Skip to content
Navigation Menu
{{ message }}
forked from rive-app/rive-runtime
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathlua_buffer_ext.cpp
More file actions
538 lines (482 loc) · 15 KB
/
Copy pathlua_buffer_ext.cpp
File metadata and controls
538 lines (482 loc) · 15 KB
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
#ifdef WITH_RIVE_SCRIPTING
#include "lualib.h"
#include <cstring>
#include <cstdint>
#if defined(__aarch64__) || defined(__ARM_NEON__) || defined(__ARM_NEON)
#include <arm_neon.h>
#if defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC)
#define RIVE_BUF_NEON_F16 1
#endif
#endif
#if defined(__F16C__)
#include <immintrin.h>
#define RIVE_BUF_F16C 1
#elif defined(__SSE2__)
#include <emmintrin.h>
#endif
#define isoutofbounds(offset, len, accessize) \
(uint64_t(unsigned(offset)) + (accessize) > uint64_t(len))
// --------------------------------------------------------------------------
// Half-float (IEEE 754 half-precision) — scalar fallback
// --------------------------------------------------------------------------
static float halfToFloat(uint16_t h)
{
uint32_t sign = (uint32_t)(h >> 15) << 31;
uint32_t exp = (h >> 10) & 0x1F;
uint32_t mant = h & 0x3FF;
if (exp == 0)
{
if (mant == 0)
{
uint32_t bits = sign;
float f;
memcpy(&f, &bits, 4);
return f;
}
exp = 1;
while ((mant & 0x400) == 0)
{
mant <<= 1;
exp--;
}
mant &= 0x3FF;
uint32_t bits = sign | ((exp + 127 - 15) << 23) | (mant << 13);
float f;
memcpy(&f, &bits, 4);
return f;
}
if (exp == 31)
{
uint32_t bits = sign | 0x7F800000 | (mant << 13);
float f;
memcpy(&f, &bits, 4);
return f;
}
uint32_t bits = sign | ((exp + 127 - 15) << 23) | (mant << 13);
float f;
memcpy(&f, &bits, 4);
return f;
}
static uint16_t floatToHalf(float value)
{
uint32_t bits;
memcpy(&bits, &value, 4);
uint32_t sign = (bits >> 16) & 0x8000;
int32_t exp = ((bits >> 23) & 0xFF) - 127 + 15;
uint32_t mant = bits & 0x7FFFFF;
if (exp <= 0)
{
if (exp < -10)
return (uint16_t)sign;
mant |= 0x800000;
uint32_t shift = (uint32_t)(1 - exp);
uint32_t round = mant & ((1u << (shift + 13)) - 1);
mant >>= (shift + 13);
if (round > (1u << (shift + 12)) ||
(round == (1u << (shift + 12)) && (mant & 1)))
mant++;
return (uint16_t)(sign | mant);
}
if (exp == 0xFF - 127 + 15)
{
if (mant == 0)
return (uint16_t)(sign | 0x7C00);
return (uint16_t)(sign | 0x7C00 | (mant >> 13));
}
uint32_t round = mant & 0x1FFF;
mant >>= 13;
if (round > 0x1000 || (round == 0x1000 && (mant & 1)))
{
mant++;
if (mant >= 0x400)
{
mant = 0;
exp++;
}
}
if (exp >= 31)
return (uint16_t)(sign | 0x7C00);
return (uint16_t)(sign | ((uint32_t)exp << 10) | mant);
}
// --------------------------------------------------------------------------
// SIMD bulk f32↔f16 conversion
// --------------------------------------------------------------------------
static void convertF32ToF16Bulk(const float* src, uint16_t* dst, int count)
{
int i = 0;
#if RIVE_BUF_NEON_F16
for (; i + 3 < count; i += 4)
{
float32x4_t v = vld1q_f32(src + i);
float16x4_t h = vcvt_f16_f32(v);
vst1_u16(dst + i, vreinterpret_u16_f16(h));
}
#elif RIVE_BUF_F16C
for (; i + 3 < count; i += 4)
{
__m128 v = _mm_loadu_ps(src + i);
__m128i h = _mm_cvtps_ph(v, _MM_FROUND_TO_NEAREST_INT);
_mm_storel_epi64((__m128i*)(dst + i), h);
}
#endif
for (; i < count; i++)
dst[i] = floatToHalf(src[i]);
}
static void convertF16ToF32Bulk(const uint16_t* src, float* dst, int count)
{
int i = 0;
#if RIVE_BUF_NEON_F16
for (; i + 3 < count; i += 4)
{
uint16x4_t h = vld1_u16(src + i);
float32x4_t v = vcvt_f32_f16(vreinterpret_f16_u16(h));
vst1q_f32(dst + i, v);
}
#elif RIVE_BUF_F16C
for (; i + 3 < count; i += 4)
{
__m128i h = _mm_loadl_epi64((const __m128i*)(src + i));
__m128 v = _mm_cvtph_ps(h);
_mm_storeu_ps(dst + i, v);
}
#endif
for (; i < count; i++)
dst[i] = halfToFloat(src[i]);
}
// --------------------------------------------------------------------------
// buffer.readf16 / buffer.writef16
// --------------------------------------------------------------------------
static int buffer_readf16(lua_State* L)
{
size_t len = 0;
void* buf = luaL_checkbuffer(L, 1, &len);
int offset = luaL_checkinteger(L, 2);
if (isoutofbounds(offset, len, 2))
luaL_error(L, "buffer access out of bounds");
uint16_t half;
memcpy(&half, (char*)buf + offset, 2);
lua_pushnumber(L, (double)halfToFloat(half));
return 1;
}
static int buffer_writef16(lua_State* L)
{
size_t len = 0;
void* buf = luaL_checkbuffer(L, 1, &len);
int offset = luaL_checkinteger(L, 2);
double value = luaL_checknumber(L, 3);
if (isoutofbounds(offset, len, 2))
luaL_error(L, "buffer access out of bounds");
uint16_t half = floatToHalf((float)value);
memcpy((char*)buf + offset, &half, 2);
return 0;
}
// --------------------------------------------------------------------------
// buffer.stridedcopy
// --------------------------------------------------------------------------
static int buffer_stridedcopy(lua_State* L)
{
size_t dstLen = 0, srcLen = 0;
void* dst = luaL_checkbuffer(L, 1, &dstLen);
int dstOffset = luaL_checkinteger(L, 2);
int dstStride = luaL_checkinteger(L, 3);
void* src = luaL_checkbuffer(L, 4, &srcLen);
int srcOffset = luaL_checkinteger(L, 5);
int srcStride = luaL_checkinteger(L, 6);
int elementSize = luaL_checkinteger(L, 7);
int count = luaL_checkinteger(L, 8);
if (elementSize < 0)
luaL_error(L, "elementSize must be non-negative");
if (count < 0)
luaL_error(L, "count must be non-negative");
if (count == 0)
return 0;
if (srcStride < elementSize || dstStride < elementSize)
luaL_error(L, "stride must be >= elementSize");
int64_t srcEnd =
(int64_t)srcOffset + (int64_t)(count - 1) * srcStride + elementSize;
if (srcOffset < 0 || srcEnd > (int64_t)srcLen)
luaL_error(L, "buffer access out of bounds");
int64_t dstEnd =
(int64_t)dstOffset + (int64_t)(count - 1) * dstStride + elementSize;
if (dstOffset < 0 || dstEnd > (int64_t)dstLen)
luaL_error(L, "buffer access out of bounds");
char* d = (char*)dst + dstOffset;
const char* s = (const char*)src + srcOffset;
for (int i = 0; i < count; i++)
{
memcpy(d + (int64_t)i * dstStride,
s + (int64_t)i * srcStride,
(size_t)elementSize);
}
return 0;
}
// --------------------------------------------------------------------------
// buffer.convert — bulk format conversion with optional stride + SIMD
// --------------------------------------------------------------------------
enum BufFormat
{
BufFormat_f16,
BufFormat_f32,
BufFormat_u8,
BufFormat_u8norm,
BufFormat_i8norm,
BufFormat_u16,
BufFormat_u16norm,
BufFormat_i16norm,
BufFormat_u32,
};
static BufFormat parseFormat(lua_State* L, int arg)
{
const char* s = luaL_checkstring(L, arg);
if (strcmp(s, "f16") == 0)
return BufFormat_f16;
if (strcmp(s, "f32") == 0)
return BufFormat_f32;
if (strcmp(s, "u8") == 0)
return BufFormat_u8;
if (strcmp(s, "u8norm") == 0)
return BufFormat_u8norm;
if (strcmp(s, "i8norm") == 0)
return BufFormat_i8norm;
if (strcmp(s, "u16") == 0)
return BufFormat_u16;
if (strcmp(s, "u16norm") == 0)
return BufFormat_u16norm;
if (strcmp(s, "i16norm") == 0)
return BufFormat_i16norm;
if (strcmp(s, "u32") == 0)
return BufFormat_u32;
luaL_error(L, "unknown buffer format '%s'", s);
return BufFormat_f32;
}
static int formatByteSize(BufFormat fmt)
{
switch (fmt)
{
case BufFormat_f16:
case BufFormat_u16:
case BufFormat_u16norm:
case BufFormat_i16norm:
return 2;
case BufFormat_f32:
case BufFormat_u32:
return 4;
case BufFormat_u8:
case BufFormat_u8norm:
case BufFormat_i8norm:
return 1;
}
return 0;
}
static double readElement(const char* ptr, BufFormat fmt)
{
switch (fmt)
{
case BufFormat_f16:
{
uint16_t h;
memcpy(&h, ptr, 2);
return (double)halfToFloat(h);
}
case BufFormat_f32:
{
float f;
memcpy(&f, ptr, 4);
return (double)f;
}
case BufFormat_u8:
return (double)(uint8_t)*ptr;
case BufFormat_u8norm:
return (double)(uint8_t)*ptr / 255.0;
case BufFormat_i8norm:
{
int8_t v;
memcpy(&v, ptr, 1);
double d = (double)v / 127.0;
return d < -1.0 ? -1.0 : d;
}
case BufFormat_u16:
{
uint16_t v;
memcpy(&v, ptr, 2);
return (double)v;
}
case BufFormat_u16norm:
{
uint16_t v;
memcpy(&v, ptr, 2);
return (double)v / 65535.0;
}
case BufFormat_i16norm:
{
int16_t v;
memcpy(&v, ptr, 2);
double d = (double)v / 32767.0;
return d < -1.0 ? -1.0 : d;
}
case BufFormat_u32:
{
uint32_t v;
memcpy(&v, ptr, 4);
return (double)v;
}
}
return 0.0;
}
static void writeElement(char* ptr, BufFormat fmt, double value)
{
switch (fmt)
{
case BufFormat_f16:
{
uint16_t h = floatToHalf((float)value);
memcpy(ptr, &h, 2);
break;
}
case BufFormat_f32:
{
float f = (float)value;
memcpy(ptr, &f, 4);
break;
}
case BufFormat_u8:
{
*ptr = (char)(uint8_t)(unsigned int)value;
break;
}
case BufFormat_u8norm:
{
double c = value < 0.0 ? 0.0 : (value > 1.0 ? 1.0 : value);
*ptr = (char)(uint8_t)(c * 255.0 + 0.5);
break;
}
case BufFormat_i8norm:
{
double c = value < -1.0 ? -1.0 : (value > 1.0 ? 1.0 : value);
int8_t v = (int8_t)(c * 127.0 + (c >= 0 ? 0.5 : -0.5));
memcpy(ptr, &v, 1);
break;
}
case BufFormat_u16:
{
uint16_t v = (uint16_t)(unsigned int)value;
memcpy(ptr, &v, 2);
break;
}
case BufFormat_u16norm:
{
double c = value < 0.0 ? 0.0 : (value > 1.0 ? 1.0 : value);
uint16_t v = (uint16_t)(c * 65535.0 + 0.5);
memcpy(ptr, &v, 2);
break;
}
case BufFormat_i16norm:
{
double c = value < -1.0 ? -1.0 : (value > 1.0 ? 1.0 : value);
int16_t v = (int16_t)(c * 32767.0 + (c >= 0 ? 0.5 : -0.5));
memcpy(ptr, &v, 2);
break;
}
case BufFormat_u32:
{
uint32_t v = (uint32_t)value;
memcpy(ptr, &v, 4);
break;
}
}
}
// buffer.convert(dst, dstOff, dstFmt, src, srcOff, srcFmt, count,
// components?, dstStride?, srcStride?)
static int buffer_convert(lua_State* L)
{
size_t dstLen = 0, srcLen = 0;
void* dst = luaL_checkbuffer(L, 1, &dstLen);
int dstOffset = luaL_checkinteger(L, 2);
BufFormat dstFmt = parseFormat(L, 3);
void* src = luaL_checkbuffer(L, 4, &srcLen);
int srcOffset = luaL_checkinteger(L, 5);
BufFormat srcFmt = parseFormat(L, 6);
int count = luaL_checkinteger(L, 7);
int components = luaL_optinteger(L, 8, 1);
int srcElemSize = formatByteSize(srcFmt);
int dstElemSize = formatByteSize(dstFmt);
int dstStride = luaL_optinteger(L, 9, components * dstElemSize);
int srcStride = luaL_optinteger(L, 10, components * srcElemSize);
if (count < 0)
luaL_error(L, "count must be non-negative");
if (components < 1)
luaL_error(L, "components must be at least 1");
if (count == 0)
return 0;
if (srcStride < 0 || dstStride < 0)
luaL_error(L, "stride must be non-negative");
int64_t srcSpan = (int64_t)components * srcElemSize;
int64_t dstSpan = (int64_t)components * dstElemSize;
if (srcStride > 0 && srcStride < srcSpan)
luaL_error(L, "srcStride must be >= components * element size");
if (dstStride > 0 && dstStride < dstSpan)
luaL_error(L, "dstStride must be >= components * element size");
// Bounds check: last element at offset + (count-1)*stride + span
{
int64_t srcEnd =
(int64_t)srcOffset + (int64_t)(count - 1) * srcStride + srcSpan;
if (srcOffset < 0 || srcEnd > (int64_t)srcLen)
luaL_error(L, "buffer access out of bounds");
int64_t dstEnd =
(int64_t)dstOffset + (int64_t)(count - 1) * dstStride + dstSpan;
if (dstOffset < 0 || dstEnd > (int64_t)dstLen)
luaL_error(L, "buffer access out of bounds");
}
const char* s = (const char*)src + srcOffset;
char* d = (char*)dst + dstOffset;
bool packed = (srcStride == (int)srcSpan && dstStride == (int)dstSpan);
int totalScalars = count * components;
// Fast path: same format, packed → memcpy
if (srcFmt == dstFmt && packed)
{
memcpy(d, s, (size_t)totalScalars * srcElemSize);
return 0;
}
// SIMD fast path: packed f32 → f16
if (srcFmt == BufFormat_f32 && dstFmt == BufFormat_f16 && packed)
{
convertF32ToF16Bulk((const float*)s, (uint16_t*)d, totalScalars);
return 0;
}
// SIMD fast path: packed f16 → f32
if (srcFmt == BufFormat_f16 && dstFmt == BufFormat_f32 && packed)
{
convertF16ToF32Bulk((const uint16_t*)s, (float*)d, totalScalars);
return 0;
}
// General path: per-element with stride support
for (int i = 0; i < count; i++)
{
const char* sp = s + (int64_t)i * srcStride;
char* dp = d + (int64_t)i * dstStride;
for (int c = 0; c < components; c++)
{
double val = readElement(sp + c * srcElemSize, srcFmt);
writeElement(dp + c * dstElemSize, dstFmt, val);
}
}
return 0;
}
// --------------------------------------------------------------------------
// Registration
// --------------------------------------------------------------------------
extern "C" int luaopen_rive_buffer_ext(lua_State* L)
{
lua_getglobal(L, "buffer");
lua_pushcfunction(L, buffer_readf16, "buffer.readf16");
lua_setfield(L, -2, "readf16");
lua_pushcfunction(L, buffer_writef16, "buffer.writef16");
lua_setfield(L, -2, "writef16");
lua_pushcfunction(L, buffer_stridedcopy, "buffer.stridedcopy");
lua_setfield(L, -2, "stridedcopy");
lua_pushcfunction(L, buffer_convert, "buffer.convert");
lua_setfield(L, -2, "convert");
lua_pop(L, 1);
return 0;
}
#endif // WITH_RIVE_SCRIPTING
You can’t perform that action at this time.