• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1/*===---- complex - CUDA wrapper for <complex> ------------------------------===
2 *
3 * Permission is hereby granted, free of charge, to any person obtaining a copy
4 * of this software and associated documentation files (the "Software"), to deal
5 * in the Software without restriction, including without limitation the rights
6 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
7 * copies of the Software, and to permit persons to whom the Software is
8 * furnished to do so, subject to the following conditions:
9 *
10 * The above copyright notice and this permission notice shall be included in
11 * all copies or substantial portions of the Software.
12 *
13 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
14 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
15 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
16 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
17 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
18 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
19 * THE SOFTWARE.
20 *
21 *===-----------------------------------------------------------------------===
22 */
23
24#ifndef __CLANG_CUDA_WRAPPERS_COMPLEX
25#define __CLANG_CUDA_WRAPPERS_COMPLEX
26
27// Wrapper around <complex> that forces its functions to be __host__
28// __device__.
29
30// First, include host-only headers we think are likely to be included by
31// <complex>, so that the pragma below only applies to <complex> itself.
32#if __cplusplus >= 201103L
33#include <type_traits>
34#endif
35#include <stdexcept>
36#include <cmath>
37#include <sstream>
38
39// Next, include our <algorithm> wrapper, to ensure that device overloads of
40// std::min/max are available.
41#include <algorithm>
42
43#pragma clang force_cuda_host_device begin
44
45// When compiling for device, ask libstdc++ to use its own implements of
46// complex functions, rather than calling builtins (which resolve to library
47// functions that don't exist when compiling CUDA device code).
48//
49// This is a little dicey, because it causes libstdc++ to define a different
50// set of overloads on host and device.
51//
52//   // Present only when compiling for host.
53//   __host__ __device__ void complex<float> sin(const complex<float>& x) {
54//     return __builtin_csinf(x);
55//   }
56//
57//   // Present when compiling for host and for device.
58//   template <typename T>
59//   void __host__ __device__ complex<T> sin(const complex<T>& x) {
60//     return complex<T>(sin(x.real()) * cosh(x.imag()),
61//                       cos(x.real()), sinh(x.imag()));
62//   }
63//
64// This is safe because when compiling for device, all function calls in
65// __host__ code to sin() will still resolve to *something*, even if they don't
66// resolve to the same function as they resolve to when compiling for host.  We
67// don't care that they don't resolve to the right function because we won't
68// codegen this host code when compiling for device.
69
70#pragma push_macro("_GLIBCXX_USE_C99_COMPLEX")
71#pragma push_macro("_GLIBCXX_USE_C99_COMPLEX_TR1")
72#define _GLIBCXX_USE_C99_COMPLEX 0
73#define _GLIBCXX_USE_C99_COMPLEX_TR1 0
74
75#include_next <complex>
76
77#pragma pop_macro("_GLIBCXX_USE_C99_COMPLEX_TR1")
78#pragma pop_macro("_GLIBCXX_USE_C99_COMPLEX")
79
80#pragma clang force_cuda_host_device end
81
82#endif // include guard
83