NEON/Complex.h
Go to the documentation of this file.
1 // This file is part of Eigen, a lightweight C++ template library
2 // for linear algebra.
3 //
4 // Copyright (C) 2010 Gael Guennebaud <gael.guennebaud@inria.fr>
5 //
6 // This Source Code Form is subject to the terms of the Mozilla
7 // Public License v. 2.0. If a copy of the MPL was not distributed
8 // with this file, You can obtain one at http://mozilla.org/MPL/2.0/.
9 
10 #ifndef EIGEN_COMPLEX_NEON_H
11 #define EIGEN_COMPLEX_NEON_H
12 
13 namespace Eigen {
14 
15 namespace internal {
16 
17 static uint32x4_t p4ui_CONJ_XOR = EIGEN_INIT_NEON_PACKET4(0x00000000, 0x80000000, 0x00000000, 0x80000000);
18 static uint32x2_t p2ui_CONJ_XOR = EIGEN_INIT_NEON_PACKET2(0x00000000, 0x80000000);
19 
20 //---------- float ----------
21 struct Packet2cf
22 {
24  EIGEN_STRONG_INLINE explicit Packet2cf(const Packet4f& a) : v(a) {}
25  Packet4f v;
26 };
27 
28 template<> struct packet_traits<std::complex<float> > : default_packet_traits
29 {
30  typedef Packet2cf type;
31  enum {
32  Vectorizable = 1,
33  AlignedOnScalar = 1,
34  size = 2,
35 
36  HasAdd = 1,
37  HasSub = 1,
38  HasMul = 1,
39  HasDiv = 1,
40  HasNegate = 1,
41  HasAbs = 0,
42  HasAbs2 = 0,
43  HasMin = 0,
44  HasMax = 0,
45  HasSetLinear = 0
46  };
47 };
48 
49 template<> struct unpacket_traits<Packet2cf> { typedef std::complex<float> type; enum {size=2}; };
50 
51 template<> EIGEN_STRONG_INLINE Packet2cf pset1<Packet2cf>(const std::complex<float>& from)
52 {
53  float32x2_t r64;
54  r64 = vld1_f32((float *)&from);
55 
56  return Packet2cf(vcombine_f32(r64, r64));
57 }
58 
59 template<> EIGEN_STRONG_INLINE Packet2cf padd<Packet2cf>(const Packet2cf& a, const Packet2cf& b) { return Packet2cf(padd<Packet4f>(a.v,b.v)); }
60 template<> EIGEN_STRONG_INLINE Packet2cf psub<Packet2cf>(const Packet2cf& a, const Packet2cf& b) { return Packet2cf(psub<Packet4f>(a.v,b.v)); }
61 template<> EIGEN_STRONG_INLINE Packet2cf pnegate(const Packet2cf& a) { return Packet2cf(pnegate<Packet4f>(a.v)); }
62 template<> EIGEN_STRONG_INLINE Packet2cf pconj(const Packet2cf& a)
63 {
64  Packet4ui b = vreinterpretq_u32_f32(a.v);
65  return Packet2cf(vreinterpretq_f32_u32(veorq_u32(b, p4ui_CONJ_XOR)));
66 }
67 
69 {
70  Packet4f v1, v2;
71 
72  // Get the real values of a | a1_re | a1_re | a2_re | a2_re |
73  v1 = vcombine_f32(vdup_lane_f32(vget_low_f32(a.v), 0), vdup_lane_f32(vget_high_f32(a.v), 0));
74  // Get the real values of a | a1_im | a1_im | a2_im | a2_im |
75  v2 = vcombine_f32(vdup_lane_f32(vget_low_f32(a.v), 1), vdup_lane_f32(vget_high_f32(a.v), 1));
76  // Multiply the real a with b
77  v1 = vmulq_f32(v1, b.v);
78  // Multiply the imag a with b
79  v2 = vmulq_f32(v2, b.v);
80  // Conjugate v2
81  v2 = vreinterpretq_f32_u32(veorq_u32(vreinterpretq_u32_f32(v2), p4ui_CONJ_XOR));
82  // Swap real/imag elements in v2.
83  v2 = vrev64q_f32(v2);
84  // Add and return the result
85  return Packet2cf(vaddq_f32(v1, v2));
86 }
87 
89 {
90  return Packet2cf(vreinterpretq_f32_u32(vorrq_u32(vreinterpretq_u32_f32(a.v),vreinterpretq_u32_f32(b.v))));
91 }
93 {
94  return Packet2cf(vreinterpretq_f32_u32(vorrq_u32(vreinterpretq_u32_f32(a.v),vreinterpretq_u32_f32(b.v))));
95 }
97 {
98  return Packet2cf(vreinterpretq_f32_u32(veorq_u32(vreinterpretq_u32_f32(a.v),vreinterpretq_u32_f32(b.v))));
99 }
101 {
102  return Packet2cf(vreinterpretq_f32_u32(vbicq_u32(vreinterpretq_u32_f32(a.v),vreinterpretq_u32_f32(b.v))));
103 }
104 
105 template<> EIGEN_STRONG_INLINE Packet2cf pload<Packet2cf>(const std::complex<float>* from) { EIGEN_DEBUG_ALIGNED_LOAD return Packet2cf(pload<Packet4f>((const float*)from)); }
106 template<> EIGEN_STRONG_INLINE Packet2cf ploadu<Packet2cf>(const std::complex<float>* from) { EIGEN_DEBUG_UNALIGNED_LOAD return Packet2cf(ploadu<Packet4f>((const float*)from)); }
107 
108 template<> EIGEN_STRONG_INLINE Packet2cf ploaddup<Packet2cf>(const std::complex<float>* from) { return pset1<Packet2cf>(*from); }
109 
110 template<> EIGEN_STRONG_INLINE void pstore <std::complex<float> >(std::complex<float> * to, const Packet2cf& from) { EIGEN_DEBUG_ALIGNED_STORE pstore((float*)to, from.v); }
111 template<> EIGEN_STRONG_INLINE void pstoreu<std::complex<float> >(std::complex<float> * to, const Packet2cf& from) { EIGEN_DEBUG_UNALIGNED_STORE pstoreu((float*)to, from.v); }
112 
113 template<> EIGEN_STRONG_INLINE void prefetch<std::complex<float> >(const std::complex<float> * addr) { __pld((float *)addr); }
114 
115 template<> EIGEN_STRONG_INLINE std::complex<float> pfirst<Packet2cf>(const Packet2cf& a)
116 {
117  std::complex<float> EIGEN_ALIGN16 x[2];
118  vst1q_f32((float *)x, a.v);
119  return x[0];
120 }
121 
122 template<> EIGEN_STRONG_INLINE Packet2cf preverse(const Packet2cf& a)
123 {
124  float32x2_t a_lo, a_hi;
125  Packet4f a_r128;
126 
127  a_lo = vget_low_f32(a.v);
128  a_hi = vget_high_f32(a.v);
129  a_r128 = vcombine_f32(a_hi, a_lo);
130 
131  return Packet2cf(a_r128);
132 }
133 
135 {
136  return Packet2cf(vrev64q_f32(a.v));
137 }
138 
139 template<> EIGEN_STRONG_INLINE std::complex<float> predux<Packet2cf>(const Packet2cf& a)
140 {
141  float32x2_t a1, a2;
142  std::complex<float> s;
143 
144  a1 = vget_low_f32(a.v);
145  a2 = vget_high_f32(a.v);
146  a2 = vadd_f32(a1, a2);
147  vst1_f32((float *)&s, a2);
148 
149  return s;
150 }
151 
153 {
154  Packet4f sum1, sum2, sum;
155 
156  // Add the first two 64-bit float32x2_t of vecs[0]
157  sum1 = vcombine_f32(vget_low_f32(vecs[0].v), vget_low_f32(vecs[1].v));
158  sum2 = vcombine_f32(vget_high_f32(vecs[0].v), vget_high_f32(vecs[1].v));
159  sum = vaddq_f32(sum1, sum2);
160 
161  return Packet2cf(sum);
162 }
163 
164 template<> EIGEN_STRONG_INLINE std::complex<float> predux_mul<Packet2cf>(const Packet2cf& a)
165 {
166  float32x2_t a1, a2, v1, v2, prod;
167  std::complex<float> s;
168 
169  a1 = vget_low_f32(a.v);
170  a2 = vget_high_f32(a.v);
171  // Get the real values of a | a1_re | a1_re | a2_re | a2_re |
172  v1 = vdup_lane_f32(a1, 0);
173  // Get the real values of a | a1_im | a1_im | a2_im | a2_im |
174  v2 = vdup_lane_f32(a1, 1);
175  // Multiply the real a with b
176  v1 = vmul_f32(v1, a2);
177  // Multiply the imag a with b
178  v2 = vmul_f32(v2, a2);
179  // Conjugate v2
180  v2 = vreinterpret_f32_u32(veor_u32(vreinterpret_u32_f32(v2), p2ui_CONJ_XOR));
181  // Swap real/imag elements in v2.
182  v2 = vrev64_f32(v2);
183  // Add v1, v2
184  prod = vadd_f32(v1, v2);
185 
186  vst1_f32((float *)&s, prod);
187 
188  return s;
189 }
190 
191 template<int Offset>
192 struct palign_impl<Offset,Packet2cf>
193 {
194  EIGEN_STRONG_INLINE static void run(Packet2cf& first, const Packet2cf& second)
195  {
196  if (Offset==1)
197  {
198  first.v = vextq_f32(first.v, second.v, 2);
199  }
200  }
201 };
202 
203 template<> struct conj_helper<Packet2cf, Packet2cf, false,true>
204 {
205  EIGEN_STRONG_INLINE Packet2cf pmadd(const Packet2cf& x, const Packet2cf& y, const Packet2cf& c) const
206  { return padd(pmul(x,y),c); }
207 
208  EIGEN_STRONG_INLINE Packet2cf pmul(const Packet2cf& a, const Packet2cf& b) const
209  {
210  return internal::pmul(a, pconj(b));
211  }
212 };
213 
214 template<> struct conj_helper<Packet2cf, Packet2cf, true,false>
215 {
216  EIGEN_STRONG_INLINE Packet2cf pmadd(const Packet2cf& x, const Packet2cf& y, const Packet2cf& c) const
217  { return padd(pmul(x,y),c); }
218 
219  EIGEN_STRONG_INLINE Packet2cf pmul(const Packet2cf& a, const Packet2cf& b) const
220  {
221  return internal::pmul(pconj(a), b);
222  }
223 };
224 
225 template<> struct conj_helper<Packet2cf, Packet2cf, true,true>
226 {
227  EIGEN_STRONG_INLINE Packet2cf pmadd(const Packet2cf& x, const Packet2cf& y, const Packet2cf& c) const
228  { return padd(pmul(x,y),c); }
229 
230  EIGEN_STRONG_INLINE Packet2cf pmul(const Packet2cf& a, const Packet2cf& b) const
231  {
232  return pconj(internal::pmul(a, b));
233  }
234 };
235 
236 template<> EIGEN_STRONG_INLINE Packet2cf pdiv<Packet2cf>(const Packet2cf& a, const Packet2cf& b)
237 {
238  // TODO optimize it for AltiVec
239  Packet2cf res = conj_helper<Packet2cf,Packet2cf,false,true>().pmul(a,b);
240  Packet4f s, rev_s;
241 
242  // this computes the norm
243  s = vmulq_f32(b.v, b.v);
244  rev_s = vrev64q_f32(s);
245 
246  return Packet2cf(pdiv(res.v, vaddq_f32(s,rev_s)));
247 }
248 
249 } // end namespace internal
250 
251 } // end namespace Eigen
252 
253 #endif // EIGEN_COMPLEX_NEON_H
EIGEN_STRONG_INLINE Packet2cf pmul(const Packet2cf &a, const Packet2cf &b) const
Definition: NEON/Complex.h:230
#define EIGEN_ALIGN16
__vector float Packet4f
#define EIGEN_STRONG_INLINE
EIGEN_STRONG_INLINE Packet2cf por< Packet2cf >(const Packet2cf &a, const Packet2cf &b)
EIGEN_STRONG_INLINE Packet2cf preduxp< Packet2cf >(const Packet2cf *vecs)
#define EIGEN_DEBUG_UNALIGNED_LOAD
EIGEN_STRONG_INLINE Packet2cf pmul< Packet2cf >(const Packet2cf &a, const Packet2cf &b)
static uint32x2_t p2ui_CONJ_XOR
Definition: NEON/Complex.h:18
EIGEN_STRONG_INLINE Packet2cf ploadu< Packet2cf >(const std::complex< float > *from)
#define EIGEN_DEBUG_ALIGNED_STORE
iterative scaling algorithm to equilibrate rows and column norms in matrices
Definition: matrix.hpp:471
EIGEN_STRONG_INLINE Packet2cf pmadd(const Packet2cf &x, const Packet2cf &y, const Packet2cf &c) const
Definition: NEON/Complex.h:205
EIGEN_STRONG_INLINE Packet2cf pmadd(const Packet2cf &x, const Packet2cf &y, const Packet2cf &c) const
Definition: NEON/Complex.h:216
EIGEN_STRONG_INLINE Packet2cf pload< Packet2cf >(const std::complex< float > *from)
EIGEN_STRONG_INLINE Packet2cf pset1< Packet2cf >(const std::complex< float > &from)
EIGEN_STRONG_INLINE std::complex< float > pfirst< Packet2cf >(const Packet2cf &a)
#define EIGEN_DEBUG_UNALIGNED_STORE
#define EIGEN_DEBUG_ALIGNED_LOAD
EIGEN_STRONG_INLINE Packet2cf psub< Packet2cf >(const Packet2cf &a, const Packet2cf &b)
void pstore(Scalar *to, const Packet &from)
EIGEN_STRONG_INLINE Packet2cf pmul(const Packet2cf &a, const Packet2cf &b) const
Definition: NEON/Complex.h:219
EIGEN_STRONG_INLINE Packet2cf(const Packet4f &a)
Definition: NEON/Complex.h:24
EIGEN_STRONG_INLINE Packet2cf pmadd(const Packet2cf &x, const Packet2cf &y, const Packet2cf &c) const
Definition: NEON/Complex.h:227
void pstoreu(Scalar *to, const Packet &from)
#define EIGEN_INIT_NEON_PACKET2(X, Y)
EIGEN_STRONG_INLINE Packet4f ploadu< Packet4f >(const float *from)
EIGEN_STRONG_INLINE Packet2cf pand< Packet2cf >(const Packet2cf &a, const Packet2cf &b)
EIGEN_STRONG_INLINE Packet2cf()
Definition: NEON/Complex.h:23
#define __pld(x)
EIGEN_STRONG_INLINE std::complex< float > predux_mul< Packet2cf >(const Packet2cf &a)
EIGEN_STRONG_INLINE Packet2cf pmul(const Packet2cf &a, const Packet2cf &b) const
Definition: NEON/Complex.h:208
EIGEN_STRONG_INLINE Packet2cf pconj(const Packet2cf &a)
#define EIGEN_INIT_NEON_PACKET4(X, Y, Z, W)
EIGEN_STRONG_INLINE std::complex< float > predux< Packet2cf >(const Packet2cf &a)
EIGEN_STRONG_INLINE Packet2cf ploaddup< Packet2cf >(const std::complex< float > *from)
EIGEN_STRONG_INLINE Packet4f pload< Packet4f >(const float *from)
Packet pdiv(const Packet &a, const Packet &b)
EIGEN_STRONG_INLINE Packet2cf pnegate(const Packet2cf &a)
static EIGEN_STRONG_INLINE void run(Packet2cf &first, const Packet2cf &second)
Definition: NEON/Complex.h:194
EIGEN_STRONG_INLINE Packet4f padd< Packet4f >(const Packet4f &a, const Packet4f &b)
__vector unsigned int Packet4ui
static Packet4ui p4ui_CONJ_XOR
Packet pmul(const Packet &a, const Packet &b)
EIGEN_STRONG_INLINE Packet2cf pxor< Packet2cf >(const Packet2cf &a, const Packet2cf &b)
EIGEN_STRONG_INLINE Packet2cf padd< Packet2cf >(const Packet2cf &a, const Packet2cf &b)
EIGEN_STRONG_INLINE Packet4f psub< Packet4f >(const Packet4f &a, const Packet4f &b)
EIGEN_STRONG_INLINE Packet2cf pandnot< Packet2cf >(const Packet2cf &a, const Packet2cf &b)
Packet padd(const Packet &a, const Packet &b)
EIGEN_STRONG_INLINE Packet2cf pdiv< Packet2cf >(const Packet2cf &a, const Packet2cf &b)
EIGEN_STRONG_INLINE Packet2cf pcplxflip< Packet2cf >(const Packet2cf &x)
EIGEN_STRONG_INLINE Packet2cf preverse(const Packet2cf &a)


acado
Author(s): Milan Vukov, Rien Quirynen
autogenerated on Mon Jun 10 2019 12:34:30