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
|
// -*-C++-*-
#ifndef MATHFUNCS_EXP_H
#define MATHFUNCS_EXP_H
#include "mathfuncs_base.h"
#include <cmath>
namespace vecmathlib {
template<typename realvec_t>
realvec_t mathfuncs<realvec_t>::vml_exp2(realvec_t x)
{
// Rescale
realvec_t x0 = x;
// realvec_t round_x = rint(x);
// intvec_t iround_x = convert_int(round_x);
// r = ldexp(r, iround_x);
// Round by adding, then subtracting again a large number
// Add a large number to move the mantissa bits to the right
int_t large = (U(1) << FP::mantissa_bits) + FP::exponent_offset;
realvec_t tmp = x + RV(R(large));
tmp.barrier();
realvec_t round_x = tmp - RV(R(large));
x -= round_x;
VML_ASSERT(all(x >= RV(-0.5) && x <= RV(0.5)));
// Polynomial expansion
realvec_t r;
switch (sizeof(real_t)) {
case 4:
#ifdef VML_HAVE_FP_CONTRACT
// float, error=4.55549108005200277750378992345e-9
r = RV(0.000154653240842602623787395880898);
r = fma(r, x, RV(0.00133952915439234389712105060319));
r = fma(r, x, RV(0.0096180399118156827664944870552));
r = fma(r, x, RV(0.055503406540531310853149866446));
r = fma(r, x, RV(0.240226511015459465468737123346));
r = fma(r, x, RV(0.69314720007380208630542805293));
r = fma(r, x, RV(0.99999999997182023878745628977));
#else
// float, error=1.62772721960621336664735896836e-7
r = RV(0.00133952915439234389712105060319);
r = fma(r, x, RV(0.009670773148229417605024318985));
r = fma(r, x, RV(0.055503406540531310853149866446));
r = fma(r, x, RV(0.240222115700585316818177639177));
r = fma(r, x, RV(0.69314720007380208630542805293));
r = fma(r, x, RV(1.00000005230745711373079206024));
#endif
break;
case 8:
#ifdef VML_HAVE_FP_CONTRACT
// double, error=9.32016781355638010975628074746e-18
r = RV(4.45623165388261696886670014471e-10);
r = fma(r, x, RV(7.0733589360775271430968224806e-9));
r = fma(r, x, RV(1.01780540270960163558119510246e-7));
r = fma(r, x, RV(1.3215437348041505269462510712e-6));
r = fma(r, x, RV(0.000015252733849766201174247690629));
r = fma(r, x, RV(0.000154035304541242555115696403795));
r = fma(r, x, RV(0.00133335581463968601407096905671));
r = fma(r, x, RV(0.0096181291075949686712855561931));
r = fma(r, x, RV(0.055504108664821672870565883052));
r = fma(r, x, RV(0.240226506959101382690753994082));
r = fma(r, x, RV(0.69314718055994530864272481773));
r = fma(r, x, RV(0.9999999999999999978508676375));
#else
// double, error=3.74939899823302048807873981077e-14
r = RV(1.02072375599725694063203809188e-7);
r = fma(r, x, RV(1.32573274434801314145133004073e-6));
r = fma(r, x, RV(0.0000152526647170731944840736190013));
r = fma(r, x, RV(0.000154034441925859828261898614555));
r = fma(r, x, RV(0.00133335582175770747495287552557));
r = fma(r, x, RV(0.0096181291794939392517233403183));
r = fma(r, x, RV(0.055504108664525029438908798685));
r = fma(r, x, RV(0.240226506957026959772247598695));
r = fma(r, x, RV(0.6931471805599487321347668143));
r = fma(r, x, RV(1.00000000000000942892870993489));
#endif
break;
default:
__builtin_unreachable();
}
// Undo rescaling
// Extract integer as lowest mantissa bits (highest bits still
// contain offset, exponent, and sign)
intvec_t itmp = as_int(tmp);
// Construct scale factor by setting exponent (this shifts out the
// highest bits)
realvec_t scale = as_float(itmp << I(FP::mantissa_bits));
scale = ifthen(x0 < RV(R(FP::min_exponent)), RV(0.0), scale);
r *= scale;
return r;
}
template<typename realvec_t>
inline
realvec_t mathfuncs<realvec_t>::vml_exp(realvec_t x)
{
return exp2(RV(M_LOG2E) * x);
}
template<typename realvec_t>
inline
realvec_t mathfuncs<realvec_t>::vml_exp10(realvec_t x)
{
return exp2(RV(M_LOG2E * M_LN10) * x);
}
template<typename realvec_t>
inline
realvec_t mathfuncs<realvec_t>::vml_expm1(realvec_t x)
{
return exp(x) - RV(1.0);
#if 0
r = exp(x) - RV(1.0);
return ifthen(r == RV(0.0), x, r);
#endif
}
}; // namespace vecmathlib
#endif // #ifndef MATHFUNCS_EXP_H
|
