-
Notifications
You must be signed in to change notification settings - Fork 0
/
fixed_point_op.cpp
219 lines (179 loc) · 6.2 KB
/
fixed_point_op.cpp
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
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
#include <stdio.h>
#include <math.h>
#include "fixed_point_op.h"
// check fraction bits is enough
fixed_t float_to_fixed(float input, uint8_t fixed_bits, uint8_t fraction_bits)
{
uint32_t max_integer;
fixed_t output;
uint32_t value;
printf("\n-------Convert float to fixed-------------\n");
printf("Set float %.2f to %d bits fixed point with fraction bits %d\n", input, fixed_bits, fraction_bits);
max_integer = (1 << (fixed_bits - fraction_bits - 1)) - 1;
if (fabs(input) > max_integer)
{
printf("\nInteger bits of %.2f is not enough!\nModify fraction bits %d to ", input, fraction_bits);
fraction_bits = fixed_bits - int(log2((1 << fixed_bits) - fabs(int(input)))) + 1;
printf("%d\n\n", fraction_bits);
printf("Reset float %.2f to %d bits fixed point with fraction bits %d\n", input, fixed_bits, fraction_bits);
}
value = input * (1 << fraction_bits);
output.fixed_bits = fixed_bits;
output.fraction_bits = fraction_bits;
output.value = MASK_BIT(value, fixed_bits);
return output;
}
float fixed_to_float(fixed_t input)
{
float output;
if (MSB_BIT(input.value, input.fixed_bits))
{
uint32_t tmp;
tmp = TWOSCOMP(input.value, input.fixed_bits);
output = -(float)tmp/(1<<input.fraction_bits);
}
else
{
output = (float)input.value/(1<<input.fraction_bits);
}
return output;
}
// fixed point operation
fixed_t fixed_add(fixed_t fixed_a, fixed_t fixed_b)
{
fixed_t output;
uint8_t fraction_bits, fixed_bits;
uint32_t tmp_fixed_a_bits, tmp_fixed_b_bits;
uint32_t tmp_value_a, tmp_value_b, tmp_value;
fraction_bits = (fixed_a.fraction_bits > fixed_b.fraction_bits) ? fixed_a.fraction_bits : fixed_b.fraction_bits;
output.fraction_bits = fraction_bits;
tmp_value_a = fixed_a.value << (fraction_bits - fixed_a.fraction_bits);
tmp_value_b = fixed_b.value << (fraction_bits - fixed_b.fraction_bits);
tmp_fixed_a_bits = fixed_a.fixed_bits + (fraction_bits - fixed_a.fraction_bits);
tmp_fixed_b_bits = fixed_b.fixed_bits + (fraction_bits - fixed_b.fraction_bits);
fixed_bits = tmp_fixed_a_bits > tmp_fixed_b_bits ? tmp_fixed_a_bits :tmp_fixed_b_bits;
// add 1 bit to avoid overflow
fixed_bits += 1;
output.fixed_bits = fixed_bits;
// if fixed point is positive mask with MAX_FIXED_BITS
// if fixed point is neagtive fill 1 to MAX_FIXED_BITS
if (MSB_BIT(tmp_value_a, tmp_fixed_a_bits))
{
tmp_value_a = FILL_ONE(tmp_value_a, fixed_bits, tmp_fixed_a_bits);
}
else
{
tmp_value_a = MASK_BIT(tmp_value_a, fixed_bits);
}
if (MSB_BIT(tmp_value_b, tmp_fixed_b_bits))
{
tmp_value_b = FILL_ONE(tmp_value_b, fixed_bits, tmp_fixed_b_bits);
}
else
{
tmp_value_b = MASK_BIT(tmp_value_b, fixed_bits);
}
output.value = MASK_BIT((tmp_value_a + tmp_value_b), output.fixed_bits);
return output;
}
fixed_t fixed_mult(fixed_t fixed_a, fixed_t fixed_b)
{
fixed_t output;
uint32_t tmp_value_output, tmp_value_a, tmp_value_b;
int8_t signed_a, signed_b;
uint8_t fixed_bits;
fixed_bits = fixed_a.fixed_bits + fixed_b.fixed_bits;
if (MSB_BIT(fixed_a.value, fixed_a.fixed_bits))
{
tmp_value_a = TWOSCOMP(fixed_a.value, fixed_a.fixed_bits);
signed_a = -1;
}
else
{
tmp_value_a = fixed_a.value;
signed_a = 1;
}
if (MSB_BIT(fixed_b.value, fixed_b.fixed_bits))
{
tmp_value_b = TWOSCOMP(fixed_b.value, fixed_b.fixed_bits);
signed_b = -1;
}
else
{
tmp_value_b = fixed_b.value;
signed_b = 1;
}
tmp_value_output = MASK_BIT((tmp_value_a * tmp_value_b), fixed_bits);
if ((signed_a * signed_b) == -1)
{
output.value = TWOSCOMP(tmp_value_output, fixed_bits);
}
else
{
output.value = tmp_value_output;
}
output.fixed_bits = fixed_bits;
output.fraction_bits = fixed_a.fraction_bits + fixed_b.fraction_bits;
return output;
}
fixed_t fixed_div(fixed_t fixed_a, fixed_t fixed_b)
{
fixed_t output;
uint32_t tmp_value_a, tmp_value_b, tmp_value;
int8_t signed_a, signed_b;
uint8_t diff_fraction_bits,fixed_bits, fraction_bits;
if (MSB_BIT(fixed_a.value, fixed_a.fixed_bits))
{
tmp_value_a = TWOSCOMP(fixed_a.value, fixed_a.fixed_bits);
signed_a = -1;
}
else
{
tmp_value_a = fixed_a.value;
signed_a = 1;
}
if (MSB_BIT(fixed_b.value, fixed_b.fixed_bits))
{
tmp_value_b = TWOSCOMP(fixed_b.value, fixed_b.fixed_bits);
signed_b = -1;
}
else
{
tmp_value_b = fixed_b.value;
signed_b = 1;
}
output.fraction_bits = 0;
if (fixed_a.fraction_bits == fixed_b.fraction_bits)
{
fixed_bits = fixed_a.fixed_bits > fixed_b.fixed_bits ? fixed_a.fixed_bits: fixed_b.fixed_bits;
tmp_value_a = MASK_BIT(tmp_value_a,fixed_bits);
tmp_value_b = MASK_BIT(tmp_value_b,fixed_bits);
output.fixed_bits = fixed_bits;
}
else if (fixed_a.fraction_bits > fixed_b.fraction_bits)
{
diff_fraction_bits = fixed_a.fraction_bits - fixed_b.fraction_bits;
tmp_value_b = tmp_value_b << diff_fraction_bits;
fixed_bits = fixed_a.fixed_bits > (fixed_b.fixed_bits+diff_fraction_bits) ? fixed_a.fixed_bits: (fixed_b.fixed_bits+diff_fraction_bits);
tmp_value_a = MASK_BIT(tmp_value_a,fixed_bits);
tmp_value_b = MASK_BIT(tmp_value_b,fixed_bits);
output.fixed_bits = fixed_bits;
}else{
diff_fraction_bits = fixed_b.fraction_bits - fixed_a.fraction_bits;
tmp_value_a = tmp_value_a << diff_fraction_bits;
fixed_bits = fixed_b.fixed_bits > (fixed_a.fixed_bits+diff_fraction_bits) ? fixed_b.fixed_bits: (fixed_a.fixed_bits+diff_fraction_bits);
tmp_value_a = MASK_BIT(tmp_value_a,fixed_bits);
tmp_value_b = MASK_BIT(tmp_value_b,fixed_bits);
output.fixed_bits = fixed_bits;
}
tmp_value = tmp_value_a / tmp_value_b;
if ((signed_a * signed_b) == -1)
{
output.value = TWOSCOMP(tmp_value, output.fixed_bits);
}
else
{
output.value = tmp_value;
}
return output;
}