summaryrefslogtreecommitdiffstats
path: root/src/3rdparty/angle/src/common/mathutil.h
blob: 1015bd23120a07a36dbf793d6f515755fce99b03 (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
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
//
// Copyright (c) 2002-2013 The ANGLE Project Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
//

// mathutil.h: Math and bit manipulation functions.

#ifndef COMMON_MATHUTIL_H_
#define COMMON_MATHUTIL_H_

#include "common/debug.h"
#include "common/platform.h"

#include <limits>
#include <algorithm>
#include <string.h>
#include <stdlib.h>

namespace gl
{

const unsigned int Float32One = 0x3F800000;
const unsigned short Float16One = 0x3C00;

struct Vector4
{
    Vector4() {}
    Vector4(float x, float y, float z, float w) : x(x), y(y), z(z), w(w) {}

    float x;
    float y;
    float z;
    float w;
};

inline bool isPow2(int x)
{
    return (x & (x - 1)) == 0 && (x != 0);
}

inline int log2(int x)
{
    int r = 0;
    while ((x >> r) > 1) r++;
    return r;
}

inline unsigned int ceilPow2(unsigned int x)
{
    if (x != 0) x--;
    x |= x >> 1;
    x |= x >> 2;
    x |= x >> 4;
    x |= x >> 8;
    x |= x >> 16;
    x++;

    return x;
}

inline int clampToInt(unsigned int x)
{
    return static_cast<int>(std::min(x, static_cast<unsigned int>(std::numeric_limits<int>::max())));
}

template <typename DestT, typename SrcT>
inline DestT clampCast(SrcT value)
{
    // This assumes SrcT can properly represent DestT::min/max
    // Unfortunately we can't use META_ASSERT without C++11 constexpr support
    ASSERT(static_cast<DestT>(static_cast<SrcT>(std::numeric_limits<DestT>::min())) == std::numeric_limits<DestT>::min());
    ASSERT(static_cast<DestT>(static_cast<SrcT>(std::numeric_limits<DestT>::max())) == std::numeric_limits<DestT>::max());

    SrcT lo = static_cast<SrcT>(std::numeric_limits<DestT>::min());
    SrcT hi = static_cast<SrcT>(std::numeric_limits<DestT>::max());
    return static_cast<DestT>(value > lo ? (value > hi ? hi : value) : lo);
}

template<typename T, typename MIN, typename MAX>
inline T clamp(T x, MIN min, MAX max)
{
    // Since NaNs fail all comparison tests, a NaN value will default to min
    return x > min ? (x > max ? max : x) : min;
}

inline float clamp01(float x)
{
    return clamp(x, 0.0f, 1.0f);
}

template<const int n>
inline unsigned int unorm(float x)
{
    const unsigned int max = 0xFFFFFFFF >> (32 - n);

    if (x > 1)
    {
        return max;
    }
    else if (x < 0)
    {
        return 0;
    }
    else
    {
        return (unsigned int)(max * x + 0.5f);
    }
}

inline bool supportsSSE2()
{
#if defined(ANGLE_PLATFORM_WINDOWS) && !defined(_M_ARM)
    static bool checked = false;
    static bool supports = false;

    if (checked)
    {
        return supports;
    }

#if defined(__GNUC__)
    supports = __builtin_cpu_supports("sse2");
#else
    int info[4];
    __cpuid(info, 0);

    if (info[0] >= 1)
    {
        __cpuid(info, 1);

        supports = (info[3] >> 26) & 1;
    }
#endif

    checked = true;

    return supports;
#else
    UNIMPLEMENTED();
    return false;
#endif
}

template <typename destType, typename sourceType>
destType bitCast(const sourceType &source)
{
    size_t copySize = std::min(sizeof(destType), sizeof(sourceType));
    destType output;
    memcpy(&output, &source, copySize);
    return output;
}

inline unsigned short float32ToFloat16(float fp32)
{
    unsigned int fp32i = (unsigned int&)fp32;
    unsigned int sign = (fp32i & 0x80000000) >> 16;
    unsigned int abs = fp32i & 0x7FFFFFFF;

    if(abs > 0x47FFEFFF)   // Infinity
    {
        return sign | 0x7FFF;
    }
    else if(abs < 0x38800000)   // Denormal
    {
        unsigned int mantissa = (abs & 0x007FFFFF) | 0x00800000;
        int e = 113 - (abs >> 23);

        if(e < 24)
        {
            abs = mantissa >> e;
        }
        else
        {
            abs = 0;
        }

        return sign | (abs + 0x00000FFF + ((abs >> 13) & 1)) >> 13;
    }
    else
    {
        return sign | (abs + 0xC8000000 + 0x00000FFF + ((abs >> 13) & 1)) >> 13;
    }
}

float float16ToFloat32(unsigned short h);

unsigned int convertRGBFloatsTo999E5(float red, float green, float blue);
void convert999E5toRGBFloats(unsigned int input, float *red, float *green, float *blue);

inline unsigned short float32ToFloat11(float fp32)
{
    const unsigned int float32MantissaMask = 0x7FFFFF;
    const unsigned int float32ExponentMask = 0x7F800000;
    const unsigned int float32SignMask = 0x80000000;
    const unsigned int float32ValueMask = ~float32SignMask;
    const unsigned int float32ExponentFirstBit = 23;
    const unsigned int float32ExponentBias = 127;

    const unsigned short float11Max = 0x7BF;
    const unsigned short float11MantissaMask = 0x3F;
    const unsigned short float11ExponentMask = 0x7C0;
    const unsigned short float11BitMask = 0x7FF;
    const unsigned int float11ExponentBias = 14;

    const unsigned int float32Maxfloat11 = 0x477E0000;
    const unsigned int float32Minfloat11 = 0x38800000;

    const unsigned int float32Bits = bitCast<unsigned int>(fp32);
    const bool float32Sign = (float32Bits & float32SignMask) == float32SignMask;

    unsigned int float32Val = float32Bits & float32ValueMask;

    if ((float32Val & float32ExponentMask) == float32ExponentMask)
    {
        // INF or NAN
        if ((float32Val & float32MantissaMask) != 0)
        {
            return float11ExponentMask | (((float32Val >> 17) | (float32Val >> 11) | (float32Val >> 6) | (float32Val)) & float11MantissaMask);
        }
        else if (float32Sign)
        {
            // -INF is clamped to 0 since float11 is positive only
            return 0;
        }
        else
        {
            return float11ExponentMask;
        }
    }
    else if (float32Sign)
    {
        // float11 is positive only, so clamp to zero
        return 0;
    }
    else if (float32Val > float32Maxfloat11)
    {
        // The number is too large to be represented as a float11, set to max
        return float11Max;
    }
    else
    {
        if (float32Val < float32Minfloat11)
        {
            // The number is too small to be represented as a normalized float11
            // Convert it to a denormalized value.
            const unsigned int shift = (float32ExponentBias - float11ExponentBias) - (float32Val >> float32ExponentFirstBit);
            float32Val = ((1 << float32ExponentFirstBit) | (float32Val & float32MantissaMask)) >> shift;
        }
        else
        {
            // Rebias the exponent to represent the value as a normalized float11
            float32Val += 0xC8000000;
        }

        return ((float32Val + 0xFFFF + ((float32Val >> 17) & 1)) >> 17) & float11BitMask;
    }
}

inline unsigned short float32ToFloat10(float fp32)
{
    const unsigned int float32MantissaMask = 0x7FFFFF;
    const unsigned int float32ExponentMask = 0x7F800000;
    const unsigned int float32SignMask = 0x80000000;
    const unsigned int float32ValueMask = ~float32SignMask;
    const unsigned int float32ExponentFirstBit = 23;
    const unsigned int float32ExponentBias = 127;

    const unsigned short float10Max = 0x3DF;
    const unsigned short float10MantissaMask = 0x1F;
    const unsigned short float10ExponentMask = 0x3E0;
    const unsigned short float10BitMask = 0x3FF;
    const unsigned int float10ExponentBias = 14;

    const unsigned int float32Maxfloat10 = 0x477C0000;
    const unsigned int float32Minfloat10 = 0x38800000;

    const unsigned int float32Bits = bitCast<unsigned int>(fp32);
    const bool float32Sign = (float32Bits & float32SignMask) == float32SignMask;

    unsigned int float32Val = float32Bits & float32ValueMask;

    if ((float32Val & float32ExponentMask) == float32ExponentMask)
    {
        // INF or NAN
        if ((float32Val & float32MantissaMask) != 0)
        {
            return float10ExponentMask | (((float32Val >> 18) | (float32Val >> 13) | (float32Val >> 3) | (float32Val)) & float10MantissaMask);
        }
        else if (float32Sign)
        {
            // -INF is clamped to 0 since float11 is positive only
            return 0;
        }
        else
        {
            return float10ExponentMask;
        }
    }
    else if (float32Sign)
    {
        // float10 is positive only, so clamp to zero
        return 0;
    }
    else if (float32Val > float32Maxfloat10)
    {
        // The number is too large to be represented as a float11, set to max
        return float10Max;
    }
    else
    {
        if (float32Val < float32Minfloat10)
        {
            // The number is too small to be represented as a normalized float11
            // Convert it to a denormalized value.
            const unsigned int shift = (float32ExponentBias - float10ExponentBias) - (float32Val >> float32ExponentFirstBit);
            float32Val = ((1 << float32ExponentFirstBit) | (float32Val & float32MantissaMask)) >> shift;
        }
        else
        {
            // Rebias the exponent to represent the value as a normalized float11
            float32Val += 0xC8000000;
        }

        return ((float32Val + 0x1FFFF + ((float32Val >> 18) & 1)) >> 18) & float10BitMask;
    }
}

inline float float11ToFloat32(unsigned short fp11)
{
    unsigned short exponent = (fp11 >> 6) & 0x1F;
    unsigned short mantissa = fp11 & 0x3F;

    if (exponent == 0x1F)
    {
        // INF or NAN
        return bitCast<float>(0x7f800000 | (mantissa << 17));
    }
    else
    {
        if (exponent != 0)
        {
            // normalized
        }
        else if (mantissa != 0)
        {
            // The value is denormalized
            exponent = 1;

            do
            {
                exponent--;
                mantissa <<= 1;
            }
            while ((mantissa & 0x40) == 0);

            mantissa = mantissa & 0x3F;
        }
        else // The value is zero
        {
            exponent = static_cast<unsigned short>(-112);
        }

        return bitCast<float>(((exponent + 112) << 23) | (mantissa << 17));
    }
}

inline float float10ToFloat32(unsigned short fp11)
{
    unsigned short exponent = (fp11 >> 5) & 0x1F;
    unsigned short mantissa = fp11 & 0x1F;

    if (exponent == 0x1F)
    {
        // INF or NAN
        return bitCast<float>(0x7f800000 | (mantissa << 17));
    }
    else
    {
        if (exponent != 0)
        {
            // normalized
        }
        else if (mantissa != 0)
        {
            // The value is denormalized
            exponent = 1;

            do
            {
                exponent--;
                mantissa <<= 1;
            }
            while ((mantissa & 0x20) == 0);

            mantissa = mantissa & 0x1F;
        }
        else // The value is zero
        {
            exponent = static_cast<unsigned short>(-112);
        }

        return bitCast<float>(((exponent + 112) << 23) | (mantissa << 18));
    }
}

template <typename T>
inline float normalizedToFloat(T input)
{
    static_assert(std::numeric_limits<T>::is_integer, "T must be an integer.");

    const float inverseMax = 1.0f / std::numeric_limits<T>::max();
    return input * inverseMax;
}

template <unsigned int inputBitCount, typename T>
inline float normalizedToFloat(T input)
{
    static_assert(std::numeric_limits<T>::is_integer, "T must be an integer.");
    static_assert(inputBitCount < (sizeof(T) * 8), "T must have more bits than inputBitCount.");

    const float inverseMax = 1.0f / ((1 << inputBitCount) - 1);
    return input * inverseMax;
}

template <typename T>
inline T floatToNormalized(float input)
{
    return std::numeric_limits<T>::max() * input + 0.5f;
}

template <unsigned int outputBitCount, typename T>
inline T floatToNormalized(float input)
{
    static_assert(outputBitCount < (sizeof(T) * 8), "T must have more bits than outputBitCount.");
    return ((1 << outputBitCount) - 1) * input + 0.5f;
}

template <unsigned int inputBitCount, unsigned int inputBitStart, typename T>
inline T getShiftedData(T input)
{
    static_assert(inputBitCount + inputBitStart <= (sizeof(T) * 8),
                  "T must have at least as many bits as inputBitCount + inputBitStart.");
    const T mask = (1 << inputBitCount) - 1;
    return (input >> inputBitStart) & mask;
}

template <unsigned int inputBitCount, unsigned int inputBitStart, typename T>
inline T shiftData(T input)
{
    static_assert(inputBitCount + inputBitStart <= (sizeof(T) * 8),
                  "T must have at least as many bits as inputBitCount + inputBitStart.");
    const T mask = (1 << inputBitCount) - 1;
    return (input & mask) << inputBitStart;
}


inline unsigned char average(unsigned char a, unsigned char b)
{
    return ((a ^ b) >> 1) + (a & b);
}

inline signed char average(signed char a, signed char b)
{
    return ((short)a + (short)b) / 2;
}

inline unsigned short average(unsigned short a, unsigned short b)
{
    return ((a ^ b) >> 1) + (a & b);
}

inline signed short average(signed short a, signed short b)
{
    return ((int)a + (int)b) / 2;
}

inline unsigned int average(unsigned int a, unsigned int b)
{
    return ((a ^ b) >> 1) + (a & b);
}

inline signed int average(signed int a, signed int b)
{
    return ((long long)a + (long long)b) / 2;
}

inline float average(float a, float b)
{
    return (a + b) * 0.5f;
}

inline unsigned short averageHalfFloat(unsigned short a, unsigned short b)
{
    return float32ToFloat16((float16ToFloat32(a) + float16ToFloat32(b)) * 0.5f);
}

inline unsigned int averageFloat11(unsigned int a, unsigned int b)
{
    return float32ToFloat11((float11ToFloat32(a) + float11ToFloat32(b)) * 0.5f);
}

inline unsigned int averageFloat10(unsigned int a, unsigned int b)
{
    return float32ToFloat10((float10ToFloat32(a) + float10ToFloat32(b)) * 0.5f);
}

}

namespace rx
{

// Represents intervals of the type [a, b)
template <typename T>
struct Range
{
    Range() {}
    Range(T lo, T hi) : start(lo), end(hi) { ASSERT(lo <= hi); }

    T start;
    T end;

    T length() const { return end - start; }

    bool intersects(Range<T> other)
    {
        if (start <= other.start)
        {
            return other.start < end;
        }
        else
        {
            return start < other.end;
        }
    }
};

typedef Range<int> RangeI;
typedef Range<unsigned int> RangeUI;

template <typename T>
T roundUp(const T value, const T alignment)
{
    return value + alignment - 1 - (value - 1) % alignment;
}

inline unsigned int UnsignedCeilDivide(unsigned int value, unsigned int divisor)
{
    unsigned int divided = value / divisor;
    return (divided + ((value % divisor == 0) ? 0 : 1));
}

template <class T>
inline bool IsUnsignedAdditionSafe(T lhs, T rhs)
{
    static_assert(!std::numeric_limits<T>::is_signed, "T must be unsigned.");
    return (rhs <= std::numeric_limits<T>::max() - lhs);
}

template <class T>
inline bool IsUnsignedMultiplicationSafe(T lhs, T rhs)
{
    static_assert(!std::numeric_limits<T>::is_signed, "T must be unsigned.");
    return (lhs == T(0) || rhs == T(0) || (rhs <= std::numeric_limits<T>::max() / lhs));
}

template <class SmallIntT, class BigIntT>
inline bool IsIntegerCastSafe(BigIntT bigValue)
{
    return (static_cast<BigIntT>(static_cast<SmallIntT>(bigValue)) == bigValue);
}

#if defined(_MSC_VER)

#define ANGLE_ROTL(x,y) _rotl(x,y)

#else

inline uint32_t RotL(uint32_t x, int8_t r)
{
    return (x << r) | (x >> (32 - r));
}

#define ANGLE_ROTL(x,y) RotL(x,y)

#endif // namespace rx

}

#endif   // COMMON_MATHUTIL_H_