aboutsummaryrefslogtreecommitdiffstats
path: root/lib/lib8tion/scale8.h
blob: 9895fd4d7919e6191679c75660d75254f2c8f123 (plain)
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
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
#ifndef __INC_LIB8TION_SCALE_H
#define __INC_LIB8TION_SCALE_H

///@ingroup lib8tion

///@defgroup Scaling Scaling functions
/// Fast, efficient 8-bit scaling functions specifically
/// designed for high-performance LED programming.
///
/// Because of the AVR(Arduino) and ARM assembly language
/// implementations provided, using these functions often
/// results in smaller and faster code than the equivalent
/// program using plain "C" arithmetic and logic.
///@{

///  scale one byte by a second one, which is treated as
///  the numerator of a fraction whose denominator is 256
///  In other words, it computes i * (scale / 256)
///  4 clocks AVR with MUL, 2 clocks ARM
LIB8STATIC_ALWAYS_INLINE uint8_t scale8( uint8_t i, fract8 scale)
{
#if SCALE8_C == 1
#if (FASTLED_SCALE8_FIXED == 1)
    return (((uint16_t)i) * (1+(uint16_t)(scale))) >> 8;
#else
    return ((uint16_t)i * (uint16_t)(scale) ) >> 8;
#endif
#elif SCALE8_AVRASM == 1
#if defined(LIB8_ATTINY)
#if (FASTLED_SCALE8_FIXED == 1)
    uint8_t work=i;
#else
    uint8_t work=0;
#endif
    uint8_t cnt=0x80;
    asm volatile(
#if (FASTLED_SCALE8_FIXED == 1)
        "  inc %[scale]                 \n\t"
        "  breq DONE_%=                 \n\t"
        "  clr %[work]                  \n\t"
#endif
        "LOOP_%=:                       \n\t"
        /*"  sbrc %[scale], 0             \n\t"
        "  add %[work], %[i]            \n\t"
        "  ror %[work]                  \n\t"
        "  lsr %[scale]                 \n\t"
        "  clc                          \n\t"*/
        "  sbrc %[scale], 0             \n\t"
        "  add %[work], %[i]            \n\t"
        "  ror %[work]                  \n\t"
        "  lsr %[scale]                 \n\t"
        "  lsr %[cnt]                   \n\t"
        "brcc LOOP_%=                   \n\t"
        "DONE_%=:                       \n\t"
        : [work] "+r" (work), [cnt] "+r" (cnt)
        : [scale] "r" (scale), [i] "r" (i)
        :
      );
    return work;
#else
    asm volatile(
#if (FASTLED_SCALE8_FIXED==1)
        // Multiply 8-bit i * 8-bit scale, giving 16-bit r1,r0
        "mul %0, %1          \n\t"
        // Add i to r0, possibly setting the carry flag
        "add r0, %0         \n\t"
        // load the immediate 0 into i (note, this does _not_ touch any flags)
        "ldi %0, 0x00       \n\t"
        // walk and chew gum at the same time
        "adc %0, r1          \n\t"
#else
         /* Multiply 8-bit i * 8-bit scale, giving 16-bit r1,r0 */
         "mul %0, %1          \n\t"
         /* Move the high 8-bits of the product (r1) back to i */
         "mov %0, r1          \n\t"
         /* Restore r1 to "0"; it's expected to always be that */
#endif
         "clr __zero_reg__    \n\t"

         : "+a" (i)      /* writes to i */
         : "a"  (scale)  /* uses scale */
         : "r0", "r1"    /* clobbers r0, r1 */ );

    /* Return the result */
    return i;
#endif
#else
#error "No implementation for scale8 available."
#endif
}


///  The "video" version of scale8 guarantees that the output will
///  be only be zero if one or both of the inputs are zero.  If both
///  inputs are non-zero, the output is guaranteed to be non-zero.
///  This makes for better 'video'/LED dimming, at the cost of
///  several additional cycles.
LIB8STATIC_ALWAYS_INLINE uint8_t scale8_video( uint8_t i, fract8 scale)
{
#if SCALE8_C == 1 || defined(LIB8_ATTINY)
    uint8_t j = (((int)i * (int)scale) >> 8) + ((i&&scale)?1:0);
    // uint8_t nonzeroscale = (scale != 0) ? 1 : 0;
    // uint8_t j = (i == 0) ? 0 : (((int)i * (int)(scale) ) >> 8) + nonzeroscale;
    return j;
#elif SCALE8_AVRASM == 1
    uint8_t j=0;
    asm volatile(
        "  tst %[i]\n\t"
        "  breq L_%=\n\t"
        "  mul %[i], %[scale]\n\t"
        "  mov %[j], r1\n\t"
        "  clr __zero_reg__\n\t"
        "  cpse %[scale], r1\n\t"
        "  subi %[j], 0xFF\n\t"
        "L_%=: \n\t"
        : [j] "+a" (j)
        : [i] "a" (i), [scale] "a" (scale)
        : "r0", "r1");

    return j;
    // uint8_t nonzeroscale = (scale != 0) ? 1 : 0;
    // asm volatile(
    //      "      tst %0           \n"
    //      "      breq L_%=        \n"
    //      "      mul %0, %1       \n"
    //      "      mov %0, r1       \n"
    //      "      add %0, %2       \n"
    //      "      clr __zero_reg__ \n"
    //      "L_%=:                  \n"

    //      : "+a" (i)
    //      : "a" (scale), "a" (nonzeroscale)
    //      : "r0", "r1");

    // // Return the result
    // return i;
#else
#error "No implementation for scale8_video available."
#endif
}


/// This version of scale8 does not clean up the R1 register on AVR
/// If you are doing several 'scale8's in a row, use this, and
/// then explicitly call cleanup_R1.
LIB8STATIC_ALWAYS_INLINE uint8_t scale8_LEAVING_R1_DIRTY( uint8_t i, fract8 scale)
{
#if SCALE8_C == 1
#if (FASTLED_SCALE8_FIXED == 1)
    return (((uint16_t)i) * ((uint16_t)(scale)+1)) >> 8;
#else
    return ((int)i * (int)(scale) ) >> 8;
#endif
#elif SCALE8_AVRASM == 1
    asm volatile(
      #if (FASTLED_SCALE8_FIXED==1)
              // Multiply 8-bit i * 8-bit scale, giving 16-bit r1,r0
              "mul %0, %1          \n\t"
              // Add i to r0, possibly setting the carry flag
              "add r0, %0         \n\t"
              // load the immediate 0 into i (note, this does _not_ touch any flags)
              "ldi %0, 0x00       \n\t"
              // walk and chew gum at the same time
              "adc %0, r1          \n\t"
      #else
         /* Multiply 8-bit i * 8-bit scale, giving 16-bit r1,r0 */
         "mul %0, %1    \n\t"
         /* Move the high 8-bits of the product (r1) back to i */
         "mov %0, r1    \n\t"
      #endif
         /* R1 IS LEFT DIRTY HERE; YOU MUST ZERO IT OUT YOURSELF  */
         /* "clr __zero_reg__    \n\t" */

         : "+a" (i)      /* writes to i */
         : "a"  (scale)  /* uses scale */
         : "r0", "r1"    /* clobbers r0, r1 */ );

    // Return the result
    return i;
#else
#error "No implementation for scale8_LEAVING_R1_DIRTY available."
#endif
}


/// This version of scale8_video does not clean up the R1 register on AVR
/// If you are doing several 'scale8_video's in a row, use this, and
/// then explicitly call cleanup_R1.
LIB8STATIC_ALWAYS_INLINE uint8_t scale8_video_LEAVING_R1_DIRTY( uint8_t i, fract8 scale)
{
#if SCALE8_C == 1 || defined(LIB8_ATTINY)
    uint8_t j = (((int)i * (int)scale) >> 8) + ((i&&scale)?1:0);
    // uint8_t nonzeroscale = (scale != 0) ? 1 : 0;
    // uint8_t j = (i == 0) ? 0 : (((int)i * (int)(scale) ) >> 8) + nonzeroscale;
    return j;
#elif SCALE8_AVRASM == 1
    uint8_t j=0;
    asm volatile(
        "  tst %[i]\n\t"
        "  breq L_%=\n\t"
        "  mul %[i], %[scale]\n\t"
        "  mov %[j], r1\n\t"
        "  breq L_%=\n\t"
        "  subi %[j], 0xFF\n\t"
        "L_%=: \n\t"
        : [j] "+a" (j)
        : [i] "a" (i), [scale] "a" (scale)
        : "r0", "r1");

    return j;
    // uint8_t nonzeroscale = (scale != 0) ? 1 : 0;
    // asm volatile(
    //      "      tst %0           \n"
    //      "      breq L_%=        \n"
    //      "      mul %0, %1       \n"
    //      "      mov %0, r1       \n"
    //      "      add %0, %2       \n"
    //      "      clr __zero_reg__ \n"
    //      "L_%=:                  \n"

    //      : "+a" (i)
    //      : "a" (scale), "a" (nonzeroscale)
    //      : "r0", "r1");

    // // Return the result
    // return i;
#else
#error "No implementation for scale8_video_LEAVING_R1_DIRTY available."
#endif
}

/// Clean up the r1 register after a series of *LEAVING_R1_DIRTY calls
LIB8STATIC_ALWAYS_INLINE void cleanup_R1(void)
{
#if CLEANUP_R1_AVRASM == 1
    // Restore r1 to "0"; it's expected to always be that
    asm volatile( "clr __zero_reg__  \n\t" : : : "r1" );
#endif
}


/// scale a 16-bit unsigned value by an 8-bit value,
///         considered as numerator of a fraction whose denominator
///         is 256. In other words, it computes i * (scale / 256)

LIB8STATIC_ALWAYS_INLINE uint16_t scale16by8( uint16_t i, fract8 scale )
{
#if SCALE16BY8_C == 1
    uint16_t result;
#if FASTLED_SCALE8_FIXED == 1
    result = (i * (1+((uint16_t)scale))) >> 8;
#else
    result = (i * scale) / 256;
#endif
    return result;
#elif SCALE16BY8_AVRASM == 1
#if FASTLED_SCALE8_FIXED == 1
    uint16_t result = 0;
    asm volatile(
                 // result.A = HighByte( (i.A x scale) + i.A )
                 "  mul %A[i], %[scale]                 \n\t"
                 "  add r0, %A[i]                       \n\t"
            //   "  adc r1, [zero]                      \n\t"
            //   "  mov %A[result], r1                  \n\t"
                 "  adc %A[result], r1                  \n\t"

                 // result.A-B += i.B x scale
                 "  mul %B[i], %[scale]                 \n\t"
                 "  add %A[result], r0                  \n\t"
                 "  adc %B[result], r1                  \n\t"

                 // cleanup r1
                 "  clr __zero_reg__                    \n\t"

                 // result.A-B += i.B
                 "  add %A[result], %B[i]               \n\t"
                 "  adc %B[result], __zero_reg__        \n\t"

                 : [result] "+r" (result)
                 : [i] "r" (i), [scale] "r" (scale)
                 : "r0", "r1"
                 );
    return result;
#else
    uint16_t result = 0;
    asm volatile(
         // result.A = HighByte(i.A x j )
         "  mul %A[i], %[scale]                 \n\t"
         "  mov %A[result], r1                  \n\t"
         //"  clr %B[result]                      \n\t"

         // result.A-B += i.B x j
         "  mul %B[i], %[scale]                 \n\t"
         "  add %A[result], r0                  \n\t"
         "  adc %B[result], r1                  \n\t"

         // cleanup r1
         "  clr __zero_reg__                    \n\t"

         : [result] "+r" (result)
         : [i] "r" (i), [scale] "r" (scale)
         : "r0", "r1"
         );
    return result;
#endif
#else
    #error "No implementation for scale16by8 available."
#endif
}

/// scale a 16-bit unsigned value by a 16-bit value,
///         considered as numerator of a fraction whose denominator
///         is 65536. In other words, it computes i * (scale / 65536)

LIB8STATIC uint16_t scale16( uint16_t i, fract16 scale )
{
  #if SCALE16_C == 1
    uint16_t result;
#if FASTLED_SCALE8_FIXED == 1
    result = ((uint32_t)(i) * (1+(uint32_t)(scale))) / 65536;
#else
    result = ((uint32_t)(i) * (uint32_t)(scale)) / 65536;
#endif
    return result;
#elif SCALE16_AVRASM == 1
#if FASTLED_SCALE8_FIXED == 1
    // implemented sort of like
    //   result = ((i * scale) + i ) / 65536
    //
    // why not like this, you may ask?
    //   result = (i * (scale+1)) / 65536
    // the answer is that if scale is 65535, then scale+1
    // will be zero, which is not what we want.
    uint32_t result;
    asm volatile(
                 // result.A-B  = i.A x scale.A
                 "  mul %A[i], %A[scale]                 \n\t"
                 //  save results...
                 // basic idea:
                 //"  mov %A[result], r0                 \n\t"
                 //"  mov %B[result], r1                 \n\t"
                 // which can be written as...
                 "  movw %A[result], r0                   \n\t"
                 // Because we're going to add i.A-B to
                 // result.A-D, we DO need to keep both
                 // the r0 and r1 portions of the product
                 // UNlike in the 'unfixed scale8' version.
                 // So the movw here is needed.
                 : [result] "=r" (result)
                 : [i] "r" (i),
                 [scale] "r" (scale)
                 : "r0", "r1"
                 );

    asm volatile(
                 // result.C-D  = i.B x scale.B
                 "  mul %B[i], %B[scale]                 \n\t"
                 //"  mov %C[result], r0                 \n\t"
                 //"  mov %D[result], r1                 \n\t"
                 "  movw %C[result], r0                   \n\t"
                 : [result] "+r" (result)
                 : [i] "r" (i),
                 [scale] "r" (scale)
                 : "r0", "r1"
                 );

    const uint8_t  zero = 0;
    asm volatile(
                 // result.B-D += i.B x scale.A
                 "  mul %B[i], %A[scale]                 \n\t"

                 "  add %B[result], r0                   \n\t"
                 "  adc %C[result], r1                   \n\t"
                 "  adc %D[result], %[zero]              \n\t"

                 // result.B-D += i.A x scale.B
                 "  mul %A[i], %B[scale]                 \n\t"

                 "  add %B[result], r0                   \n\t"
                 "  adc %C[result], r1                   \n\t"
                 "  adc %D[result], %[zero]              \n\t"

                 // cleanup r1
                 "  clr r1                               \n\t"

                 : [result] "+r" (result)
                 : [i] "r" (i),
                 [scale] "r" (scale),
                 [zero] "r" (zero)
                 : "r0", "r1"
                 );

    asm volatile(
                 // result.A-D += i.A-B
                 "  add %A[result], %A[i]                \n\t"
                 "  adc %B[result], %B[i]                \n\t"
                 "  adc %C[result], %[zero]              \n\t"
                 "  adc %D[result], %[zero]              \n\t"
                 : [result] "+r" (result)
                 : [i] "r" (i),
                 [zero] "r" (zero)
                 );

    result = result >> 16;
    return result;
#else
    uint32_t result;
    asm volatile(
                 // result.A-B  = i.A x scale.A
                 "  mul %A[i], %A[scale]                 \n\t"
                 //  save results...
                 // basic idea:
                 //"  mov %A[result], r0                 \n\t"
                 //"  mov %B[result], r1                 \n\t"
                 // which can be written as...
                 "  movw %A[result], r0                   \n\t"
                 // We actually don't need to do anything with r0,
                 // as result.A is never used again here, so we
                 // could just move the high byte, but movw is
                 // one clock cycle, just like mov, so might as
                 // well, in case we want to use this code for
                 // a generic 16x16 multiply somewhere.

                 : [result] "=r" (result)
                 : [i] "r" (i),
                   [scale] "r" (scale)
                 : "r0", "r1"
                 );

    asm volatile(
                 // result.C-D  = i.B x scale.B
                 "  mul %B[i], %B[scale]                 \n\t"
                 //"  mov %C[result], r0                 \n\t"
                 //"  mov %D[result], r1                 \n\t"
                 "  movw %C[result], r0                   \n\t"
                 : [result] "+r" (result)
                 : [i] "r" (i),
                   [scale] "r" (scale)
                 : "r0", "r1"
                 );

    const uint8_t  zero = 0;
    asm volatile(
                 // result.B-D += i.B x scale.A
                 "  mul %B[i], %A[scale]                 \n\t"

                 "  add %B[result], r0                   \n\t"
                 "  adc %C[result], r1                   \n\t"
                 "  adc %D[result], %[zero]              \n\t"

                 // result.B-D += i.A x scale.B
                 "  mul %A[i], %B[scale]                 \n\t"

                 "  add %B[result], r0                   \n\t"
                 "  adc %C[result], r1                   \n\t"
                 "  adc %D[result], %[zero]              \n\t"

                 // cleanup r1
                 "  clr r1                               \n\t"

                 : [result] "+r" (result)
                 : [i] "r" (i),
                   [scale] "r" (scale),
                   [zero] "r" (zero)
                 : "r0", "r1"
                 );

    result = result >> 16;
    return result;
#endif
#else
    #error "No implementation for scale16 available."
#endif
}
///@}

///@defgroup Dimming Dimming and brightening functions
///
/// Dimming and brightening functions
///
/// The eye does not respond in a linear way to light.
/// High speed PWM'd LEDs at 50% duty cycle appear far
/// brighter then the 'half as bright' you might expect.
///
/// If you want your midpoint brightness leve (128) to
/// appear half as bright as 'full' brightness (255), you
/// have to apply a 'dimming function'.
///@{

/// Adjust a scaling value for dimming
LIB8STATIC uint8_t dim8_raw( uint8_t x)
{
    return scale8( x, x);
}

/// Adjust a scaling value for dimming for video (value will never go below 1)
LIB8STATIC uint8_t dim8_video( uint8_t x)
{
    return scale8_video( x, x);
}

/// Linear version of the dimming function that halves for values < 128
LIB8STATIC uint8_t dim8_lin( uint8_t x )
{
    if( x & 0x80 ) {
        x = scale8( x, x);
    } else {
        x += 1;
        x /= 2;
    }
    return x;
}

/// inverse of the dimming function, brighten a value
LIB8STATIC uint8_t brighten8_raw( uint8_t x)
{
    uint8_t ix = 255 - x;
    return 255 - scale8( ix, ix);
}

/// inverse of the dimming function, brighten a value
LIB8STATIC uint8_t brighten8_video( uint8_t x)
{
    uint8_t ix = 255 - x;
    return 255 - scale8_video( ix, ix);
}

/// inverse of the dimming function, brighten a value
LIB8STATIC uint8_t brighten8_lin( uint8_t x )
{
    uint8_t ix = 255 - x;
    if( ix & 0x80 ) {
        ix = scale8( ix, ix);
    } else {
        ix += 1;
        ix /= 2;
    }
    return 255 - ix;
}

///@}
#endif