Add Motion Compehension ARM64 Neon Code

dongzhang committed 12 years ago

Commit ad9e2da1 parent b47606a8 files · +5475 −480Browse files

Showing the changes to codec/decoder/core/src/mc.cpp (1 of 8 files). Show all

Unified · Split

@@ -85,13 +85,13 @@ static const uint8_t g_kuiABCD[8][8][4] = { //g_kA[dy][dx], g_kB[dy][dx], g_kC[d · 84 unmodified lines
85	85	};
86	86
87	87	typedef void (PWelsMcWidthHeightFunc) (const uint8_t pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
88		− int32_t iWidth, int32_t iHeight);
	88	+ int32_t iWidth, int32_t iHeight);
89	89
90	90	//***************************************************************************//
91	91	// C code implementation //
92	92	//***************************************************************************//
93	93	static inline void McCopyWidthEq2_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
94		− int32_t iHeight) {
	94	+ int32_t iHeight) {
95	95	int32_t i;
96	96	for (i = 0; i < iHeight; i++) { // iWidth == 2 only for chroma
97	97	ST16A2 (pDst, LD16 (pSrc));
@@ -101,7 +101,7 @@ static inline void McCopyWidthEq2_c (const uint8_t* pSrc, int32_t iSrcStride, ui · 3 unmodified lines
101	101	}
102	102
103	103	static inline void McCopyWidthEq4_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
104		− int32_t iHeight) {
	104	+ int32_t iHeight) {
105	105	int32_t i;
106	106	for (i = 0; i < iHeight; i++) {
107	107	ST32A4 (pDst, LD32 (pSrc));
@@ -111,7 +111,7 @@ static inline void McCopyWidthEq4_c (const uint8_t* pSrc, int32_t iSrcStride, ui · 3 unmodified lines
111	111	}
112	112
113	113	static inline void McCopyWidthEq8_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
114		− int32_t iHeight) {
	114	+ int32_t iHeight) {
115	115	int32_t i;
116	116	for (i = 0; i < iHeight; i++) {
117	117	ST64A8 (pDst, LD64 (pSrc));
@@ -121,7 +121,7 @@ static inline void McCopyWidthEq8_c (const uint8_t* pSrc, int32_t iSrcStride, ui · 3 unmodified lines
121	121	}
122	122
123	123	static inline void McCopyWidthEq16_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
124		− int32_t iHeight) {
	124	+ int32_t iHeight) {
125	125	int32_t i;
126	126	for (i = 0; i < iHeight; i++) {
127	127	ST64A8 (pDst , LD64 (pSrc));
@@ -138,7 +138,7 @@ static inline int32_t HorFilterInput16bit_c (int16_t* pSrc) { · 10 unmodified lines
138	138	int32_t iPix14 = pSrc[-1] + pSrc[2];
139	139	int32_t iPix23 = pSrc[ 0] + pSrc[1];
140	140
141		− return (iPix05 - (iPix14 * 5)+ (iPix23 * 20));
	141	+ return (iPix05 - (iPix14 * 5) + (iPix23 * 20));
142	142	}
143	143	// h: iOffset=1 / v: iOffset=iSrcStride
144	144	static inline int32_t FilterInput8bitWithStride_c (const uint8_t* pSrc, const int32_t kiOffset) {
@@ -153,7 +153,7 @@ static inline int32_t FilterInput8bitWithStride_c (const uint8_t* pSrc, const in · 8 unmodified lines
153	153	}
154	154
155	155	static inline void PixelAvg_c (uint8_t* pDst, int32_t iDstStride, const uint8_t* pSrcA, int32_t iSrcAStride,
156		− const uint8_t* pSrcB, int32_t iSrcBStride, int32_t iWidth, int32_t iHeight) {
	156	+ const uint8_t* pSrcB, int32_t iSrcBStride, int32_t iWidth, int32_t iHeight) {
157	157	int32_t i, j;
158	158	for (i = 0; i < iHeight; i++) {
159	159	for (j = 0; j < iWidth; j++) {
@@ -165,7 +165,7 @@ static inline void PixelAvg_c (uint8_t* pDst, int32_t iDstStride, const uint8_t* · 5 unmodified lines
165	165	}
166	166	}
167	167	static inline void McCopy_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
168		− int32_t iHeight) {
	168	+ int32_t iHeight) {
169	169	if (iWidth == 16)
170	170	McCopyWidthEq16_c (pSrc, iSrcStride, pDst, iDstStride, iHeight);
171	171	else if (iWidth == 8)
@@ -176,8 +176,9 @@ static inline void McCopy_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* p · 4 unmodified lines
176	176	McCopyWidthEq2_c (pSrc, iSrcStride, pDst, iDstStride, iHeight);
177	177	}
178	178
179		−static inline void McHorVer20_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
180		− int32_t iHeight) {
	179	+static inline void McHorVer20_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	180	+ int32_t iWidth,
	181	+ int32_t iHeight) {
181	182	int32_t i, j;
182	183	for (i = 0; i < iHeight; i++) {
183	184	for (j = 0; j < iWidth; j++) {
@@ -188,8 +189,9 @@ static inline void McHorVer20_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_ · 4 unmodified lines
188	189	}
189	190	}
190	191
191		−static inline void McHorVer02_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
192		− int32_t iHeight) {
	192	+static inline void McHorVer02_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	193	+ int32_t iWidth,
	194	+ int32_t iHeight) {
193	195	int32_t i, j;
194	196	for (i = 0; i < iHeight; i++) {
195	197	for (j = 0; j < iWidth; j++) {
@@ -200,8 +202,9 @@ static inline void McHorVer02_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_ · 4 unmodified lines
200	202	}
201	203	}
202	204
203		−static inline void McHorVer22_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
204		− int32_t iHeight) {
	205	+static inline void McHorVer22_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	206	+ int32_t iWidth,
	207	+ int32_t iHeight) {
205	208	int16_t iTmp[16 + 5]; //16
206	209	int32_t i, j, k;
207	210
@@ -218,88 +221,100 @@ static inline void McHorVer22_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_ · 10 unmodified lines
218	221	}
219	222
220	223	/////////////////////luma MC//////////////////////////
221		−static inline void McHorVer01_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
222		− int32_t iHeight) {
	224	+static inline void McHorVer01_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	225	+ int32_t iWidth,
	226	+ int32_t iHeight) {
223	227	uint8_t uiTmp[256];
224	228	McHorVer02_c (pSrc, iSrcStride, uiTmp, 16, iWidth, iHeight);
225	229	PixelAvg_c (pDst, iDstStride, pSrc, iSrcStride, uiTmp, 16, iWidth, iHeight);
226	230	}
227		−static inline void McHorVer03_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
228		− int32_t iHeight) {
	231	+static inline void McHorVer03_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	232	+ int32_t iWidth,
	233	+ int32_t iHeight) {
229	234	uint8_t uiTmp[256];
230	235	McHorVer02_c (pSrc, iSrcStride, uiTmp, 16, iWidth, iHeight);
231	236	PixelAvg_c (pDst, iDstStride, pSrc + iSrcStride, iSrcStride, uiTmp, 16, iWidth, iHeight);
232	237	}
233		−static inline void McHorVer10_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
234		− int32_t iHeight) {
	238	+static inline void McHorVer10_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	239	+ int32_t iWidth,
	240	+ int32_t iHeight) {
235	241	uint8_t uiTmp[256];
236	242	McHorVer20_c (pSrc, iSrcStride, uiTmp, 16, iWidth, iHeight);
237	243	PixelAvg_c (pDst, iDstStride, pSrc, iSrcStride, uiTmp, 16, iWidth, iHeight);
238	244	}
239		−static inline void McHorVer11_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
240		− int32_t iHeight) {
	245	+static inline void McHorVer11_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	246	+ int32_t iWidth,
	247	+ int32_t iHeight) {
241	248	uint8_t uiHorTmp[256];
242	249	uint8_t uiVerTmp[256];
243	250	McHorVer20_c (pSrc, iSrcStride, uiHorTmp, 16, iWidth, iHeight);
244	251	McHorVer02_c (pSrc, iSrcStride, uiVerTmp, 16, iWidth, iHeight);
245	252	PixelAvg_c (pDst, iDstStride, uiHorTmp, 16, uiVerTmp, 16, iWidth, iHeight);
246	253	}
247		−static inline void McHorVer12_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
248		− int32_t iHeight) {
	254	+static inline void McHorVer12_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	255	+ int32_t iWidth,
	256	+ int32_t iHeight) {
249	257	uint8_t uiVerTmp[256];
250	258	uint8_t uiCtrTmp[256];
251	259	McHorVer02_c (pSrc, iSrcStride, uiVerTmp, 16, iWidth, iHeight);
252	260	McHorVer22_c (pSrc, iSrcStride, uiCtrTmp, 16, iWidth, iHeight);
253	261	PixelAvg_c (pDst, iDstStride, uiVerTmp, 16, uiCtrTmp, 16, iWidth, iHeight);
254	262	}
255		−static inline void McHorVer13_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
256		− int32_t iHeight) {
	263	+static inline void McHorVer13_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	264	+ int32_t iWidth,
	265	+ int32_t iHeight) {
257	266	uint8_t uiHorTmp[256];
258	267	uint8_t uiVerTmp[256];
259	268	McHorVer20_c (pSrc + iSrcStride, iSrcStride, uiHorTmp, 16, iWidth, iHeight);
260	269	McHorVer02_c (pSrc, iSrcStride, uiVerTmp, 16, iWidth, iHeight);
261	270	PixelAvg_c (pDst, iDstStride, uiHorTmp, 16, uiVerTmp, 16, iWidth, iHeight);
262	271	}
263		−static inline void McHorVer21_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
264		− int32_t iHeight) {
	272	+static inline void McHorVer21_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	273	+ int32_t iWidth,
	274	+ int32_t iHeight) {
265	275	uint8_t uiHorTmp[256];
266	276	uint8_t uiCtrTmp[256];
267	277	McHorVer20_c (pSrc, iSrcStride, uiHorTmp, 16, iWidth, iHeight);
268	278	McHorVer22_c (pSrc, iSrcStride, uiCtrTmp, 16, iWidth, iHeight);
269	279	PixelAvg_c (pDst, iDstStride, uiHorTmp, 16, uiCtrTmp, 16, iWidth, iHeight);
270	280	}
271		−static inline void McHorVer23_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
272		− int32_t iHeight) {
	281	+static inline void McHorVer23_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	282	+ int32_t iWidth,
	283	+ int32_t iHeight) {
273	284	uint8_t uiHorTmp[256];
274	285	uint8_t uiCtrTmp[256];
275	286	McHorVer20_c (pSrc + iSrcStride, iSrcStride, uiHorTmp, 16, iWidth, iHeight);
276	287	McHorVer22_c (pSrc, iSrcStride, uiCtrTmp, 16, iWidth, iHeight);
277	288	PixelAvg_c (pDst, iDstStride, uiHorTmp, 16, uiCtrTmp, 16, iWidth, iHeight);
278	289	}
279		−static inline void McHorVer30_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
280		− int32_t iHeight) {
	290	+static inline void McHorVer30_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	291	+ int32_t iWidth,
	292	+ int32_t iHeight) {
281	293	uint8_t uiHorTmp[256];
282	294	McHorVer20_c (pSrc, iSrcStride, uiHorTmp, 16, iWidth, iHeight);
283	295	PixelAvg_c (pDst, iDstStride, pSrc + 1, iSrcStride, uiHorTmp, 16, iWidth, iHeight);
284	296	}
285		−static inline void McHorVer31_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
286		− int32_t iHeight) {
	297	+static inline void McHorVer31_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	298	+ int32_t iWidth,
	299	+ int32_t iHeight) {
287	300	uint8_t uiHorTmp[256];
288	301	uint8_t uiVerTmp[256];
289	302	McHorVer20_c (pSrc, iSrcStride, uiHorTmp, 16, iWidth, iHeight);
290	303	McHorVer02_c (pSrc + 1, iSrcStride, uiVerTmp, 16, iWidth, iHeight);
291	304	PixelAvg_c (pDst, iDstStride, uiHorTmp, 16, uiVerTmp, 16, iWidth, iHeight);
292	305	}
293		−static inline void McHorVer32_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
294		− int32_t iHeight) {
	306	+static inline void McHorVer32_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	307	+ int32_t iWidth,
	308	+ int32_t iHeight) {
295	309	uint8_t uiVerTmp[256];
296	310	uint8_t uiCtrTmp[256];
297	311	McHorVer02_c (pSrc + 1, iSrcStride, uiVerTmp, 16, iWidth, iHeight);
298	312	McHorVer22_c (pSrc, iSrcStride, uiCtrTmp, 16, iWidth, iHeight);
299	313	PixelAvg_c (pDst, iDstStride, uiVerTmp, 16, uiCtrTmp, 16, iWidth, iHeight);
300	314	}
301		−static inline void McHorVer33_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
302		− int32_t iHeight) {
	315	+static inline void McHorVer33_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	316	+ int32_t iWidth,
	317	+ int32_t iHeight) {
303	318	uint8_t uiHorTmp[256];
304	319	uint8_t uiVerTmp[256];
305	320	McHorVer20_c (pSrc + iSrcStride, iSrcStride, uiHorTmp, 16, iWidth, iHeight);
@@ -308,7 +323,7 @@ static inline void McHorVer33_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_ · 2 unmodified lines
308	323	}
309	324
310	325	void McLuma_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
311		− int16_t iMvX, int16_t iMvY, int32_t iWidth, int32_t iHeight)
	326	+ int16_t iMvX, int16_t iMvY, int32_t iWidth, int32_t iHeight)
312	327	//pSrc has been added the offset of mv
313	328	{
314	329	static const PWelsMcWidthHeightFunc pWelsMcFunc[4][4] = { //[x][y]
@@ -326,7 +341,7 @@ static inline void McChromaWithFragMv_c (const uint8_t* pSrc, int32_t iSrcStride · 11 unmodified lines
326	341	int32_t i, j;
327	342	int32_t iA, iB, iC, iD;
328	343	const uint8_t* pSrcNext = pSrc + iSrcStride;
329		− const uint8_t *pABCD = g_kuiABCD[iMvY & 0x07][iMvX & 0x07];
	344	+ const uint8_t* pABCD = g_kuiABCD[iMvY & 0x07][iMvX & 0x07];
330	345	iA = pABCD[0];
331	346	iB = pABCD[1];
332	347	iC = pABCD[2];
@@ -342,7 +357,7 @@ static inline void McChromaWithFragMv_c (const uint8_t* pSrc, int32_t iSrcStride · 9 unmodified lines
342	357	}
343	358
344	359	void McChroma_c (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
345		− int16_t iMvX, int16_t iMvY, int32_t iWidth, int32_t iHeight)
	360	+ int16_t iMvX, int16_t iMvY, int32_t iWidth, int32_t iHeight)
346	361	//pSrc has been added the offset of mv
347	362	{
348	363	const int32_t kiD8x = iMvX & 0x07;
@@ -376,8 +391,9 @@ static inline void McHorVer22WidthEq16_sse2 (const uint8_t* pSrc, int32_t iSrcSt · 27 unmodified lines
376	391	McHorVer22WidthEq8_sse2 (&pSrc[8], iSrcStride, &pDst[8], iDstStride, iHeight);
377	392	}
378	393
379		−static inline void McCopy_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride, int32_t iWidth,
380		− int32_t iHeight) {
	394	+static inline void McCopy_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	395	+ int32_t iWidth,
	396	+ int32_t iHeight) {
381	397	if (iWidth == 16)
382	398	McCopyWidthEq16_sse2 (pSrc, iSrcStride, pDst, iDstStride, iHeight);
383	399	else if (iWidth == 8)
@@ -389,7 +405,7 @@ static inline void McCopy_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t · 5 unmodified lines
389	405	}
390	406
391	407	static inline void McHorVer20_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
392		− int32_t iWidth, int32_t iHeight) {
	408	+ int32_t iWidth, int32_t iHeight) {
393	409	if (iWidth == 16)
394	410	McHorVer20WidthEq16_sse2 (pSrc, iSrcStride, pDst, iDstStride, iHeight);
395	411	else if (iWidth == 8)
@@ -399,7 +415,7 @@ static inline void McHorVer20_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uin · 3 unmodified lines
399	415	}
400	416
401	417	static inline void McHorVer02_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
402		− int32_t iWidth, int32_t iHeight) {
	418	+ int32_t iWidth, int32_t iHeight) {
403	419	if (iWidth == 16)
404	420	McHorVer02WidthEq16_sse2 (pSrc, iSrcStride, pDst, iDstStride, iHeight);
405	421	else if (iWidth == 8)
@@ -409,7 +425,7 @@ static inline void McHorVer02_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uin · 3 unmodified lines
409	425	}
410	426
411	427	static inline void McHorVer22_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
412		− int32_t iWidth, int32_t iHeight) {
	428	+ int32_t iWidth, int32_t iHeight) {
413	429	if (iWidth == 16)
414	430	McHorVer22WidthEq16_sse2 (pSrc, iSrcStride, pDst, iDstStride, iHeight);
415	431	else if (iWidth == 8)
@@ -419,7 +435,7 @@ static inline void McHorVer22_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uin · 3 unmodified lines
419	435	}
420	436
421	437	static inline void McHorVer01_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
422		− int32_t iWidth, int32_t iHeight) {
	438	+ int32_t iWidth, int32_t iHeight) {
423	439	ENFORCE_STACK_ALIGN_1D (uint8_t, pTmp, 256, 16);
424	440	if (iWidth == 16) {
425	441	McHorVer02WidthEq16_sse2 (pSrc, iSrcStride, pTmp, 16, iHeight);
@@ -433,7 +449,7 @@ static inline void McHorVer01_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uin · 7 unmodified lines
433	449	}
434	450	}
435	451	static inline void McHorVer03_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
436		− int32_t iWidth, int32_t iHeight) {
	452	+ int32_t iWidth, int32_t iHeight) {
437	453	ENFORCE_STACK_ALIGN_1D (uint8_t, pTmp, 256, 16);
438	454	if (iWidth == 16) {
439	455	McHorVer02WidthEq16_sse2 (pSrc, iSrcStride, pTmp, 16, iHeight);
@@ -447,7 +463,7 @@ static inline void McHorVer03_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uin · 7 unmodified lines
447	463	}
448	464	}
449	465	static inline void McHorVer10_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
450		− int32_t iWidth, int32_t iHeight) {
	466	+ int32_t iWidth, int32_t iHeight) {
451	467	ENFORCE_STACK_ALIGN_1D (uint8_t, pTmp, 256, 16);
452	468	if (iWidth == 16) {
453	469	McHorVer20WidthEq16_sse2 (pSrc, iSrcStride, pTmp, 16, iHeight);
@@ -461,7 +477,7 @@ static inline void McHorVer10_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uin · 7 unmodified lines
461	477	}
462	478	}
463	479	static inline void McHorVer11_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
464		− int32_t iWidth, int32_t iHeight) {
	480	+ int32_t iWidth, int32_t iHeight) {
465	481	ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
466	482	ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
467	483	if (iWidth == 16) {
@@ -479,7 +495,7 @@ static inline void McHorVer11_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uin · 11 unmodified lines
479	495	}
480	496	}
481	497	static inline void McHorVer12_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
482		− int32_t iWidth, int32_t iHeight) {
	498	+ int32_t iWidth, int32_t iHeight) {
483	499	ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
484	500	ENFORCE_STACK_ALIGN_1D (uint8_t, pCtrTmp, 256, 16);
485	501	if (iWidth == 16) {
@@ -497,7 +513,7 @@ static inline void McHorVer12_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uin · 11 unmodified lines
497	513	}
498	514	}
499	515	static inline void McHorVer13_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
500		− int32_t iWidth, int32_t iHeight) {
	516	+ int32_t iWidth, int32_t iHeight) {
501	517	ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
502	518	ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
503	519	if (iWidth == 16) {
@@ -515,7 +531,7 @@ static inline void McHorVer13_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uin · 11 unmodified lines
515	531	}
516	532	}
517	533	static inline void McHorVer21_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
518		− int32_t iWidth, int32_t iHeight) {
	534	+ int32_t iWidth, int32_t iHeight) {
519	535	ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
520	536	ENFORCE_STACK_ALIGN_1D (uint8_t, pCtrTmp, 256, 16);
521	537	if (iWidth == 16) {
@@ -533,7 +549,7 @@ static inline void McHorVer21_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uin · 11 unmodified lines
533	549	}
534	550	}
535	551	static inline void McHorVer23_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
536		− int32_t iWidth, int32_t iHeight) {
	552	+ int32_t iWidth, int32_t iHeight) {
537	553	ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
538	554	ENFORCE_STACK_ALIGN_1D (uint8_t, pCtrTmp, 256, 16);
539	555	if (iWidth == 16) {
@@ -551,7 +567,7 @@ static inline void McHorVer23_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uin · 11 unmodified lines
551	567	}
552	568	}
553	569	static inline void McHorVer30_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
554		− int32_t iWidth, int32_t iHeight) {
	570	+ int32_t iWidth, int32_t iHeight) {
555	571	ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
556	572	if (iWidth == 16) {
557	573	McHorVer20WidthEq16_sse2 (pSrc, iSrcStride, pHorTmp, 16, iHeight);
@@ -565,7 +581,7 @@ static inline void McHorVer30_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uin · 7 unmodified lines
565	581	}
566	582	}
567	583	static inline void McHorVer31_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
568		− int32_t iWidth, int32_t iHeight) {
	584	+ int32_t iWidth, int32_t iHeight) {
569	585	ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
570	586	ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
571	587	if (iWidth == 16) {
@@ -583,7 +599,7 @@ static inline void McHorVer31_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uin · 11 unmodified lines
583	599	}
584	600	}
585	601	static inline void McHorVer32_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
586		− int32_t iWidth, int32_t iHeight) {
	602	+ int32_t iWidth, int32_t iHeight) {
587	603	ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
588	604	ENFORCE_STACK_ALIGN_1D (uint8_t, pCtrTmp, 256, 16);
589	605	if (iWidth == 16) {
@@ -601,7 +617,7 @@ static inline void McHorVer32_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uin · 11 unmodified lines
601	617	}
602	618	}
603	619	static inline void McHorVer33_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
604		− int32_t iWidth, int32_t iHeight) {
	620	+ int32_t iWidth, int32_t iHeight) {
605	621	ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
606	622	ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
607	623	if (iWidth == 16) {
@@ -620,7 +636,7 @@ static inline void McHorVer33_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uin · 12 unmodified lines
620	636	}
621	637
622	638	void McLuma_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
623		− int16_t iMvX, int16_t iMvY, int32_t iWidth, int32_t iHeight)
	639	+ int16_t iMvX, int16_t iMvY, int32_t iWidth, int32_t iHeight)
624	640	//pSrc has been added the offset of mv
625	641	{
626	642	static const PWelsMcWidthHeightFunc pWelsMcFunc[4][4] = { //[x][y]
@@ -634,9 +650,9 @@ void McLuma_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_ · 7 unmodified lines
634	650	}
635	651
636	652	void McChroma_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
637		− int16_t iMvX, int16_t iMvY, int32_t iWidth, int32_t iHeight) {
	653	+ int16_t iMvX, int16_t iMvY, int32_t iWidth, int32_t iHeight) {
638	654	static const PMcChromaWidthExtFunc kpMcChromaWidthFuncs[2] = {
639		− McChromaWidthEq4_mmx,
	655	+ McChromaWidthEq4_mmx,
640	656	McChromaWidthEq8_sse2
641	657	};
642	658	const int32_t kiD8x = iMvX & 0x07;
@@ -656,331 +672,528 @@ void McChroma_sse2 (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int3 · 13 unmodified lines
656	672	// NEON implementation //
657	673	//***************************************************************************//
658	674	#if defined(HAVE_NEON)
659		−void McCopy_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
660		− int32_t iWidth, int32_t iHeight)
661		−{
	675	+void McCopy_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	676	+ int32_t iWidth, int32_t iHeight) {
662	677	if (16 == iWidth)
663		− McCopyWidthEq16_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
664		− else if(8 == iWidth)
665		− McCopyWidthEq8_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
666		− else
667		− McCopyWidthEq4_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
668		− }
669		−void McHorVer20_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
670		− int32_t iWidth, int32_t iHeight)
671		−{
	678	+ McCopyWidthEq16_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	679	+ else if (8 == iWidth)
	680	+ McCopyWidthEq8_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	681	+ else
	682	+ McCopyWidthEq4_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	683	+}
	684	+void McHorVer20_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	685	+ int32_t iWidth, int32_t iHeight) {
672	686	if (iWidth == 16)
673		− McHorVer20WidthEq16_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
674		− else if (iWidth == 8)
675		− McHorVer20WidthEq8_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
676		− else if (iWidth == 4)
677		− McHorVer20WidthEq4_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
678		−}
679		−void McHorVer02_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
680		− int32_t iWidth, int32_t iHeight)
681		−{
682		− if (iWidth == 16)
683		− McHorVer02WidthEq16_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
684		− else if (iWidth == 8)
685		− McHorVer02WidthEq8_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
686		− else if (iWidth == 4)
687		− McHorVer02WidthEq4_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
688		−}
689		−void McHorVer22_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
690		− int32_t iWidth, int32_t iHeight)
691		−{
692		− if (iWidth == 16)
693		− McHorVer22WidthEq16_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
694		− else if (iWidth == 8)
695		− McHorVer22WidthEq8_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
696		− else if (iWidth == 4)
697		− McHorVer22WidthEq4_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
	687	+ McHorVer20WidthEq16_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	688	+ else if (iWidth == 8)
	689	+ McHorVer20WidthEq8_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	690	+ else if (iWidth == 4)
	691	+ McHorVer20WidthEq4_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	692	+}
	693	+void McHorVer02_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	694	+ int32_t iWidth, int32_t iHeight) {
	695	+ if (iWidth == 16)
	696	+ McHorVer02WidthEq16_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	697	+ else if (iWidth == 8)
	698	+ McHorVer02WidthEq8_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	699	+ else if (iWidth == 4)
	700	+ McHorVer02WidthEq4_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	701	+}
	702	+void McHorVer22_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	703	+ int32_t iWidth, int32_t iHeight) {
	704	+ if (iWidth == 16)
	705	+ McHorVer22WidthEq16_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	706	+ else if (iWidth == 8)
	707	+ McHorVer22WidthEq8_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	708	+ else if (iWidth == 4)
	709	+ McHorVer22WidthEq4_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
698	710	}
699	711
700		−void McHorVer01_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
701		− int32_t iWidth, int32_t iHeight)
702		−{
	712	+void McHorVer01_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	713	+ int32_t iWidth, int32_t iHeight) {
703	714	if (iWidth == 16)
704		− McHorVer01WidthEq16_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
	715	+ McHorVer01WidthEq16_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
705	716	else if (iWidth == 8)
706		− McHorVer01WidthEq8_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
	717	+ McHorVer01WidthEq8_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
707	718	else if (iWidth == 4)
708		− McHorVer01WidthEq4_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
	719	+ McHorVer01WidthEq4_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
709	720	}
710		−void McHorVer03_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
711		− int32_t iWidth, int32_t iHeight)
712		−{
	721	+void McHorVer03_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	722	+ int32_t iWidth, int32_t iHeight) {
713	723	if (iWidth == 16)
714		− McHorVer03WidthEq16_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
715		− else if (iWidth == 8)
716		− McHorVer03WidthEq8_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
717		− else if (iWidth == 4)
718		− McHorVer03WidthEq4_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
719		−}
720		−void McHorVer10_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
721		− int32_t iWidth, int32_t iHeight)
722		−{
	724	+ McHorVer03WidthEq16_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	725	+ else if (iWidth == 8)
	726	+ McHorVer03WidthEq8_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	727	+ else if (iWidth == 4)
	728	+ McHorVer03WidthEq4_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	729	+}
	730	+void McHorVer10_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	731	+ int32_t iWidth, int32_t iHeight) {
723	732	if (iWidth == 16)
724		− McHorVer10WidthEq16_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
725		− else if (iWidth == 8)
726		− McHorVer10WidthEq8_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
727		− else if (iWidth == 4)
728		− McHorVer10WidthEq4_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
729		−}
730		−void McHorVer11_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
731		− int32_t iWidth, int32_t iHeight)
732		−{
733		− ENFORCE_STACK_ALIGN_1D( uint8_t, pHorTmp, 256, 16 );
734		− ENFORCE_STACK_ALIGN_1D( uint8_t, pVerTmp, 256, 16 );
	733	+ McHorVer10WidthEq16_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	734	+ else if (iWidth == 8)
	735	+ McHorVer10WidthEq8_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	736	+ else if (iWidth == 4)
	737	+ McHorVer10WidthEq4_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	738	+}
	739	+void McHorVer11_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	740	+ int32_t iWidth, int32_t iHeight) {
	741	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
	742	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
	743	+ if (iWidth == 16) {
	744	+ McHorVer20WidthEq16_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	745	+ McHorVer02WidthEq16_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	746	+ PixelAvgWidthEq16_neon (pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
	747	+ } else if (iWidth == 8) {
	748	+ McHorVer20WidthEq8_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	749	+ McHorVer02WidthEq8_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	750	+ PixelAvgWidthEq8_neon (pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
	751	+ } else if (iWidth == 4) {
	752	+ McHorVer20WidthEq4_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	753	+ McHorVer02WidthEq4_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	754	+ PixelAvgWidthEq4_neon (pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
	755	+ }
	756	+}
	757	+void McHorVer12_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	758	+ int32_t iWidth, int32_t iHeight) {
	759	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
	760	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pCtrTmp, 256, 16);
	761	+ if (iWidth == 16) {
	762	+ McHorVer02WidthEq16_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	763	+ McHorVer22WidthEq16_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	764	+ PixelAvgWidthEq16_neon (pDst, iDstStride, pVerTmp, pCtrTmp, iHeight);
	765	+ } else if (iWidth == 8) {
	766	+ McHorVer02WidthEq8_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	767	+ McHorVer22WidthEq8_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	768	+ PixelAvgWidthEq8_neon (pDst, iDstStride, pVerTmp, pCtrTmp, iHeight);
	769	+ } else if (iWidth == 4) {
	770	+ McHorVer02WidthEq4_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	771	+ McHorVer22WidthEq4_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	772	+ PixelAvgWidthEq4_neon (pDst, iDstStride, pVerTmp, pCtrTmp, iHeight);
	773	+ }
	774	+}
	775	+void McHorVer13_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	776	+ int32_t iWidth, int32_t iHeight) {
	777	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
	778	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
	779	+ if (iWidth == 16) {
	780	+ McHorVer20WidthEq16_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	781	+ McHorVer02WidthEq16_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	782	+ PixelAvgWidthEq16_neon (pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
	783	+ } else if (iWidth == 8) {
	784	+ McHorVer20WidthEq8_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	785	+ McHorVer02WidthEq8_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	786	+ PixelAvgWidthEq8_neon (pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
	787	+ } else if (iWidth == 4) {
	788	+ McHorVer20WidthEq4_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	789	+ McHorVer02WidthEq4_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	790	+ PixelAvgWidthEq4_neon (pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
	791	+ }
	792	+}
	793	+void McHorVer21_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	794	+ int32_t iWidth, int32_t iHeight) {
	795	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
	796	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pCtrTmp, 256, 16);
	797	+ if (iWidth == 16) {
	798	+ McHorVer20WidthEq16_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	799	+ McHorVer22WidthEq16_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	800	+ PixelAvgWidthEq16_neon (pDst, iDstStride, pHorTmp, pCtrTmp, iHeight);
	801	+ } else if (iWidth == 8) {
	802	+ McHorVer20WidthEq8_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	803	+ McHorVer22WidthEq8_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	804	+ PixelAvgWidthEq8_neon (pDst, iDstStride, pHorTmp, pCtrTmp, iHeight);
	805	+ } else if (iWidth == 4) {
	806	+ McHorVer20WidthEq4_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	807	+ McHorVer22WidthEq4_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	808	+ PixelAvgWidthEq4_neon (pDst, iDstStride, pHorTmp, pCtrTmp, iHeight);
	809	+ }
	810	+}
	811	+void McHorVer23_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	812	+ int32_t iWidth, int32_t iHeight) {
	813	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
	814	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pCtrTmp, 256, 16);
	815	+ if (iWidth == 16) {
	816	+ McHorVer20WidthEq16_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	817	+ McHorVer22WidthEq16_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	818	+ PixelAvgWidthEq16_neon (pDst, iDstStride, pHorTmp, pCtrTmp, iHeight);
	819	+ } else if (iWidth == 8) {
	820	+ McHorVer20WidthEq8_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	821	+ McHorVer22WidthEq8_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	822	+ PixelAvgWidthEq8_neon (pDst, iDstStride, pHorTmp, pCtrTmp, iHeight);
	823	+ } else if (iWidth == 4) {
	824	+ McHorVer20WidthEq4_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	825	+ McHorVer22WidthEq4_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	826	+ PixelAvgWidthEq4_neon (pDst, iDstStride, pHorTmp, pCtrTmp, iHeight);
	827	+ }
	828	+}
	829	+void McHorVer30_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	830	+ int32_t iWidth, int32_t iHeight) {
735	831	if (iWidth == 16)
736		− {
737		− McHorVer20WidthEq16_neon(pSrc, iSrcStride, pHorTmp, 16, iHeight);
738		− McHorVer02WidthEq16_neon(pSrc, iSrcStride, pVerTmp, 16, iHeight);
739		− PixelAvgWidthEq16_neon(pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
740		− }
741		− else if (iWidth == 8)
742		− {
743		− McHorVer20WidthEq8_neon(pSrc, iSrcStride, pHorTmp, 16, iHeight);
744		− McHorVer02WidthEq8_neon(pSrc, iSrcStride, pVerTmp, 16, iHeight);
745		− PixelAvgWidthEq8_neon(pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
746		− }
747		− else if (iWidth == 4)
748		− {
749		− McHorVer20WidthEq4_neon(pSrc, iSrcStride, pHorTmp, 16, iHeight);
750		− McHorVer02WidthEq4_neon(pSrc, iSrcStride, pVerTmp, 16, iHeight);
751		− PixelAvgWidthEq4_neon(pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
752		− }
753		−}
754		−void McHorVer12_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
755		− int32_t iWidth, int32_t iHeight)
756		−{
757		− ENFORCE_STACK_ALIGN_1D( uint8_t, pVerTmp, 256, 16 );
758		− ENFORCE_STACK_ALIGN_1D( uint8_t, pCtrTmp, 256, 16 );
	832	+ McHorVer30WidthEq16_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	833	+ else if (iWidth == 8)
	834	+ McHorVer30WidthEq8_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	835	+ else if (iWidth == 4)
	836	+ McHorVer30WidthEq4_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	837	+}
	838	+void McHorVer31_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	839	+ int32_t iWidth, int32_t iHeight) {
	840	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
	841	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
	842	+ if (iWidth == 16) {
	843	+ McHorVer20WidthEq16_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	844	+ McHorVer02WidthEq16_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	845	+ PixelAvgWidthEq16_neon (pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
	846	+ } else if (iWidth == 8) {
	847	+ McHorVer20WidthEq8_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	848	+ McHorVer02WidthEq8_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	849	+ PixelAvgWidthEq8_neon (pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
	850	+ } else if (iWidth == 4) {
	851	+ McHorVer20WidthEq4_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	852	+ McHorVer02WidthEq4_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	853	+ PixelAvgWidthEq4_neon (pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
	854	+ }
	855	+}
	856	+void McHorVer32_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	857	+ int32_t iWidth, int32_t iHeight) {
	858	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
	859	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pCtrTmp, 256, 16);
	860	+ if (iWidth == 16) {
	861	+ McHorVer02WidthEq16_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	862	+ McHorVer22WidthEq16_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	863	+ PixelAvgWidthEq16_neon (pDst, iDstStride, pVerTmp, pCtrTmp, iHeight);
	864	+ } else if (iWidth == 8) {
	865	+ McHorVer02WidthEq8_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	866	+ McHorVer22WidthEq8_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	867	+ PixelAvgWidthEq8_neon (pDst, iDstStride, pVerTmp, pCtrTmp, iHeight);
	868	+ } else if (iWidth == 4) {
	869	+ McHorVer02WidthEq4_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	870	+ McHorVer22WidthEq4_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	871	+ PixelAvgWidthEq4_neon (pDst, iDstStride, pVerTmp, pCtrTmp, iHeight);
	872	+ }
	873	+}
	874	+void McHorVer33_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	875	+ int32_t iWidth, int32_t iHeight) {
	876	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
	877	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
	878	+ if (iWidth == 16) {
	879	+ McHorVer20WidthEq16_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	880	+ McHorVer02WidthEq16_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	881	+ PixelAvgWidthEq16_neon (pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
	882	+ } else if (iWidth == 8) {
	883	+ McHorVer20WidthEq8_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	884	+ McHorVer02WidthEq8_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	885	+ PixelAvgWidthEq8_neon (pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
	886	+ } else if (iWidth == 4) {
	887	+ McHorVer20WidthEq4_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	888	+ McHorVer02WidthEq4_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	889	+ PixelAvgWidthEq4_neon (pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
	890	+ }
	891	+}
	892	+
	893	+void McLuma_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	894	+ int16_t iMvX, int16_t iMvY, int32_t iWidth, int32_t iHeight) {
	895	+ static PWelsMcWidthHeightFunc pWelsMcFunc[4][4] = { //[x][y]
	896	+ {McCopy_neon, McHorVer01_neon, McHorVer02_neon, McHorVer03_neon},
	897	+ {McHorVer10_neon, McHorVer11_neon, McHorVer12_neon, McHorVer13_neon},
	898	+ {McHorVer20_neon, McHorVer21_neon, McHorVer22_neon, McHorVer23_neon},
	899	+ {McHorVer30_neon, McHorVer31_neon, McHorVer32_neon, McHorVer33_neon},
	900	+ };
	901	+ // pSrc += (iMvY >> 2) * iSrcStride + (iMvX >> 2);
	902	+ pWelsMcFunc[iMvX & 0x03][iMvY & 0x03] (pSrc, iSrcStride, pDst, iDstStride, iWidth, iHeight);
	903	+}
	904	+void McChroma_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	905	+ int16_t iMvX, int16_t iMvY, int32_t iWidth, int32_t iHeight) {
	906	+ if (0 == iMvX && 0 == iMvY) {
	907	+ if (8 == iWidth)
	908	+ McCopyWidthEq8_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	909	+ else if (iWidth == 4)
	910	+ McCopyWidthEq4_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	911	+ else //here iWidth == 2
	912	+ McCopyWidthEq2_c (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	913	+ } else {
	914	+ const int32_t kiD8x = iMvX & 0x07;
	915	+ const int32_t kiD8y = iMvY & 0x07;
	916	+ if (8 == iWidth)
	917	+ McChromaWidthEq8_neon (pSrc, iSrcStride, pDst, iDstStride, (int32_t*) (g_kuiABCD[kiD8y][kiD8x]), iHeight);
	918	+ else if (4 == iWidth)
	919	+ McChromaWidthEq4_neon (pSrc, iSrcStride, pDst, iDstStride, (int32_t*) (g_kuiABCD[kiD8y][kiD8x]), iHeight);
	920	+ else //here iWidth == 2
	921	+ McChromaWithFragMv_c (pSrc, iSrcStride, pDst, iDstStride, iMvX, iMvY, iWidth, iHeight);
	922	+ }
	923	+}
	924	+#endif
	925	+#if defined(HAVE_NEON_AARCH64)
	926	+void McCopy_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	927	+ int32_t iWidth, int32_t iHeight) {
	928	+ if (16 == iWidth)
	929	+ McCopyWidthEq16_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	930	+ else if (8 == iWidth)
	931	+ McCopyWidthEq8_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	932	+ else
	933	+ McCopyWidthEq4_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	934	+}
	935	+void McHorVer20_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	936	+ int32_t iWidth, int32_t iHeight) {
759	937	if (iWidth == 16)
760		− {
761		− McHorVer02WidthEq16_neon(pSrc, iSrcStride, pVerTmp, 16, iHeight);
762		− McHorVer22WidthEq16_neon(pSrc, iSrcStride, pCtrTmp, 16, iHeight);
763		− PixelAvgWidthEq16_neon(pDst, iDstStride, pVerTmp, pCtrTmp, iHeight);
764		− }
765		− else if (iWidth == 8)
766		− {
767		− McHorVer02WidthEq8_neon(pSrc, iSrcStride, pVerTmp, 16, iHeight);
768		− McHorVer22WidthEq8_neon(pSrc, iSrcStride, pCtrTmp, 16, iHeight);
769		− PixelAvgWidthEq8_neon(pDst, iDstStride, pVerTmp, pCtrTmp, iHeight);
770		− }
771		− else if (iWidth == 4)
772		− {
773		− McHorVer02WidthEq4_neon(pSrc, iSrcStride, pVerTmp, 16, iHeight);
774		− McHorVer22WidthEq4_neon(pSrc, iSrcStride, pCtrTmp, 16, iHeight);
775		− PixelAvgWidthEq4_neon(pDst, iDstStride, pVerTmp, pCtrTmp, iHeight);
776		− }
777		−}
778		−void McHorVer13_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
779		− int32_t iWidth, int32_t iHeight)
780		−{
781		− ENFORCE_STACK_ALIGN_1D( uint8_t, pHorTmp, 256, 16 );
782		− ENFORCE_STACK_ALIGN_1D( uint8_t, pVerTmp, 256, 16 );
	938	+ McHorVer20WidthEq16_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	939	+ else if (iWidth == 8)
	940	+ McHorVer20WidthEq8_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	941	+ else if (iWidth == 4)
	942	+ McHorVer20WidthEq4_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	943	+}
	944	+void McHorVer02_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	945	+ int32_t iWidth, int32_t iHeight) {
783	946	if (iWidth == 16)
784		− {
785		− McHorVer20WidthEq16_neon(pSrc+iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
786		− McHorVer02WidthEq16_neon(pSrc, iSrcStride, pVerTmp, 16, iHeight);
787		− PixelAvgWidthEq16_neon(pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
788		− }
789		− else if (iWidth == 8)
790		− {
791		− McHorVer20WidthEq8_neon(pSrc+iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
792		− McHorVer02WidthEq8_neon(pSrc, iSrcStride, pVerTmp, 16, iHeight);
793		− PixelAvgWidthEq8_neon(pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
794		− }
795		− else if (iWidth == 4)
796		− {
797		− McHorVer20WidthEq4_neon(pSrc+iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
798		− McHorVer02WidthEq4_neon(pSrc, iSrcStride, pVerTmp, 16, iHeight);
799		− PixelAvgWidthEq4_neon(pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
800		− }
801		−}
802		−void McHorVer21_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
803		− int32_t iWidth, int32_t iHeight)
804		−{
805		− ENFORCE_STACK_ALIGN_1D( uint8_t, pHorTmp, 256, 16 );
806		− ENFORCE_STACK_ALIGN_1D( uint8_t, pCtrTmp, 256, 16 );
	947	+ McHorVer02WidthEq16_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	948	+ else if (iWidth == 8)
	949	+ McHorVer02WidthEq8_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	950	+ else if (iWidth == 4)
	951	+ McHorVer02WidthEq4_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	952	+}
	953	+void McHorVer22_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	954	+ int32_t iWidth, int32_t iHeight) {
807	955	if (iWidth == 16)
808		− {
809		− McHorVer20WidthEq16_neon(pSrc, iSrcStride, pHorTmp, 16, iHeight);
810		− McHorVer22WidthEq16_neon(pSrc, iSrcStride, pCtrTmp, 16, iHeight);
811		− PixelAvgWidthEq16_neon(pDst, iDstStride, pHorTmp, pCtrTmp, iHeight);
812		− }
813		− else if (iWidth == 8)
814		− {
815		− McHorVer20WidthEq8_neon(pSrc, iSrcStride, pHorTmp, 16, iHeight);
816		− McHorVer22WidthEq8_neon(pSrc, iSrcStride, pCtrTmp, 16, iHeight);
817		− PixelAvgWidthEq8_neon(pDst, iDstStride, pHorTmp, pCtrTmp, iHeight);
818		− }
819		− else if (iWidth == 4)
820		− {
821		− McHorVer20WidthEq4_neon(pSrc, iSrcStride, pHorTmp, 16, iHeight);
822		− McHorVer22WidthEq4_neon(pSrc, iSrcStride, pCtrTmp, 16, iHeight);
823		− PixelAvgWidthEq4_neon(pDst, iDstStride, pHorTmp, pCtrTmp, iHeight);
824		− }
825		−}
826		−void McHorVer23_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
827		− int32_t iWidth, int32_t iHeight)
828		−{
829		− ENFORCE_STACK_ALIGN_1D( uint8_t, pHorTmp, 256, 16 );
830		− ENFORCE_STACK_ALIGN_1D( uint8_t, pCtrTmp, 256, 16 );
	956	+ McHorVer22WidthEq16_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	957	+ else if (iWidth == 8)
	958	+ McHorVer22WidthEq8_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	959	+ else if (iWidth == 4)
	960	+ McHorVer22WidthEq4_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	961	+}
	962	+
	963	+void McHorVer01_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	964	+ int32_t iWidth, int32_t iHeight) {
831	965	if (iWidth == 16)
832		− {
833		− McHorVer20WidthEq16_neon(pSrc+iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
834		− McHorVer22WidthEq16_neon(pSrc, iSrcStride, pCtrTmp, 16, iHeight);
835		− PixelAvgWidthEq16_neon(pDst, iDstStride, pHorTmp, pCtrTmp, iHeight);
836		− }
	966	+ McHorVer01WidthEq16_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
837	967	else if (iWidth == 8)
838		− {
839		− McHorVer20WidthEq8_neon(pSrc+iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
840		− McHorVer22WidthEq8_neon(pSrc, iSrcStride, pCtrTmp, 16, iHeight);
841		− PixelAvgWidthEq8_neon(pDst, iDstStride, pHorTmp, pCtrTmp, iHeight);
842		− }
843		− else if (iWidth == 4)
844		− {
845		− McHorVer20WidthEq4_neon(pSrc+iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
846		− McHorVer22WidthEq4_neon(pSrc, iSrcStride, pCtrTmp, 16, iHeight);
847		− PixelAvgWidthEq4_neon(pDst, iDstStride, pHorTmp, pCtrTmp, iHeight);
848		− }
849		−}
850		−void McHorVer30_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
851		− int32_t iWidth, int32_t iHeight)
852		−{
	968	+ McHorVer01WidthEq8_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	969	+ else if (iWidth == 4)
	970	+ McHorVer01WidthEq4_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	971	+}
	972	+void McHorVer03_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	973	+ int32_t iWidth, int32_t iHeight) {
853	974	if (iWidth == 16)
854		− McHorVer30WidthEq16_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
855		− else if (iWidth == 8)
856		− McHorVer30WidthEq8_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
857		− else if (iWidth == 4)
858		− McHorVer30WidthEq4_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
859		−}
860		−void McHorVer31_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
861		− int32_t iWidth, int32_t iHeight)
862		−{
863		− ENFORCE_STACK_ALIGN_1D( uint8_t, pHorTmp, 256, 16 );
864		− ENFORCE_STACK_ALIGN_1D( uint8_t, pVerTmp, 256, 16 );
865		− if (iWidth == 16) {
866		− McHorVer20WidthEq16_neon(pSrc, iSrcStride, pHorTmp, 16, iHeight);
867		− McHorVer02WidthEq16_neon(pSrc+1, iSrcStride, pVerTmp, 16, iHeight);
868		− PixelAvgWidthEq16_neon(pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
869		− }
870		− else if (iWidth == 8){
871		− McHorVer20WidthEq8_neon(pSrc, iSrcStride, pHorTmp, 16, iHeight);
872		− McHorVer02WidthEq8_neon(pSrc+1, iSrcStride, pVerTmp, 16, iHeight);
873		− PixelAvgWidthEq8_neon(pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
874		− }
875		− else if (iWidth == 4)
876		− {
877		− McHorVer20WidthEq4_neon(pSrc, iSrcStride, pHorTmp, 16, iHeight);
878		− McHorVer02WidthEq4_neon(pSrc+1, iSrcStride, pVerTmp, 16, iHeight);
879		− PixelAvgWidthEq4_neon(pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
880		− }
881		−}
882		−void McHorVer32_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
883		− int32_t iWidth, int32_t iHeight)
884		−{
885		− ENFORCE_STACK_ALIGN_1D( uint8_t, pVerTmp, 256, 16 );
886		− ENFORCE_STACK_ALIGN_1D( uint8_t, pCtrTmp, 256, 16 );
	975	+ McHorVer03WidthEq16_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	976	+ else if (iWidth == 8)
	977	+ McHorVer03WidthEq8_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	978	+ else if (iWidth == 4)
	979	+ McHorVer03WidthEq4_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	980	+}
	981	+void McHorVer10_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	982	+ int32_t iWidth, int32_t iHeight) {
887	983	if (iWidth == 16)
888		− {
889		− McHorVer02WidthEq16_neon(pSrc+1, iSrcStride, pVerTmp, 16, iHeight);
890		− McHorVer22WidthEq16_neon(pSrc, iSrcStride, pCtrTmp, 16, iHeight);
891		− PixelAvgWidthEq16_neon(pDst, iDstStride, pVerTmp, pCtrTmp, iHeight);
892		− }
893		− else if (iWidth == 8)
894		− {
895		− McHorVer02WidthEq8_neon(pSrc+1, iSrcStride, pVerTmp, 16, iHeight);
896		− McHorVer22WidthEq8_neon(pSrc, iSrcStride, pCtrTmp, 16, iHeight);
897		− PixelAvgWidthEq8_neon(pDst, iDstStride, pVerTmp, pCtrTmp, iHeight);
898		− }
899		− else if (iWidth == 4)
900		− {
901		− McHorVer02WidthEq4_neon(pSrc+1, iSrcStride, pVerTmp, 16, iHeight);
902		− McHorVer22WidthEq4_neon(pSrc, iSrcStride, pCtrTmp, 16, iHeight);
903		− PixelAvgWidthEq4_neon(pDst, iDstStride, pVerTmp, pCtrTmp, iHeight);
904		− }
905		−}
906		−void McHorVer33_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
907		− int32_t iWidth, int32_t iHeight)
908		−{
909		− ENFORCE_STACK_ALIGN_1D( uint8_t, pHorTmp, 256, 16 );
910		− ENFORCE_STACK_ALIGN_1D( uint8_t, pVerTmp, 256, 16 );
	984	+ McHorVer10WidthEq16_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	985	+ else if (iWidth == 8)
	986	+ McHorVer10WidthEq8_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	987	+ else if (iWidth == 4)
	988	+ McHorVer10WidthEq4_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	989	+}
	990	+void McHorVer11_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	991	+ int32_t iWidth, int32_t iHeight) {
	992	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
	993	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
	994	+ if (iWidth == 16) {
	995	+ McHorVer20WidthEq16_AArch64_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	996	+ McHorVer02WidthEq16_AArch64_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	997	+ PixelAvgWidthEq16_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pVerTmp, 16, iHeight);
	998	+ } else if (iWidth == 8) {
	999	+ McHorVer20WidthEq8_AArch64_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	1000	+ McHorVer02WidthEq8_AArch64_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	1001	+ PixelAvgWidthEq8_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pVerTmp, 16, iHeight);
	1002	+ } else if (iWidth == 4) {
	1003	+ McHorVer20WidthEq4_AArch64_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	1004	+ McHorVer02WidthEq4_AArch64_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	1005	+ PixelAvgWidthEq4_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pVerTmp, 16, iHeight);
	1006	+ }
	1007	+}
	1008	+void McHorVer12_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	1009	+ int32_t iWidth, int32_t iHeight) {
	1010	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
	1011	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pCtrTmp, 256, 16);
	1012	+ if (iWidth == 16) {
	1013	+ McHorVer02WidthEq16_AArch64_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	1014	+ McHorVer22WidthEq16_AArch64_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	1015	+ PixelAvgWidthEq16_AArch64_neon (pDst, iDstStride, pVerTmp, 16, pCtrTmp, 16, iHeight);
	1016	+ } else if (iWidth == 8) {
	1017	+ McHorVer02WidthEq8_AArch64_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	1018	+ McHorVer22WidthEq8_AArch64_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	1019	+ PixelAvgWidthEq8_AArch64_neon (pDst, iDstStride, pVerTmp, 16, pCtrTmp, 16, iHeight);
	1020	+ } else if (iWidth == 4) {
	1021	+ McHorVer02WidthEq4_AArch64_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	1022	+ McHorVer22WidthEq4_AArch64_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	1023	+ PixelAvgWidthEq4_AArch64_neon (pDst, iDstStride, pVerTmp, 16, pCtrTmp, 16, iHeight);
	1024	+ }
	1025	+}
	1026	+void McHorVer13_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	1027	+ int32_t iWidth, int32_t iHeight) {
	1028	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
	1029	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
	1030	+ if (iWidth == 16) {
	1031	+ McHorVer20WidthEq16_AArch64_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	1032	+ McHorVer02WidthEq16_AArch64_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	1033	+ PixelAvgWidthEq16_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pVerTmp, 16, iHeight);
	1034	+ } else if (iWidth == 8) {
	1035	+ McHorVer20WidthEq8_AArch64_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	1036	+ McHorVer02WidthEq8_AArch64_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	1037	+ PixelAvgWidthEq8_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pVerTmp, 16, iHeight);
	1038	+ } else if (iWidth == 4) {
	1039	+ McHorVer20WidthEq4_AArch64_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	1040	+ McHorVer02WidthEq4_AArch64_neon (pSrc, iSrcStride, pVerTmp, 16, iHeight);
	1041	+ PixelAvgWidthEq4_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pVerTmp, 16, iHeight);
	1042	+ }
	1043	+}
	1044	+void McHorVer21_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	1045	+ int32_t iWidth, int32_t iHeight) {
	1046	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
	1047	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pCtrTmp, 256, 16);
	1048	+ if (iWidth == 16) {
	1049	+ McHorVer20WidthEq16_AArch64_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	1050	+ McHorVer22WidthEq16_AArch64_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	1051	+ PixelAvgWidthEq16_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pCtrTmp, 16, iHeight);
	1052	+ } else if (iWidth == 8) {
	1053	+ McHorVer20WidthEq8_AArch64_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	1054	+ McHorVer22WidthEq8_AArch64_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	1055	+ PixelAvgWidthEq8_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pCtrTmp, 16, iHeight);
	1056	+ } else if (iWidth == 4) {
	1057	+ McHorVer20WidthEq4_AArch64_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	1058	+ McHorVer22WidthEq4_AArch64_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	1059	+ PixelAvgWidthEq4_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pCtrTmp, 16, iHeight);
	1060	+ }
	1061	+}
	1062	+void McHorVer23_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	1063	+ int32_t iWidth, int32_t iHeight) {
	1064	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
	1065	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pCtrTmp, 256, 16);
	1066	+ if (iWidth == 16) {
	1067	+ McHorVer20WidthEq16_AArch64_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	1068	+ McHorVer22WidthEq16_AArch64_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	1069	+ PixelAvgWidthEq16_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pCtrTmp, 16, iHeight);
	1070	+ } else if (iWidth == 8) {
	1071	+ McHorVer20WidthEq8_AArch64_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	1072	+ McHorVer22WidthEq8_AArch64_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	1073	+ PixelAvgWidthEq8_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pCtrTmp, 16, iHeight);
	1074	+ } else if (iWidth == 4) {
	1075	+ McHorVer20WidthEq4_AArch64_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	1076	+ McHorVer22WidthEq4_AArch64_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	1077	+ PixelAvgWidthEq4_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pCtrTmp, 16, iHeight);
	1078	+ }
	1079	+}
	1080	+void McHorVer30_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	1081	+ int32_t iWidth, int32_t iHeight) {
911	1082	if (iWidth == 16)
912		− {
913		− McHorVer20WidthEq16_neon(pSrc+iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
914		− McHorVer02WidthEq16_neon(pSrc+1, iSrcStride, pVerTmp, 16, iHeight);
915		− PixelAvgWidthEq16_neon(pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
916		− }
917		− else if (iWidth == 8)
918		− {
919		− McHorVer20WidthEq8_neon(pSrc+iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
920		− McHorVer02WidthEq8_neon(pSrc+1, iSrcStride, pVerTmp, 16, iHeight);
921		− PixelAvgWidthEq8_neon(pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
922		− }
923		− else if (iWidth == 4)
924		− {
925		− McHorVer20WidthEq4_neon(pSrc+iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
926		− McHorVer02WidthEq4_neon(pSrc+1, iSrcStride, pVerTmp, 16, iHeight);
927		− PixelAvgWidthEq4_neon(pDst, iDstStride, pHorTmp, pVerTmp, iHeight);
928		− }
929		−}
930		−
931		−void McLuma_neon(const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
932		− int16_t iMvX, int16_t iMvY, int32_t iWidth, int32_t iHeight)
933		−{
934		− static PWelsMcWidthHeightFunc pWelsMcFunc[4][4] = //[x][y]
935		− {
936		− {McCopy_neon, McHorVer01_neon, McHorVer02_neon, McHorVer03_neon},
937		− {McHorVer10_neon, McHorVer11_neon, McHorVer12_neon, McHorVer13_neon},
938		− {McHorVer20_neon, McHorVer21_neon, McHorVer22_neon, McHorVer23_neon},
939		− {McHorVer30_neon, McHorVer31_neon, McHorVer32_neon, McHorVer33_neon},
940		− };
	1083	+ McHorVer30WidthEq16_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	1084	+ else if (iWidth == 8)
	1085	+ McHorVer30WidthEq8_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	1086	+ else if (iWidth == 4)
	1087	+ McHorVer30WidthEq4_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	1088	+}
	1089	+void McHorVer31_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	1090	+ int32_t iWidth, int32_t iHeight) {
	1091	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
	1092	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
	1093	+ if (iWidth == 16) {
	1094	+ McHorVer20WidthEq16_AArch64_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	1095	+ McHorVer02WidthEq16_AArch64_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	1096	+ PixelAvgWidthEq16_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pVerTmp, 16, iHeight);
	1097	+ } else if (iWidth == 8) {
	1098	+ McHorVer20WidthEq8_AArch64_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	1099	+ McHorVer02WidthEq8_AArch64_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	1100	+ PixelAvgWidthEq8_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pVerTmp, 16, iHeight);
	1101	+ } else if (iWidth == 4) {
	1102	+ McHorVer20WidthEq4_AArch64_neon (pSrc, iSrcStride, pHorTmp, 16, iHeight);
	1103	+ McHorVer02WidthEq4_AArch64_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	1104	+ PixelAvgWidthEq4_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pVerTmp, 16, iHeight);
	1105	+ }
	1106	+}
	1107	+void McHorVer32_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	1108	+ int32_t iWidth, int32_t iHeight) {
	1109	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
	1110	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pCtrTmp, 256, 16);
	1111	+ if (iWidth == 16) {
	1112	+ McHorVer02WidthEq16_AArch64_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	1113	+ McHorVer22WidthEq16_AArch64_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	1114	+ PixelAvgWidthEq16_AArch64_neon (pDst, iDstStride, pVerTmp, 16, pCtrTmp, 16, iHeight);
	1115	+ } else if (iWidth == 8) {
	1116	+ McHorVer02WidthEq8_AArch64_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	1117	+ McHorVer22WidthEq8_AArch64_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	1118	+ PixelAvgWidthEq8_AArch64_neon (pDst, iDstStride, pVerTmp, 16, pCtrTmp, 16, iHeight);
	1119	+ } else if (iWidth == 4) {
	1120	+ McHorVer02WidthEq4_AArch64_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	1121	+ McHorVer22WidthEq4_AArch64_neon (pSrc, iSrcStride, pCtrTmp, 16, iHeight);
	1122	+ PixelAvgWidthEq4_AArch64_neon (pDst, iDstStride, pVerTmp, 16, pCtrTmp, 16, iHeight);
	1123	+ }
	1124	+}
	1125	+void McHorVer33_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	1126	+ int32_t iWidth, int32_t iHeight) {
	1127	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pHorTmp, 256, 16);
	1128	+ ENFORCE_STACK_ALIGN_1D (uint8_t, pVerTmp, 256, 16);
	1129	+ if (iWidth == 16) {
	1130	+ McHorVer20WidthEq16_AArch64_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	1131	+ McHorVer02WidthEq16_AArch64_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	1132	+ PixelAvgWidthEq16_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pVerTmp, 16, iHeight);
	1133	+ } else if (iWidth == 8) {
	1134	+ McHorVer20WidthEq8_AArch64_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	1135	+ McHorVer02WidthEq8_AArch64_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	1136	+ PixelAvgWidthEq8_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pVerTmp, 16, iHeight);
	1137	+ } else if (iWidth == 4) {
	1138	+ McHorVer20WidthEq4_AArch64_neon (pSrc + iSrcStride, iSrcStride, pHorTmp, 16, iHeight);
	1139	+ McHorVer02WidthEq4_AArch64_neon (pSrc + 1, iSrcStride, pVerTmp, 16, iHeight);
	1140	+ PixelAvgWidthEq4_AArch64_neon (pDst, iDstStride, pHorTmp, 16, pVerTmp, 16, iHeight);
	1141	+ }
	1142	+}
	1143	+
	1144	+void McLuma_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	1145	+ int16_t iMvX, int16_t iMvY, int32_t iWidth, int32_t iHeight) {
	1146	+ static PWelsMcWidthHeightFunc pWelsMcFunc[4][4] = { //[x][y]
	1147	+ {McCopy_AArch64_neon, McHorVer01_AArch64_neon, McHorVer02_AArch64_neon, McHorVer03_AArch64_neon},
	1148	+ {McHorVer10_AArch64_neon, McHorVer11_AArch64_neon, McHorVer12_AArch64_neon, McHorVer13_AArch64_neon},
	1149	+ {McHorVer20_AArch64_neon, McHorVer21_AArch64_neon, McHorVer22_AArch64_neon, McHorVer23_AArch64_neon},
	1150	+ {McHorVer30_AArch64_neon, McHorVer31_AArch64_neon, McHorVer32_AArch64_neon, McHorVer33_AArch64_neon},
	1151	+ };
941	1152	// pSrc += (iMvY >> 2) * iSrcStride + (iMvX >> 2);
942		− pWelsMcFunc[iMvX&0x03][iMvY&0x03](pSrc, iSrcStride, pDst, iDstStride, iWidth, iHeight);
	1153	+ pWelsMcFunc[iMvX & 0x03][iMvY & 0x03] (pSrc, iSrcStride, pDst, iDstStride, iWidth, iHeight);
943	1154	}
944		−void McChroma_neon(const uint8_t pSrc, int32_t iSrcStride, uint8_t pDst, int32_t iDstStride,
945		− int16_t iMvX, int16_t iMvY, int32_t iWidth, int32_t iHeight)
946		−{
947		− if (0 == iMvX && 0 == iMvY)
948		− {
949		− if(8 == iWidth)
950		− McCopyWidthEq8_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
951		− else if(iWidth == 4)
952		− McCopyWidthEq4_neon(pSrc, iSrcStride, pDst, iDstStride, iHeight);
953		− else //here iWidth == 2
954		− McCopyWidthEq2_c(pSrc,iSrcStride,pDst,iDstStride,iHeight);
955		− }
956		− else
957		− {
958		− const int32_t kiD8x = iMvX & 0x07;
959		− const int32_t kiD8y = iMvY & 0x07;
960		− if(8 == iWidth)
961		− McChromaWidthEq8_neon(pSrc, iSrcStride, pDst, iDstStride, (int32_t*)(g_kuiABCD[kiD8y][kiD8x]), iHeight);
962		− else if(4 == iWidth)
963		− McChromaWidthEq4_neon(pSrc, iSrcStride, pDst, iDstStride, (int32_t*)(g_kuiABCD[kiD8y][kiD8x]), iHeight);
964		− else //here iWidth == 2
965		− McChromaWithFragMv_c(pSrc, iSrcStride, pDst, iDstStride, iMvX, iMvY, iWidth, iHeight);
966		− }
	1155	+void McChroma_AArch64_neon (const uint8_t* pSrc, int32_t iSrcStride, uint8_t* pDst, int32_t iDstStride,
	1156	+ int16_t iMvX, int16_t iMvY, int32_t iWidth, int32_t iHeight) {
	1157	+ if (0 == iMvX && 0 == iMvY) {
	1158	+ if (8 == iWidth)
	1159	+ McCopyWidthEq8_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	1160	+ else if (iWidth == 4)
	1161	+ McCopyWidthEq4_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	1162	+ else //here iWidth == 2
	1163	+ McCopyWidthEq2_c (pSrc, iSrcStride, pDst, iDstStride, iHeight);
	1164	+ } else {
	1165	+ const int32_t kiD8x = iMvX & 0x07;
	1166	+ const int32_t kiD8y = iMvY & 0x07;
	1167	+ if (8 == iWidth)
	1168	+ McChromaWidthEq8_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, (int32_t*) (g_kuiABCD[kiD8y][kiD8x]), iHeight);
	1169	+ else if (4 == iWidth)
	1170	+ McChromaWidthEq4_AArch64_neon (pSrc, iSrcStride, pDst, iDstStride, (int32_t*) (g_kuiABCD[kiD8y][kiD8x]), iHeight);
	1171	+ else //here iWidth == 2
	1172	+ McChromaWithFragMv_c (pSrc, iSrcStride, pDst, iDstStride, iMvX, iMvY, iWidth, iHeight);
	1173	+ }
967	1174	}
968	1175	#endif
	1176	+
969	1177	void InitMcFunc (SMcFunc* pMcFunc, int32_t iCpu) {
970	1178	pMcFunc->pMcLumaFunc = McLuma_c;
971	1179	pMcFunc->pMcChromaFunc = McChroma_c;
972	1180
973	1181	#ifdef HAVE_NEON
974		− if ( iCpu & WELS_CPU_NEON ) {
975		− pMcFunc->pMcLumaFunc = McLuma_neon;
976		− pMcFunc->pMcChromaFunc = McChroma_neon;
977		− }
	1182	+ if (iCpu & WELS_CPU_NEON) {
	1183	+ pMcFunc->pMcLumaFunc = McLuma_neon;
	1184	+ pMcFunc->pMcChromaFunc = McChroma_neon;
	1185	+ }
	1186	+#endif
	1187	+#ifdef HAVE_NEON_AARCH64
	1188	+ if (iCpu & WELS_CPU_NEON) {
	1189	+ pMcFunc->pMcLumaFunc = McLuma_AArch64_neon;
	1190	+ pMcFunc->pMcChromaFunc = McChroma_AArch64_neon;
	1191	+ }
978	1192	#endif
979		−
980	1193	#if defined (X86_ASM)
981	1194	if (iCpu & WELS_CPU_SSE2) {
982		− pMcFunc->pMcLumaFunc = McLuma_sse2;
983		− pMcFunc->pMcChromaFunc = McChroma_sse2;
	1195	+ pMcFunc->pMcLumaFunc = McLuma_sse2;
	1196	+ pMcFunc->pMcChromaFunc = McChroma_sse2;
984	1197	}
985	1198	#endif //(X86_ASM)
986	1199	}

Add Motion Compehension ARM64 Neon Code

Keyboard shortcuts

General

Issues and pull requests

Issues, pull requests and Inbox

Repository code