : {
623 : ParallelFor(box1,box2,std::forward<L1>(f1),std::forward<L2>(f2));
624 : }
625 :
626 : template <int MT, typename L1, typename L2>
627 : void HostDeviceParallelFor (Box const& box1, Box const& box2, L1&& f1, L2&& f2) noexcept
628 : {
629 : amrex::ignore_unused(MT);
630 : ParallelFor(box1,box2,std::forward<L1>(f1),std::forward<L2>(f2));
631 : }
632 :
633 : template <typename L1, typename L2, typename L3>
634 : void HostDeviceParallelFor (Box const& box1, Box const& box2, Box const& box3,
635 : L1&& f1, L2&& f2, L3&& f3) noexcept
636 : {
637 : ParallelFor(box1,box2,box3,std::forward<L1>(f1),std::forward<L2>(f2),std::forward<L3>(f3));
638 : }
639 :
640 : template <int MT, typename L1, typename L2, typename L3>
641 : void HostDeviceParallelFor (Box const& box1, Box const& box2, Box const& box3,
642 : L1&& f1, L2&& f2, L3&& f3) noexcept
643 : {
644 : amrex::ignore_unused(MT);
645 : ParallelFor(box1,box2,box3,std::forward<L1>(f1),std::forward<L2>(f2),std::forward<L3>(f3));
646 : }
647 :
648 : template <typename T1, typename T2, typename L1, typename L2,
649 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
650 : typename M2=std::enable_if_t<std::is_integral_v<T2>> >
651 : void HostDeviceParallelFor (Box const& box1, T1 ncomp1, L1&& f1,
652 : Box const& box2, T2 ncomp2, L2&& f2) noexcept
653 : {
654 : ParallelFor(box1,ncomp1,std::forward<L1>(f1),box2,ncomp2,std::forward<L2>(f2));
655 : }
656 :
657 : template <int MT, typename T1, typename T2, typename L1, typename L2,
658 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
659 : typename M2=std::enable_if_t<std::is_integral_v<T2>> >
660 : void HostDeviceParallelFor (Box const& box1, T1 ncomp1, L1&& f1,
661 : Box const& box2, T2 ncomp2, L2&& f2) noexcept
662 : {
663 : amrex::ignore_unused(MT);
664 : ParallelFor(box1,ncomp1,std::forward<L1>(f1),box2,ncomp2,std::forward<L2>(f2));
665 : }
666 :
667 : template <typename T1, typename T2, typename T3, typename L1, typename L2, typename L3,
668 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
669 : typename M2=std::enable_if_t<std::is_integral_v<T2>>,
670 : typename M3=std::enable_if_t<std::is_integral_v<T3>> >
671 : void HostDeviceParallelFor (Box const& box1, T1 ncomp1, L1&& f1,
672 : Box const& box2, T2 ncomp2, L2&& f2,
673 : Box const& box3, T3 ncomp3, L3&& f3) noexcept
674 : {
675 : ParallelFor(box1,ncomp1,std::forward<L1>(f1),
676 : box2,ncomp2,std::forward<L2>(f2),
677 : box3,ncomp3,std::forward<L3>(f3));
678 : }
679 :
680 : template <int MT, typename T1, typename T2, typename T3, typename L1, typename L2, typename L3,
681 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
682 : typename M2=std::enable_if_t<std::is_integral_v<T2>>,
683 : typename M3=std::enable_if_t<std::is_integral_v<T3>> >
684 : void HostDeviceParallelFor (Box const& box1, T1 ncomp1, L1&& f1,
685 : Box const& box2, T2 ncomp2, L2&& f2,
686 : Box const& box3, T3 ncomp3, L3&& f3) noexcept
687 : {
688 : amrex::ignore_unused(MT);
689 : ParallelFor(box1,ncomp1,std::forward<L1>(f1),
690 : box2,ncomp2,std::forward<L2>(f2),
691 : box3,ncomp3,std::forward<L3>(f3));
692 : }
693 :
694 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
695 : void HostDeviceFor (T n, L&& f) noexcept
696 : {
697 : For(n,std::forward<L>(f));
698 : }
699 :
700 : template <int MT, typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
701 : void HostDeviceFor (T n, L&& f) noexcept
702 : {
703 : amrex::ignore_unused(MT);
704 : For(n,std::forward<L>(f));
705 : }
706 :
707 : template <typename L>
708 : void HostDeviceFor (Box const& box, L&& f) noexcept
709 : {
710 : For(box,std::forward<L>(f));
711 : }
712 :
713 : template <int MT, typename L>
714 : void HostDeviceFor (Box const& box, L&& f) noexcept
715 : {
716 : amrex::ignore_unused(MT);
717 : For(box,std::forward<L>(f));
718 : }
719 :
720 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
721 : void HostDeviceFor (Box const& box, T ncomp, L&& f) noexcept
722 : {
723 : For(box,ncomp,std::forward<L>(f));
724 : }
725 :
726 : template <int MT, typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
727 : void HostDeviceFor (Box const& box, T ncomp, L&& f) noexcept
728 : {
729 : amrex::ignore_unused(MT);
730 : For(box,ncomp,std::forward<L>(f));
731 : }
732 :
733 : template <typename L1, typename L2>
734 : void HostDeviceFor (Box const& box1, Box const& box2, L1&& f1, L2&& f2) noexcept
735 : {
736 : For(box1,box2,std::forward<L1>(f1),std::forward<L2>(f2));
737 : }
738 :
739 : template <int MT, typename L1, typename L2>
740 : void HostDeviceFor (Box const& box1, Box const& box2, L1&& f1, L2&& f2) noexcept
741 : {
742 : amrex::ignore_unused(MT);
743 : For(box1,box2,std::forward<L1>(f1),std::forward<L2>(f2));
744 : }
745 :
746 : template <typename L1, typename L2, typename L3>
747 : void HostDeviceFor (Box const& box1, Box const& box2, Box const& box3,
748 : L1&& f1, L2&& f2, L3&& f3) noexcept
749 : {
750 : For(box1,box2,box3,std::forward<L1>(f1),std::forward<L2>(f2),std::forward<L3>(f3));
751 : }
752 :
753 : template <int MT, typename L1, typename L2, typename L3>
754 : void HostDeviceFor (Box const& box1, Box const& box2, Box const& box3,
755 : L1&& f1, L2&& f2, L3&& f3) noexcept
756 : {
757 : amrex::ignore_unused(MT);
758 : For(box1,box2,box3,std::forward<L1>(f1),std::forward<L2>(f2),std::forward<L3>(f3));
759 : }
760 :
761 : template <typename T1, typename T2, typename L1, typename L2,
762 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
763 : typename M2=std::enable_if_t<std::is_integral_v<T2>> >
764 : void HostDeviceFor (Box const& box1, T1 ncomp1, L1&& f1,
765 : Box const& box2, T2 ncomp2, L2&& f2) noexcept
766 : {
767 : For(box1,ncomp1,std::forward<L1>(f1),box2,ncomp2,std::forward<L2>(f2));
768 : }
769 :
770 : template <int MT, typename T1, typename T2, typename L1, typename L2,
771 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
772 : typename M2=std::enable_if_t<std::is_integral_v<T2>> >
773 : void HostDeviceFor (Box const& box1, T1 ncomp1, L1&& f1,
774 : Box const& box2, T2 ncomp2, L2&& f2) noexcept
775 : {
776 : amrex::ignore_unused(MT);
777 : For(box1,ncomp1,std::forward<L1>(f1),box2,ncomp2,std::forward<L2>(f2));
778 : }
779 :
780 : template <typename T1, typename T2, typename T3, typename L1, typename L2, typename L3,
781 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
782 : typename M2=std::enable_if_t<std::is_integral_v<T2>>,
783 : typename M3=std::enable_if_t<std::is_integral_v<T3>> >
784 : void HostDeviceFor (Box const& box1, T1 ncomp1, L1&& f1,
785 : Box const& box2, T2 ncomp2, L2&& f2,
786 : Box const& box3, T3 ncomp3, L3&& f3) noexcept
787 : {
788 : For(box1,ncomp1,std::forward<L1>(f1),
789 : box2,ncomp2,std::forward<L2>(f2),
790 : box3,ncomp3,std::forward<L3>(f3));
791 : }
792 :
793 : template <int MT, typename T1, typename T2, typename T3, typename L1, typename L2, typename L3,
794 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
795 : typename M2=std::enable_if_t<std::is_integral_v<T2>>,
796 : typename M3=std::enable_if_t<std::is_integral_v<T3>> >
797 : void HostDeviceFor (Box const& box1, T1 ncomp1, L1&& f1,
798 : Box const& box2, T2 ncomp2, L2&& f2,
799 : Box const& box3, T3 ncomp3, L3&& f3) noexcept
800 : {
801 : amrex::ignore_unused(MT);
802 : For(box1,ncomp1,std::forward<L1>(f1),
803 : box2,ncomp2,std::forward<L2>(f2),
804 : box3,ncomp3,std::forward<L3>(f3));
805 : }
806 :
807 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
808 : void HostDeviceParallelFor (Gpu::KernelInfo const&, T n, L&& f) noexcept
809 : {
810 : ParallelFor(n,std::forward<L>(f));
811 : }
812 :
813 : template <int MT, typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
814 : void HostDeviceParallelFor (Gpu::KernelInfo const&, T n, L&& f) noexcept
815 : {
816 : amrex::ignore_unused(MT);
817 : ParallelFor(n,std::forward<L>(f));
818 : }
819 :
820 : template <typename L>
821 : void HostDeviceParallelFor (Gpu::KernelInfo const&, Box const& box, L&& f) noexcept
822 : {
823 : ParallelFor(box,std::forward<L>(f));
824 : }
825 :
826 : template <int MT, typename L>
827 : void HostDeviceParallelFor (Gpu::KernelInfo const&, Box const& box, L&& f) noexcept
828 : {
829 : amrex::ignore_unused(MT);
830 : ParallelFor(box,std::forward<L>(f));
831 : }
832 :
833 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
834 : void HostDeviceParallelFor (Gpu::KernelInfo const&, Box const& box, T ncomp, L&& f) noexcept
835 : {
836 : ParallelFor(box,ncomp,std::forward<L>(f));
837 : }
838 :
839 : template <int MT, typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
840 : void HostDeviceParallelFor (Gpu::KernelInfo const&, Box const& box, T ncomp, L&& f) noexcept
841 : {
842 : amrex::ignore_unused(MT);
843 : ParallelFor(box,ncomp,std::forward<L>(f));
844 : }
845 :
846 : template <typename L1, typename L2>
847 : void HostDeviceParallelFor (Gpu::KernelInfo const&, Box const& box1, Box const& box2, L1&& f1, L2&& f2) noexcept
848 : {
849 : ParallelFor(box1,box2,std::forward<L1>(f1),std::forward<L2>(f2));
850 : }
851 :
852 : template <int MT, typename L1, typename L2>
853 : void HostDeviceParallelFor (Gpu::KernelInfo const&, Box const& box1, Box const& box2, L1&& f1, L2&& f2) noexcept
854 : {
855 : amrex::ignore_unused(MT);
856 : ParallelFor(box1,box2,std::forward<L1>(f1),std::forward<L2>(f2));
857 : }
858 :
859 : template <typename L1, typename L2, typename L3>
860 : void HostDeviceParallelFor (Gpu::KernelInfo const&,
861 : Box const& box1, Box const& box2, Box const& box3,
862 : L1&& f1, L2&& f2, L3&& f3) noexcept
863 : {
864 : ParallelFor(box1,box2,box3,std::forward<L1>(f1),std::forward<L2>(f2),std::forward<L3>(f3));
865 : }
866 :
867 : template <int MT, typename L1, typename L2, typename L3>
868 : void HostDeviceParallelFor (Gpu::KernelInfo const&,
869 : Box const& box1, Box const& box2, Box const& box3,
870 : L1&& f1, L2&& f2, L3&& f3) noexcept
871 : {
872 : amrex::ignore_unused(MT);
873 : ParallelFor(box1,box2,box3,std::forward<L1>(f1),std::forward<L2>(f2),std::forward<L3>(f3));
874 : }
875 :
876 : template <typename T1, typename T2, typename L1, typename L2,
877 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
878 : typename M2=std::enable_if_t<std::is_integral_v<T2>> >
879 : void HostDeviceParallelFor (Gpu::KernelInfo const&,
880 : Box const& box1, T1 ncomp1, L1&& f1,
881 : Box const& box2, T2 ncomp2, L2&& f2) noexcept
882 : {
883 : ParallelFor(box1,ncomp1,std::forward<L1>(f1),box2,ncomp2,std::forward<L2>(f2));
884 : }
885 :
886 : template <int MT, typename T1, typename T2, typename L1, typename L2,
887 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
888 : typename M2=std::enable_if_t<std::is_integral_v<T2>> >
889 : void HostDeviceParallelFor (Gpu::KernelInfo const&,
890 : Box const& box1, T1 ncomp1, L1&& f1,
891 : Box const& box2, T2 ncomp2, L2&& f2) noexcept
892 : {
893 : amrex::ignore_unused(MT);
894 : ParallelFor(box1,ncomp1,std::forward<L1>(f1),box2,ncomp2,std::forward<L2>(f2));
895 : }
896 :
897 : template <typename T1, typename T2, typename T3, typename L1, typename L2, typename L3,
898 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
899 : typename M2=std::enable_if_t<std::is_integral_v<T2>>,
900 : typename M3=std::enable_if_t<std::is_integral_v<T3>> >
901 : void HostDeviceParallelFor (Gpu::KernelInfo const&,
902 : Box const& box1, T1 ncomp1, L1&& f1,
903 : Box const& box2, T2 ncomp2, L2&& f2,
904 : Box const& box3, T3 ncomp3, L3&& f3) noexcept
905 : {
906 : ParallelFor(box1,ncomp1,std::forward<L1>(f1),
907 : box2,ncomp2,std::forward<L2>(f2),
908 : box3,ncomp3,std::forward<L3>(f3));
909 : }
910 :
911 : template <int MT, typename T1, typename T2, typename T3, typename L1, typename L2, typename L3,
912 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
913 : typename M2=std::enable_if_t<std::is_integral_v<T2>>,
914 : typename M3=std::enable_if_t<std::is_integral_v<T3>> >
915 : void HostDeviceParallelFor (Gpu::KernelInfo const&,
916 : Box const& box1, T1 ncomp1, L1&& f1,
917 : Box const& box2, T2 ncomp2, L2&& f2,
918 : Box const& box3, T3 ncomp3, L3&& f3) noexcept
919 : {
920 : amrex::ignore_unused(MT);
921 : ParallelFor(box1,ncomp1,std::forward<L1>(f1),
922 : box2,ncomp2,std::forward<L2>(f2),
923 : box3,ncomp3,std::forward<L3>(f3));
924 : }
925 :
926 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
927 : void HostDeviceFor (Gpu::KernelInfo const&, T n, L&& f) noexcept
928 : {
929 : For(n,std::forward<L>(f));
930 : }
931 :
932 : template <int MT, typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
933 : void HostDeviceFor (Gpu::KernelInfo const&, T n, L&& f) noexcept
934 : {
935 : amrex::ignore_unused(MT);
936 : For(n,std::forward<L>(f));
937 : }
938 :
939 : template <typename L>
940 : void HostDeviceFor (Gpu::KernelInfo const&, Box const& box, L&& f) noexcept
941 : {
942 : For(box,std::forward<L>(f));
943 : }
944 :
945 : template <int MT, typename L>
946 : void HostDeviceFor (Gpu::KernelInfo const&, Box const& box, L&& f) noexcept
947 : {
948 : amrex::ignore_unused(MT);
949 : For(box,std::forward<L>(f));
950 : }
951 :
952 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
953 : void HostDeviceFor (Gpu::KernelInfo const&, Box const& box, T ncomp, L&& f) noexcept
954 : {
955 : For(box,ncomp,std::forward<L>(f));
956 : }
957 :
958 : template <int MT, typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
959 : void HostDeviceFor (Gpu::KernelInfo const&, Box const& box, T ncomp, L&& f) noexcept
960 : {
961 : amrex::ignore_unused(MT);
962 : For(box,ncomp,std::forward<L>(f));
963 : }
964 :
965 : template <typename L1, typename L2>
966 : void HostDeviceFor (Gpu::KernelInfo const&, Box const& box1, Box const& box2, L1&& f1, L2&& f2) noexcept
967 : {
968 : For(box1,box2,std::forward<L1>(f1),std::forward<L2>(f2));
969 : }
970 :
971 : template <int MT, typename L1, typename L2>
972 : void HostDeviceFor (Gpu::KernelInfo const&, Box const& box1, Box const& box2, L1&& f1, L2&& f2) noexcept
973 : {
974 : amrex::ignore_unused(MT);
975 : For(box1,box2,std::forward<L1>(f1),std::forward<L2>(f2));
976 : }
977 :
978 : template <typename L1, typename L2, typename L3>
979 : void HostDeviceFor (Gpu::KernelInfo const&,
980 : Box const& box1, Box const& box2, Box const& box3,
981 : L1&& f1, L2&& f2, L3&& f3) noexcept
982 : {
983 : For(box1,box2,box3,std::forward<L1>(f1),std::forward<L2>(f2),std::forward<L3>(f3));
984 : }
985 :
986 : template <int MT, typename L1, typename L2, typename L3>
987 : void HostDeviceFor (Gpu::KernelInfo const&,
988 : Box const& box1, Box const& box2, Box const& box3,
989 : L1&& f1, L2&& f2, L3&& f3) noexcept
990 : {
991 : amrex::ignore_unused(MT);
992 : For(box1,box2,box3,std::forward<L1>(f1),std::forward<L2>(f2),std::forward<L3>(f3));
993 : }
994 :
995 : template <typename T1, typename T2, typename L1, typename L2,
996 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
997 : typename M2=std::enable_if_t<std::is_integral_v<T2>> >
998 : void HostDeviceFor (Gpu::KernelInfo const&,
999 : Box const& box1, T1 ncomp1, L1&& f1,
1000 : Box const& box2, T2 ncomp2, L2&& f2) noexcept
1001 : {
1002 : For(box1,ncomp1,std::forward<L1>(f1),box2,ncomp2,std::forward<L2>(f2));
1003 : }
1004 :
1005 : template <int MT, typename T1, typename T2, typename L1, typename L2,
1006 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
1007 : typename M2=std::enable_if_t<std::is_integral_v<T2>> >
1008 : void HostDeviceFor (Gpu::KernelInfo const&,
1009 : Box const& box1, T1 ncomp1, L1&& f1,
1010 : Box const& box2, T2 ncomp2, L2&& f2) noexcept
1011 : {
1012 : amrex::ignore_unused(MT);
1013 : For(box1,ncomp1,std::forward<L1>(f1),box2,ncomp2,std::forward<L2>(f2));
1014 : }
1015 :
1016 : template <typename T1, typename T2, typename T3, typename L1, typename L2, typename L3,
1017 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
1018 : typename M2=std::enable_if_t<std::is_integral_v<T2>>,
1019 : typename M3=std::enable_if_t<std::is_integral_v<T3>> >
1020 : void HostDeviceFor (Gpu::KernelInfo const&,
1021 : Box const& box1, T1 ncomp1, L1&& f1,
1022 : Box const& box2, T2 ncomp2, L2&& f2,
1023 : Box const& box3, T3 ncomp3, L3&& f3) noexcept
1024 : {
1025 : For(box1,ncomp1,std::forward<L1>(f1),
1026 : box2,ncomp2,std::forward<L2>(f2),
1027 : box3,ncomp3,std::forward<L3>(f3));
1028 : }
1029 :
1030 : template <int MT, typename T1, typename T2, typename T3, typename L1, typename L2, typename L3,
1031 : typename M1=std::enable_if_t<std::is_integral_v<T1>>,
1032 : typename M2=std::enable_if_t<std::is_integral_v<T2>>,
1033 : typename M3=std::enable_if_t<std::is_integral_v<T3>> >
1034 : void HostDeviceFor (Gpu::KernelInfo const&,
1035 : Box const& box1, T1 ncomp1, L1&& f1,
1036 : Box const& box2, T2 ncomp2, L2&& f2,
1037 : Box const& box3, T3 ncomp3, L3&& f3) noexcept
1038 : {
1039 : amrex::ignore_unused(MT);
1040 : For(box1,ncomp1,std::forward<L1>(f1),
1041 : box2,ncomp2,std::forward<L2>(f2),
1042 : box3,ncomp3,std::forward<L3>(f3));
1043 : }
1044 :
1045 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
1046 : AMREX_ATTRIBUTE_FLATTEN_FOR
1047 : void ParallelForRNG (T n, L const& f) noexcept
1048 : {
1049 : for (T i = 0; i < n; ++i) {
1050 : f(i,RandomEngine{});
1051 : }
1052 : }
1053 :
1054 : template <typename L>
1055 : AMREX_ATTRIBUTE_FLATTEN_FOR
1056 : void ParallelForRNG (Box const& box, L const& f) noexcept
1057 : {
1058 : const auto lo = amrex::lbound(box);
1059 : const auto hi = amrex::ubound(box);
1060 : for (int k = lo.z; k <= hi.z; ++k) {
1061 : for (int j = lo.y; j <= hi.y; ++j) {
1062 : for (int i = lo.x; i <= hi.x; ++i) {
1063 : f(i,j,k,RandomEngine{});
1064 : }}}
1065 : }
1066 :
1067 : template <typename T, typename L, typename M=std::enable_if_t<std::is_integral_v<T>> >
1068 : AMREX_ATTRIBUTE_FLATTEN_FOR
1069 : void ParallelForRNG (Box const& box, T ncomp, L const& f) noexcept
1070 : {
1071 : const auto lo = amrex::lbound(box);
1072 : const auto hi = amrex::ubound(box);
1073 : for (T n = 0; n < ncomp; ++n) {
1074 : for (int k = lo.z; k <= hi.z; ++k) {
1075 : for (int j = lo.y; j <= hi.y; ++j) {
1076 : for (int i = lo.x; i <= hi.x; ++i) {
1077 : f(i,j,k,n,RandomEngine{});
1078 : }}}
1079 : }
1080 : }
1081 :
1082 : template <typename L>
1083 : void single_task (L&& f) noexcept
1084 : {
1085 : std::forward<L>(f)();
1086 : }
1087 :
1088 : }
1089 :
1090 : #endif
|
