/
vsTTempSmooth_AVX512.cpp
334 lines (268 loc) · 15.9 KB
/
vsTTempSmooth_AVX512.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
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
#include "VCL2/vectorclass.h"
#include "vsTTempSmooth.h"
template <typename T>
AVS_FORCEINLINE static Vec16i load(const void* p)
{
if constexpr (std::is_same_v<T, uint8_t>)
return Vec16i().load_16uc(p);
else
return Vec16i().load_16us(p);
}
template <bool pfclip, bool fp>
template <typename T, bool useDiff>
void TTempSmooth<pfclip, fp>::filterI_avx512(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept
{
int src_stride[15]{};
int pf_stride[15]{};
const size_t stride{ dst->GetPitch(plane) / sizeof(T) };
const size_t width{ dst->GetRowSize(plane) / sizeof(T) };
const int height{ dst->GetHeight(plane) };
const T* srcp[15]{}, * pfp[15]{};
for (int i{ 0 }; i < _diameter; ++i)
{
src_stride[i] = src[i]->GetPitch(plane) / sizeof(T);
pf_stride[i] = pf[i]->GetPitch(plane) / sizeof(T);
srcp[i] = reinterpret_cast<const T*>(src[i]->GetReadPtr(plane));
pfp[i] = reinterpret_cast<const T*>(pf[i]->GetReadPtr(plane));
}
T* __restrict dstp{ reinterpret_cast<T*>(dst->GetWritePtr(plane)) };
const int l{ plane >> 1 };
const Vec16i thresh{ _thresh[l] << _shift };
for (int y{ 0 }; y < height; ++y)
{
for (int x{ 0 }; x < width; x += 16)
{
const auto& c{ load<T>(&pfp[_maxr][x]) };
const auto& srcp_v{ load<T>(&srcp[_maxr][x]) };
Vec16f weights{ _cw };
Vec16f sum{ to_float(srcp_v) * weights };
int frameIndex{ _maxr - 1 };
if (frameIndex > fromFrame)
{
auto t1{ load<T>(&pfp[frameIndex][x]) };
auto diff{ abs(c - t1) };
const auto check_v{ diff < thresh };
Vec16f weight;
for (int i{ 0 }; i < 16; ++i)
weight.insert(i, _weight[l][(useDiff) ? (diff.extract(i) >> _shift) : frameIndex]);
weights = select(Vec16fb(check_v), weights + weight, weights);
sum = select(Vec16fb(check_v), sum + to_float(load<T>(&srcp[frameIndex][x])) * weight, sum);
--frameIndex;
int v{ 256 };
while (frameIndex > fromFrame)
{
const auto& t2{ t1 };
t1 = load<T>(&pfp[frameIndex][x]);
diff = abs(c - t1);
const auto check_v1{ diff < thresh&& abs(t1 - t2) < thresh };
Vec16f weight;
for (int i{ 0 }; i < 16; ++i)
weight.insert(i, _weight[l][(useDiff) ? ((diff.extract(i) >> _shift) + v) : frameIndex]);
weights = select(Vec16fb(check_v1), weights + weight, weights);
sum = select(Vec16fb(check_v1), sum + to_float(load<T>(&srcp[frameIndex][x])) * weight, sum);
--frameIndex;
v += 256;
}
}
frameIndex = _maxr + 1;
if (frameIndex < toFrame)
{
auto t1{ load<T>(&pfp[frameIndex][x]) };
auto diff{ abs(c - t1) };
const auto check_v{ diff < thresh };
Vec16f weight;
for (int i{ 0 }; i < 16; ++i)
weight.insert(i, _weight[l][(useDiff) ? (diff.extract(i) >> _shift) : frameIndex]);
weights = select(Vec16fb(check_v), weights + weight, weights);
sum = select(Vec16fb(check_v), sum + to_float(load<T>(&srcp[frameIndex][x])) * weight, sum);
++frameIndex;
int v{ 256 };
while (frameIndex < toFrame)
{
const auto& t2{ t1 };
t1 = load<T>(&pfp[frameIndex][x]);
diff = abs(c - t1);
const auto check_v1{ diff < thresh&& abs(t1 - t2) < thresh };
Vec16f weight;
for (int i{ 0 }; i < 16; ++i)
weight.insert(i, _weight[l][(useDiff) ? ((diff.extract(i) >> _shift) + v) : frameIndex]);
weights = select(Vec16fb(check_v1), weights + weight, weights);
sum = select(Vec16fb(check_v1), sum + to_float(load<T>(&srcp[frameIndex][x])) * weight, sum);
++frameIndex;
v += 256;
}
}
if constexpr (std::is_same_v<T, uint8_t>)
{
if constexpr (fp)
compress_saturated_s2u(compress_saturated(truncatei(to_float(load<T>(&srcp[_maxr][x])) * (1.0f - weights) + sum + 0.5f), zero_si512()), zero_si512()).get_low().get_low().store(dstp + x);
else
compress_saturated_s2u(compress_saturated(truncatei(sum / weights + 0.5f), zero_si512()), zero_si512()).get_low().get_low().store(dstp + x);
}
else
{
if constexpr (fp)
compress_saturated_s2u(truncatei(to_float(load<T>(&srcp[_maxr][x])) * (1.0f - weights) + sum + 0.5f), zero_si512()).get_low().store(dstp + x);
else
compress_saturated_s2u(truncatei(sum / weights + 0.5f), zero_si512()).get_low().store(dstp + x);
}
}
for (int i{ 0 }; i < _diameter; ++i)
{
srcp[i] += src_stride[i];
pfp[i] += pf_stride[i];
}
dstp += stride;
}
}
template void TTempSmooth<true, true>::filterI_avx512<uint8_t, true>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<true, false>::filterI_avx512<uint8_t, true>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<true, true>::filterI_avx512<uint8_t, false>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<true, false>::filterI_avx512<uint8_t, false>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<false, true>::filterI_avx512<uint8_t, true>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<false, false>::filterI_avx512<uint8_t, true>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<false, true>::filterI_avx512<uint8_t, false>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<false, false>::filterI_avx512<uint8_t, false>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<true, true>::filterI_avx512<uint16_t, true>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<true, false>::filterI_avx512<uint16_t, true>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<true, true>::filterI_avx512<uint16_t, false>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<true, false>::filterI_avx512<uint16_t, false>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<false, true>::filterI_avx512<uint16_t, true>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<false, false>::filterI_avx512<uint16_t, true>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<false, true>::filterI_avx512<uint16_t, false>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<false, false>::filterI_avx512<uint16_t, false>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template <bool pfclip, bool fp>
template <bool useDiff>
void TTempSmooth<pfclip, fp>::filterF_avx512(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept
{
int src_stride[15]{};
int pf_stride[15]{};
const int stride{ dst->GetPitch(plane) / 4 };
const int width{ dst->GetRowSize(plane) / 4 };
const int height{ dst->GetHeight(plane) };
const float* srcp[15]{}, * pfp[15]{};
for (int i{ 0 }; i < _diameter; ++i)
{
src_stride[i] = src[i]->GetPitch(plane) / 4;
pf_stride[i] = pf[i]->GetPitch(plane) / 4;
srcp[i] = reinterpret_cast<const float*>(src[i]->GetReadPtr(plane));
pfp[i] = reinterpret_cast<const float*>(pf[i]->GetReadPtr(plane));
}
float* __restrict dstp{ reinterpret_cast<float*>(dst->GetWritePtr(plane)) };
const int l{ plane >> 1 };
const Vec16f thresh{ _threshF[l] };
for (int y{ 0 }; y < height; ++y)
{
for (int x{ 0 }; x < width; x += 16)
{
const auto& c{ Vec16f().load(&pfp[_maxr][x]) };
const auto& srcp_v{ Vec16f().load(&srcp[_maxr][x]) };
Vec16f weights{ _cw };
Vec16f sum{ srcp_v * weights };
int frameIndex{ _maxr - 1 };
if (frameIndex > fromFrame)
{
auto t1{ Vec16f().load(&pfp[frameIndex][x]) };
auto diff{ min(abs(c - t1), 1.0f) };
const auto check_v{ diff < thresh };
Vec16f weight;
for (int i{ 0 }; i < 16; ++i)
weight.insert(i, _weight[l][(useDiff) ? static_cast<int>(diff.extract(i) * 255.0f) : frameIndex]);
weights = select(check_v, weights + weight, weights);
sum = select(check_v, sum + Vec16f().load(&srcp[frameIndex][x]) * weight, sum);
--frameIndex;
int v{ 256 };
while (frameIndex > fromFrame)
{
const auto& t2{ t1 };
t1 = Vec16f().load(&pfp[frameIndex][x]);
diff = min(abs(c - t1), 1.0f);
const auto check_v1{ diff < thresh&& min(abs(t1 - t2), 1.0f) < thresh };
Vec16f weight;
for (int i{ 0 }; i < 16; ++i)
weight.insert(i, _weight[l][(useDiff) ? (static_cast<int>(diff.extract(i) * 255.0f) + v) : frameIndex]);
weights = select(Vec16fb(check_v1), weights + weight, weights);
sum = select(Vec16fb(check_v1), sum + Vec16f().load(&srcp[frameIndex][x]) * weight, sum);
--frameIndex;
v += 256;
}
}
frameIndex = _maxr + 1;
if (frameIndex < toFrame)
{
auto t1{ Vec16f().load(&pfp[frameIndex][x]) };
auto diff{ min(abs(c - t1), 1.0f) };
const auto check_v{ diff < thresh };
Vec16f weight;
for (int i{ 0 }; i < 16; ++i)
weight.insert(i, _weight[l][(useDiff) ? static_cast<int>(diff.extract(i) * 255.0f) : frameIndex]);
weights = select(check_v, weights + weight, weights);
sum = select(check_v, sum + Vec16f().load(&srcp[frameIndex][x]) * weight, sum);
++frameIndex;
int v{ 256 };
while (frameIndex < toFrame)
{
const auto& t2{ t1 };
t1 = Vec16f().load(&pfp[frameIndex][x]);
diff = min(abs(c - t1), 1.0f);
const auto check_v1{ diff < thresh&& min(abs(t1 - t2), 1.0f) < thresh };
Vec16f weight;
for (int i{ 0 }; i < 16; ++i)
weight.insert(i, _weight[l][(useDiff) ? (static_cast<int>(diff.extract(i) * 255.0f) + v) : frameIndex]);
weights = select(Vec16fb(check_v1), weights + weight, weights);
sum = select(Vec16fb(check_v1), sum + Vec16f().load(&srcp[frameIndex][x]) * weight, sum);
++frameIndex;
v += 256;
}
}
if constexpr (fp)
(Vec16f().load(&srcp[_maxr][x]) * (1.0f - weights) + sum).store(dstp + x);
else
(sum / weights).store(dstp + x);
}
for (int i{ 0 }; i < _diameter; ++i)
{
srcp[i] += src_stride[i];
pfp[i] += pf_stride[i];
}
dstp += stride;
}
}
template void TTempSmooth<true, true>::filterF_avx512<true>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<true, false>::filterF_avx512<true>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<true, true>::filterF_avx512<false>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<true, false>::filterF_avx512<false>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<false, true>::filterF_avx512<true>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<false, false>::filterF_avx512<true>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<false, true>::filterF_avx512<false>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template void TTempSmooth<false, false>::filterF_avx512<false>(PVideoFrame src[15], PVideoFrame pf[15], PVideoFrame& dst, const int fromFrame, const int toFrame, const int plane) noexcept;
template <typename T>
float ComparePlane_avx512(PVideoFrame& src, PVideoFrame& src1, const int bits_per_pixel) noexcept
{
const size_t pitch{ src->GetPitch(PLANAR_Y) / sizeof(T) };
const size_t pitch2{ src1->GetPitch(PLANAR_Y) / sizeof(T) };
const size_t width{ src->GetRowSize(PLANAR_Y) / sizeof(T) };
const int height{ src->GetHeight(PLANAR_Y) };
const T* srcp{ reinterpret_cast<const T*>(src->GetReadPtr(PLANAR_Y)) };
const T* srcp2{ reinterpret_cast<const T*>(src1->GetReadPtr(PLANAR_Y)) };
Vec16f accum{ 0.0f };
for (size_t y{ 0 }; y < height; ++y)
{
for (size_t x{ 0 }; x < width; x += 16)
{
if constexpr (std::is_integral_v<T>)
accum += abs(to_float(load<T>(&srcp[x])) - to_float(load<T>(&srcp2[x])));
else
accum += abs(Vec16f().load(&srcp[x]) - Vec16f().load(&srcp2[x]));
}
srcp += pitch;
srcp2 += pitch2;
}
if constexpr (std::is_integral_v<T>)
return horizontal_add(accum / ((1 << bits_per_pixel) - 1)) / (height * width);
else
return horizontal_add(accum) / (height * width);
}
template float ComparePlane_avx512<uint8_t>(PVideoFrame& src, PVideoFrame& src1, const int bits_per_pixel) noexcept;
template float ComparePlane_avx512<uint16_t>(PVideoFrame& src, PVideoFrame& src1, const int bits_per_pixel) noexcept;
template float ComparePlane_avx512<float>(PVideoFrame& src, PVideoFrame& src1, const int bits_per_pixel) noexcept;