forked from generalelectrix/FastLED
-
Notifications
You must be signed in to change notification settings - Fork 0
/
lib8tion.cpp
242 lines (210 loc) · 6.32 KB
/
lib8tion.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
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
#include <stdint.h>
#define RAND16_SEED 1337
uint16_t rand16seed = RAND16_SEED;
// memset8, memcpy8, memmove8:
// optimized avr replacements for the standard "C" library
// routines memset, memcpy, and memmove.
//
// There are two techniques that make these routines
// faster than the standard avr-libc routines.
// First, the loops are unrolled 2X, meaning that
// the average loop overhead is cut in half.
// And second, the compare-and-branch at the bottom
// of each loop decrements the low byte of the
// counter, and if the carry is clear, it branches
// back up immediately. Only if the low byte math
// causes carry do we bother to decrement the high
// byte and check that result for carry as well.
// Results for a 100-byte buffer are 20-40% faster
// than standard avr-libc, at a cost of a few extra
// bytes of code.
#if defined(__AVR__)
extern "C" {
//__attribute__ ((noinline))
void * memset8 ( void * ptr, uint8_t val, uint16_t num )
{
asm volatile(
" movw r26, %[ptr] \n\t"
" sbrs %A[num], 0 \n\t"
" rjmp Lseteven_%= \n\t"
" rjmp Lsetodd_%= \n\t"
"Lsetloop_%=: \n\t"
" st X+, %[val] \n\t"
"Lsetodd_%=: \n\t"
" st X+, %[val] \n\t"
"Lseteven_%=: \n\t"
" subi %A[num], 2 \n\t"
" brcc Lsetloop_%= \n\t"
" sbci %B[num], 0 \n\t"
" brcc Lsetloop_%= \n\t"
: [num] "+r" (num)
: [ptr] "r" (ptr),
[val] "r" (val)
: "memory"
);
return ptr;
}
//__attribute__ ((noinline))
void * memcpy8 ( void * dst, void* src, uint16_t num )
{
asm volatile(
" movw r30, %[src] \n\t"
" movw r26, %[dst] \n\t"
" sbrs %A[num], 0 \n\t"
" rjmp Lcpyeven_%= \n\t"
" rjmp Lcpyodd_%= \n\t"
"Lcpyloop_%=: \n\t"
" ld __tmp_reg__, Z+ \n\t"
" st X+, __tmp_reg__ \n\t"
"Lcpyodd_%=: \n\t"
" ld __tmp_reg__, Z+ \n\t"
" st X+, __tmp_reg__ \n\t"
"Lcpyeven_%=: \n\t"
" subi %A[num], 2 \n\t"
" brcc Lcpyloop_%= \n\t"
" sbci %B[num], 0 \n\t"
" brcc Lcpyloop_%= \n\t"
: [num] "+r" (num)
: [src] "r" (src),
[dst] "r" (dst)
: "memory"
);
return dst;
}
//__attribute__ ((noinline))
void * memmove8 ( void * dst, void* src, uint16_t num )
{
if( src > dst) {
// if src > dst then we can use the forward-stepping memcpy8
return memcpy8( dst, src, num);
} else {
// if src < dst then we have to step backward:
dst = (char*)dst + num;
src = (char*)src + num;
asm volatile(
" movw r30, %[src] \n\t"
" movw r26, %[dst] \n\t"
" sbrs %A[num], 0 \n\t"
" rjmp Lmoveven_%= \n\t"
" rjmp Lmovodd_%= \n\t"
"Lmovloop_%=: \n\t"
" ld __tmp_reg__, -Z \n\t"
" st -X, __tmp_reg__ \n\t"
"Lmovodd_%=: \n\t"
" ld __tmp_reg__, -Z \n\t"
" st -X, __tmp_reg__ \n\t"
"Lmoveven_%=: \n\t"
" subi %A[num], 2 \n\t"
" brcc Lmovloop_%= \n\t"
" sbci %B[num], 0 \n\t"
" brcc Lmovloop_%= \n\t"
: [num] "+r" (num)
: [src] "r" (src),
[dst] "r" (dst)
: "memory"
);
return dst;
}
}
} /* end extern "C" */
#endif /* AVR */
#if 0
// TEST / VERIFICATION CODE ONLY BELOW THIS POINT
#include <Arduino.h>
#include "lib8tion.h"
void test1abs( int8_t i)
{
Serial.print("abs("); Serial.print(i); Serial.print(") = ");
int8_t j = abs8(i);
Serial.print(j); Serial.println(" ");
}
void testabs()
{
delay(5000);
for( int8_t q = -128; q != 127; q++) {
test1abs(q);
}
for(;;){};
}
void testmul8()
{
delay(5000);
byte r, c;
Serial.println("mul8:");
for( r = 0; r <= 20; r += 1) {
Serial.print(r); Serial.print(" : ");
for( c = 0; c <= 20; c += 1) {
byte t;
t = mul8( r, c);
Serial.print(t); Serial.print(' ');
}
Serial.println(' ');
}
Serial.println("done.");
for(;;){};
}
void testscale8()
{
delay(5000);
byte r, c;
Serial.println("scale8:");
for( r = 0; r <= 240; r += 10) {
Serial.print(r); Serial.print(" : ");
for( c = 0; c <= 240; c += 10) {
byte t;
t = scale8( r, c);
Serial.print(t); Serial.print(' ');
}
Serial.println(' ');
}
Serial.println(' ');
Serial.println("scale8_video:");
for( r = 0; r <= 100; r += 4) {
Serial.print(r); Serial.print(" : ");
for( c = 0; c <= 100; c += 4) {
byte t;
t = scale8_video( r, c);
Serial.print(t); Serial.print(' ');
}
Serial.println(' ');
}
Serial.println("done.");
for(;;){};
}
void testqadd8()
{
delay(5000);
byte r, c;
for( r = 0; r <= 240; r += 10) {
Serial.print(r); Serial.print(" : ");
for( c = 0; c <= 240; c += 10) {
byte t;
t = qadd8( r, c);
Serial.print(t); Serial.print(' ');
}
Serial.println(' ');
}
Serial.println("done.");
for(;;){};
}
void testnscale8x3()
{
delay(5000);
byte r, g, b, sc;
for( byte z = 0; z < 10; z++) {
r = random8(); g = random8(); b = random8(); sc = random8();
Serial.print("nscale8x3_video( ");
Serial.print(r); Serial.print(", ");
Serial.print(g); Serial.print(", ");
Serial.print(b); Serial.print(", ");
Serial.print(sc); Serial.print(") = [ ");
nscale8x3_video( r, g, b, sc);
Serial.print(r); Serial.print(", ");
Serial.print(g); Serial.print(", ");
Serial.print(b); Serial.print("]");
Serial.println(' ');
}
Serial.println("done.");
for(;;){};
}
#endif