diff --git a/source/Lib/CommonLib/Buffer.cpp b/source/Lib/CommonLib/Buffer.cpp
index d211a000e41d7bde51e0c05cf6b1da7852b629f6..c739d7c02a31969e3e1338896a3471a2b972da1a 100644
--- a/source/Lib/CommonLib/Buffer.cpp
+++ b/source/Lib/CommonLib/Buffer.cpp
@@ -282,7 +282,7 @@ void addBIOAvgCore(const Pel* src0, int src0Stride, const Pel* src1, int src1Str
}
#if MULTI_PASS_DMVR || SAMPLE_BASED_BDOF
-void calcBIOParameterCore(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0, Pel* gradX1, Pel* gradY0, Pel* gradY1, int width, int height, const int src0Stride, const int src1Stride, const int widthG, const int bitDepth, Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGY_GX, Pel* dI)
+void calcBIOParameterCore(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0, Pel* gradX1, Pel* gradY0, Pel* gradY1, int width, int height, const int src0Stride, const int src1Stride, const int widthG, const int bitDepth, Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGyGx, Pel* dI)
{
width -= 2;
height -= 2;
@@ -293,7 +293,7 @@ void calcBIOParameterCore(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0,
gradY0 += bioParamOffset; gradY1 += bioParamOffset;
absGX += bioParamOffset; absGY += bioParamOffset;
dIX += bioParamOffset; dIY += bioParamOffset;
- signGY_GX += bioParamOffset;
+ signGyGx += bioParamOffset;
int shift4 = 4;
int shift5 = 1;
if (dI)
@@ -311,7 +311,7 @@ void calcBIOParameterCore(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0,
absGY[x] = (tmpGY < 0 ? -tmpGY : tmpGY);
dIX[x] = (tmpGX < 0 ? -tmpDI : (tmpGX == 0 ? 0 : tmpDI));
dIY[x] = (tmpGY < 0 ? -tmpDI : (tmpGY == 0 ? 0 : tmpDI));
- signGY_GX[x] = (tmpGY < 0 ? -tmpGX : (tmpGY == 0 ? 0 : tmpGX));
+ signGyGx[x] = (tmpGY < 0 ? -tmpGX : (tmpGY == 0 ? 0 : tmpGX));
}
srcY0Tmp += src0Stride;
srcY1Tmp += src1Stride;
@@ -324,7 +324,7 @@ void calcBIOParameterCore(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0,
dI += widthG;
dIX += widthG;
dIY += widthG;
- signGY_GX += widthG;
+ signGyGx += widthG;
}
return;
@@ -341,7 +341,7 @@ void calcBIOParameterCore(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0,
absGY[x] = (tmpGY < 0 ? -tmpGY : tmpGY);
dIX[x] = (tmpGX < 0 ? -tmpDI : (tmpGX == 0 ? 0 : tmpDI));
dIY[x] = (tmpGY < 0 ? -tmpDI : (tmpGY == 0 ? 0 : tmpDI));
- signGY_GX[x] = (tmpGY < 0 ? -tmpGX : (tmpGY == 0 ? 0 : tmpGX));
+ signGyGx[x] = (tmpGY < 0 ? -tmpGX : (tmpGY == 0 ? 0 : tmpGX));
}
srcY0Tmp += src0Stride;
srcY1Tmp += src1Stride;
@@ -353,55 +353,55 @@ void calcBIOParameterCore(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0,
absGY += widthG;
dIX += widthG;
dIY += widthG;
- signGY_GX += widthG;
+ signGyGx += widthG;
}
}
-void calcBIOParamSum5Core(Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGY_GX, const int widthG, const int width, const int height, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGY_GX)
+void calcBIOParamSum5Core(Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGyGx, const int widthG, const int width, const int height, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGyGx)
{
for (int y = 0; y < height; y++)
{
for (int x = 0; x < width; x++)
{
- const int pixel_idx = y * width + x;
- sumAbsGX[pixel_idx] = 0;
- sumAbsGY[pixel_idx] = 0;
- sumDIX[pixel_idx] = 0;
- sumDIY[pixel_idx] = 0;
- sumSignGY_GX[pixel_idx] = 0;
+ const int sampleIdx = y * width + x;
+ sumAbsGX[sampleIdx] = 0;
+ sumAbsGY[sampleIdx] = 0;
+ sumDIX[sampleIdx] = 0;
+ sumDIY[sampleIdx] = 0;
+ sumSignGyGx[sampleIdx] = 0;
for (int yy = 0; yy < 5; yy++)
{
for (int xx = 0; xx < 5; xx++)
{
- sumAbsGX[pixel_idx] += absGX[xx];
- sumAbsGY[pixel_idx] += absGY[xx];
- sumDIX[pixel_idx] += dIX[xx];
- sumDIY[pixel_idx] += dIY[xx];
- sumSignGY_GX[pixel_idx] += signGY_GX[xx];
+ sumAbsGX[sampleIdx] += absGX[xx];
+ sumAbsGY[sampleIdx] += absGY[xx];
+ sumDIX[sampleIdx] += dIX[xx];
+ sumDIY[sampleIdx] += dIY[xx];
+ sumSignGyGx[sampleIdx] += signGyGx[xx];
}
absGX += widthG;
absGY += widthG;
dIX += widthG;
dIY += widthG;
- signGY_GX += widthG;
+ signGyGx += widthG;
}
- sumDIX[pixel_idx] <<= 2;
- sumDIY[pixel_idx] <<= 2;
+ sumDIX[sampleIdx] <<= 2;
+ sumDIY[sampleIdx] <<= 2;
absGX += (1 - 5 * widthG);
absGY += (1 - 5 * widthG);
dIX += (1 - 5 * widthG);
dIY += (1 - 5 * widthG);
- signGY_GX += (1 - 5 * widthG);
+ signGyGx += (1 - 5 * widthG);
}
absGX += (widthG - width);
absGY += (widthG - width);
dIX += (widthG - width);
dIY += (widthG - width);
- signGY_GX += (widthG - width);
+ signGyGx += (widthG - width);
}
}
-void calcBIOParamSum4Core(Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGY_GX, int width, int height, const int widthG, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGY_GX)
+void calcBIOParamSum4Core(Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGyGx, int width, int height, const int widthG, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGyGx)
{
for (int y = 0; y < height; y++)
{
@@ -411,30 +411,30 @@ void calcBIOParamSum4Core(Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signG
*sumAbsGY += absGY[x];
*sumDIX += dIX[x];
*sumDIY += dIY[x];
- *sumSignGY_GX += signGY_GX[x];
+ *sumSignGyGx += signGyGx[x];
}
absGX += widthG;
absGY += widthG;
dIX += widthG;
dIY += widthG;
- signGY_GX += widthG;
+ signGyGx += widthG;
}
}
-void calcBIOClippedVxVyCore(int* sumDIX_pixel_32bit, int* sumAbsGX_pixel_32bit, int* sumDIY_pixel_32bit, int* sumAbsGY_pixel_32bit, int* sumSignGY_GX_pixel_32bit, const int limit, const int bioSubblockSize, int* tmpx_pixel_32bit, int* tmpy_pixel_32bit)
+void calcBIOClippedVxVyCore(int* sumDIXSample32bit, int* sumAbsGxSample32bit, int* sumDIYSample32bit, int* sumAbsGySample32bit, int* sumSignGyGxSample32bit, const int limit, const int bioSubblockSize, int* tmpxSample32bit, int* tmpySample32bit)
{
for (int idx = 0; idx < bioSubblockSize; idx++)
{
- *tmpx_pixel_32bit = Clip3(-limit, limit, (*sumDIX_pixel_32bit) >> (*sumAbsGX_pixel_32bit));
- int tmpData = ((*sumSignGY_GX_pixel_32bit) * (*tmpx_pixel_32bit)) >> 1;
- *tmpy_pixel_32bit = Clip3(-limit, limit, (((*sumDIY_pixel_32bit) - tmpData) >> (*sumAbsGY_pixel_32bit)));
- sumDIX_pixel_32bit++;
- sumAbsGX_pixel_32bit++;
- sumDIY_pixel_32bit++;
- sumAbsGY_pixel_32bit++;
- sumSignGY_GX_pixel_32bit++;
- tmpx_pixel_32bit++;
- tmpy_pixel_32bit++;
+ *tmpxSample32bit = Clip3(-limit, limit, (*sumDIXSample32bit) >> (*sumAbsGxSample32bit));
+ int tmpData = ((*sumSignGyGxSample32bit) * (*tmpxSample32bit)) >> 1;
+ *tmpySample32bit = Clip3(-limit, limit, (((*sumDIYSample32bit) - tmpData) >> (*sumAbsGySample32bit)));
+ sumDIXSample32bit++;
+ sumAbsGxSample32bit++;
+ sumDIYSample32bit++;
+ sumAbsGySample32bit++;
+ sumSignGyGxSample32bit++;
+ tmpxSample32bit++;
+ tmpySample32bit++;
}
}
#if JVET_Z0136_OOB
@@ -542,7 +542,7 @@ void calAbsSumCore(const Pel* diff, int stride, int width, int height, int* absS
}
#endif
-template<bool PAD = true>
+template<bool pad = true>
void gradFilterCore(Pel* pSrc, int srcStride, int width, int height, int gradStride, Pel* gradX, Pel* gradY, const int bitDepth)
{
Pel* srcTmp = pSrc + srcStride + 1;
@@ -569,7 +569,7 @@ void gradFilterCore(Pel* pSrc, int srcStride, int width, int height, int gradStr
}
#if !MULTI_PASS_DMVR && !SAMPLE_BASED_BDOF
- if (PAD)
+ if (pad)
{
gradXTmp = gradX + gradStride + 1;
gradYTmp = gradY + gradStride + 1;
@@ -594,7 +594,7 @@ void gradFilterCore(Pel* pSrc, int srcStride, int width, int height, int gradStr
#endif
}
-void calcBIOSumsCore(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0, Pel* gradX1, Pel* gradY0, Pel* gradY1, int xu, int yu, const int src0Stride, const int src1Stride, const int widthG, const int bitDepth, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGY_GX)
+void calcBIOSumsCore(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0, Pel* gradX1, Pel* gradY0, Pel* gradY1, int xu, int yu, const int src0Stride, const int src1Stride, const int widthG, const int bitDepth, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGyGx)
{
int shift4 = 4;
int shift5 = 1;
@@ -610,7 +610,7 @@ void calcBIOSumsCore(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0, Pel*
*sumAbsGY += (tmpGY < 0 ? -tmpGY : tmpGY);
*sumDIX += (tmpGX < 0 ? -tmpDI : (tmpGX == 0 ? 0 : tmpDI));
*sumDIY += (tmpGY < 0 ? -tmpDI : (tmpGY == 0 ? 0 : tmpDI));
- *sumSignGY_GX += (tmpGY < 0 ? -tmpGX : (tmpGY == 0 ? 0 : tmpGX));
+ *sumSignGyGx += (tmpGY < 0 ? -tmpGX : (tmpGY == 0 ? 0 : tmpGX));
}
srcY1Tmp += src1Stride;
diff --git a/source/Lib/CommonLib/Buffer.h b/source/Lib/CommonLib/Buffer.h
index 943cb213fe119d60596d9327d0051ac3d2a96abe..daf28f03bbf1cae3e5a743d5b6f6528b2f39c346 100644
--- a/source/Lib/CommonLib/Buffer.h
+++ b/source/Lib/CommonLib/Buffer.h
@@ -80,18 +80,18 @@ struct PelBufferOps
void(*bioGradFilter) (Pel* pSrc, int srcStride, int width, int height, int gradStride, Pel* gradX, Pel* gradY, const int bitDepth);
void(*calcBIOPar) (const Pel* srcY0Temp, const Pel* srcY1Temp, const Pel* gradX0, const Pel* gradX1, const Pel* gradY0, const Pel* gradY1, int* dotProductTemp1, int* dotProductTemp2, int* dotProductTemp3, int* dotProductTemp5, int* dotProductTemp6, const int src0Stride, const int src1Stride, const int gradStride, const int widthG, const int heightG, const int bitDepth);
#if MULTI_PASS_DMVR || SAMPLE_BASED_BDOF
- void(*calcBIOParameter) (const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0, Pel* gradX1, Pel* gradY0, Pel* gradY1, int width, int height, const int src0Stride, const int src1Stride, const int widthG, const int bitDepth, Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGY_GX, Pel* dI);
+ void(*calcBIOParameter) (const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0, Pel* gradX1, Pel* gradY0, Pel* gradY1, int width, int height, const int src0Stride, const int src1Stride, const int widthG, const int bitDepth, Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGyGx, Pel* dI);
void(*calAbsSum) (const Pel* diff, int stride, int width, int height, int* absDiff);
- void(*calcBIOParamSum5) (Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGY_GX, const int widthG, const int width, const int height, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGY_GX);
- void(*calcBIOParamSum4) (Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGY_GX, int width, int height, const int widthG, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGY_GX);
+ void(*calcBIOParamSum5) (Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGyGx, const int widthG, const int width, const int height, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGyGx);
+ void(*calcBIOParamSum4) (Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGyGx, int width, int height, const int widthG, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGyGx);
#if JVET_Z0136_OOB
void(*addBIOAvgN) (const Pel* src0, int src0Stride, const Pel* src1, int src1Stride, Pel *dst, int dstStride, const Pel *gradX0, const Pel *gradX1, const Pel *gradY0, const Pel *gradY1, int gradStride, int width, int height, int* tmpx, int* tmpy, int shift, int offset, const ClpRng& clpRng, bool *mcMask[2], int mcStride, bool *isOOB);
#else
void(*addBIOAvgN) (const Pel* src0, int src0Stride, const Pel* src1, int src1Stride, Pel *dst, int dstStride, const Pel *gradX0, const Pel *gradX1, const Pel *gradY0, const Pel *gradY1, int gradStride, int width, int height, int* tmpx, int* tmpy, int shift, int offset, const ClpRng& clpRng);
#endif
- void(*calcBIOClippedVxVy) (int* sumDIX_pixel_32bit, int* sumAbsGX_pixel_32bit, int* sumDIY_pixel_32bit, int* sumAbsGY_pixel_32bit, int* sumSignGY_GX_pixel_32bit, const int limit, const int bioSubblockSize, int* tmpx_pixel_32bit, int* tmpy_pixel_32bit);
+ void(*calcBIOClippedVxVy) (int* sumDIXSample32bit, int* sumAbsGxSample32bit, int* sumDIYSample32bit, int* sumAbsGySample32bit, int* sumSignGyGxSample32bit, const int limit, const int bioSubblockSize, int* tmpxSample32bit, int* tmpySample32bit);
#endif
- void(*calcBIOSums) (const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0, Pel* gradX1, Pel* gradY0, Pel* gradY1, int xu, int yu, const int src0Stride, const int src1Stride, const int widthG, const int bitDepth, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGY_GX);
+ void(*calcBIOSums) (const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0, Pel* gradX1, Pel* gradY0, Pel* gradY1, int xu, int yu, const int src0Stride, const int src1Stride, const int widthG, const int bitDepth, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGyGx);
void(*calcBlkGradient)(int sx, int sy, int *arraysGx2, int *arraysGxGy, int *arraysGxdI, int *arraysGy2, int *arraysGydI, int &sGx2, int &sGy2, int &sGxGy, int &sGxdI, int &sGydI, int width, int height, int unitSize);
void(*copyBuffer)(Pel *src, int srcStride, Pel *dst, int dstStride, int width, int height);
void(*padding)(Pel *dst, int stride, int width, int height, int padSize);
diff --git a/source/Lib/CommonLib/CodingStructure.cpp b/source/Lib/CommonLib/CodingStructure.cpp
index a21ab886c8f604202ab005653c07ffca279ff920..e8b069d86bf05a57ce98ce38011fd2ea5f6dd07b 100644
--- a/source/Lib/CommonLib/CodingStructure.cpp
+++ b/source/Lib/CommonLib/CodingStructure.cpp
@@ -2294,35 +2294,47 @@ void CodingStructure::useSubStructure( const CodingStructure& subStruct, const C
if( parent )
{
// copy data to picture
- if( cpyPred ) getPredBuf ( clippedArea ).copyFrom( subPredBuf );
-#if JVET_AC0162_ALF_RESIDUAL_SAMPLES_INPUT
+ if( cpyPred )
+ {
+ getPredBuf( clippedArea ).copyFrom( subPredBuf );
+ }
+
if (cpyResi)
{
getResiBuf(clippedArea).copyFrom(subResiBuf);
}
+#if JVET_AC0162_ALF_RESIDUAL_SAMPLES_INPUT
else
{
getResiBuf(clippedArea).copyFrom(subStruct.getResiBuf(clippedArea), true);
}
-#else
- if( cpyResi ) getResiBuf ( clippedArea ).copyFrom( subResiBuf );
#endif
- if( cpyReco ) getRecoBuf ( clippedArea ).copyFrom( subRecoBuf );
- if( cpyOrgResi ) getOrgResiBuf( clippedArea ).copyFrom( subStruct.getOrgResiBuf( clippedArea ) );
+
+ if( cpyReco )
+ {
+ getRecoBuf( clippedArea ).copyFrom( subRecoBuf );
+ }
+
+ if( cpyOrgResi )
+ {
+ getOrgResiBuf( clippedArea ).copyFrom( subStruct.getOrgResiBuf( clippedArea ) );
+ }
+ }
+
+ if( cpyPred )
+ {
+ picture->getPredBuf( clippedArea ).copyFrom( subPredBuf );
}
- if( cpyPred ) picture->getPredBuf( clippedArea ).copyFrom( subPredBuf );
-#if JVET_AC0162_ALF_RESIDUAL_SAMPLES_INPUT
if (cpyResi)
{
picture->getResiBuf(clippedArea).copyFrom(subResiBuf);
}
+#if JVET_AC0162_ALF_RESIDUAL_SAMPLES_INPUT
else
{
picture->getResiBuf(clippedArea).copyFrom(subStruct.getResiBuf(clippedArea), true);
}
-#else
- if( cpyResi ) picture->getResiBuf( clippedArea ).copyFrom( subResiBuf );
#endif
#if JVET_Z0118_GDR
@@ -3022,17 +3034,13 @@ PelBuf CodingStructure::getBuf( const CompArea &blk, const PictureType &type )
#if !KEEP_PRED_AND_RESI_SIGNALS
#if JVET_AC0162_ALF_RESIDUAL_SAMPLES_INPUT
if (!parent && (type == PIC_PREDICTION))
- {
- cFinal.x &= (pcv->maxCUWidthMask >> getComponentScaleX(blk.compID, blk.chromaFormat));
- cFinal.y &= (pcv->maxCUHeightMask >> getComponentScaleY(blk.compID, blk.chromaFormat));
- }
#else
if( !parent && ( type == PIC_RESIDUAL || type == PIC_PREDICTION ) )
+#endif
{
cFinal.x &= ( pcv->maxCUWidthMask >> getComponentScaleX( blk.compID, blk.chromaFormat ) );
cFinal.y &= ( pcv->maxCUHeightMask >> getComponentScaleY( blk.compID, blk.chromaFormat ) );
}
-#endif
#endif
return buf->getBuf( cFinal );
@@ -3067,17 +3075,13 @@ const CPelBuf CodingStructure::getBuf( const CompArea &blk, const PictureType &t
#if !KEEP_PRED_AND_RESI_SIGNALS
#if JVET_AC0162_ALF_RESIDUAL_SAMPLES_INPUT
if (!parent && (type == PIC_PREDICTION))
- {
- cFinal.x &= (pcv->maxCUWidthMask >> getComponentScaleX(blk.compID, blk.chromaFormat));
- cFinal.y &= (pcv->maxCUHeightMask >> getComponentScaleY(blk.compID, blk.chromaFormat));
- }
#else
if( !parent && ( type == PIC_RESIDUAL || type == PIC_PREDICTION ) )
+#endif
{
cFinal.x &= ( pcv->maxCUWidthMask >> getComponentScaleX( blk.compID, blk.chromaFormat ) );
cFinal.y &= ( pcv->maxCUHeightMask >> getComponentScaleY( blk.compID, blk.chromaFormat ) );
}
-#endif
#endif
return buf->getBuf( cFinal );
diff --git a/source/Lib/CommonLib/InterPrediction.cpp b/source/Lib/CommonLib/InterPrediction.cpp
index 30bc4d6b85625318d74fb7bb712feab1c2f83d67..45fec4b89934dd3c91a197b96524d2dc7ebb972a 100644
--- a/source/Lib/CommonLib/InterPrediction.cpp
+++ b/source/Lib/CommonLib/InterPrediction.cpp
@@ -107,13 +107,13 @@ InterPrediction::InterPrediction()
, m_dIy(nullptr)
, m_dI(nullptr)
, m_signGxGy(nullptr)
-, m_tmpx_pixel_32bit(nullptr)
-, m_tmpy_pixel_32bit(nullptr)
-, m_sumAbsGX_pixel_32bit(nullptr)
-, m_sumAbsGY_pixel_32bit(nullptr)
-, m_sumDIX_pixel_32bit(nullptr)
-, m_sumDIY_pixel_32bit(nullptr)
-, m_sumSignGY_GX_pixel_32bit(nullptr)
+, m_tmpxSample32bit(nullptr)
+, m_tmpySample32bit(nullptr)
+, m_sumAbsGxSample32bit(nullptr)
+, m_sumAbsGySample32bit(nullptr)
+, m_sumDIXSample32bit(nullptr)
+, m_sumDIYSample32bit(nullptr)
+, m_sumSignGyGxSample32bit(nullptr)
#endif
, m_subPuMC(false)
{
@@ -332,15 +332,15 @@ void InterPrediction::destroy()
xFree(m_absGy); m_absGy = nullptr;
xFree(m_dIx); m_dIx = nullptr;
xFree(m_dIy); m_dIy = nullptr;
- xFree(m_dI); m_dI = nullptr;
+ xFree(m_dI); m_dI = nullptr;
xFree(m_signGxGy); m_signGxGy = nullptr;
- xFree(m_tmpx_pixel_32bit); m_tmpx_pixel_32bit = nullptr;
- xFree(m_tmpy_pixel_32bit); m_tmpy_pixel_32bit = nullptr;
- xFree(m_sumAbsGX_pixel_32bit); m_sumAbsGX_pixel_32bit = nullptr;
- xFree(m_sumAbsGY_pixel_32bit); m_sumAbsGY_pixel_32bit = nullptr;
- xFree(m_sumDIX_pixel_32bit); m_sumDIX_pixel_32bit = nullptr;
- xFree(m_sumDIY_pixel_32bit); m_sumDIY_pixel_32bit = nullptr;
- xFree(m_sumSignGY_GX_pixel_32bit); m_sumSignGY_GX_pixel_32bit = nullptr;
+ xFree(m_tmpxSample32bit); m_tmpxSample32bit = nullptr;
+ xFree(m_tmpySample32bit); m_tmpySample32bit = nullptr;
+ xFree(m_sumAbsGxSample32bit); m_sumAbsGxSample32bit = nullptr;
+ xFree(m_sumAbsGySample32bit); m_sumAbsGySample32bit = nullptr;
+ xFree(m_sumDIXSample32bit); m_sumDIXSample32bit = nullptr;
+ xFree(m_sumDIYSample32bit); m_sumDIYSample32bit = nullptr;
+ xFree(m_sumSignGyGxSample32bit); m_sumSignGyGxSample32bit = nullptr;
#endif
#if ENABLE_OBMC
m_tmpObmcBufL0.destroy();
@@ -520,13 +520,13 @@ void InterPrediction::init( RdCost* pcRdCost, ChromaFormat chromaFormatIDC, cons
m_dIy = (Pel*)xMalloc(Pel, BIO_TEMP_BUFFER_SIZE);
m_dI = (Pel*)xMalloc(Pel, BIO_TEMP_BUFFER_SIZE);
m_signGxGy = (Pel*)xMalloc(Pel, BIO_TEMP_BUFFER_SIZE);
- m_tmpx_pixel_32bit = (int*)xMalloc(int, BDOF_SUBPU_SIZE);
- m_tmpy_pixel_32bit = (int*)xMalloc(int, BDOF_SUBPU_SIZE);
- m_sumAbsGX_pixel_32bit = (int*)xMalloc(int, BDOF_SUBPU_SIZE);
- m_sumAbsGY_pixel_32bit = (int*)xMalloc(int, BDOF_SUBPU_SIZE);
- m_sumDIX_pixel_32bit = (int*)xMalloc(int, BDOF_SUBPU_SIZE);
- m_sumDIY_pixel_32bit = (int*)xMalloc(int, BDOF_SUBPU_SIZE);
- m_sumSignGY_GX_pixel_32bit = (int*)xMalloc(int, BDOF_SUBPU_SIZE);
+ m_tmpxSample32bit = (int*)xMalloc(int, BDOF_SUBPU_SIZE);
+ m_tmpySample32bit = (int*)xMalloc(int, BDOF_SUBPU_SIZE);
+ m_sumAbsGxSample32bit = (int*)xMalloc(int, BDOF_SUBPU_SIZE);
+ m_sumAbsGySample32bit = (int*)xMalloc(int, BDOF_SUBPU_SIZE);
+ m_sumDIXSample32bit = (int*)xMalloc(int, BDOF_SUBPU_SIZE);
+ m_sumDIYSample32bit = (int*)xMalloc(int, BDOF_SUBPU_SIZE);
+ m_sumSignGyGxSample32bit = (int*)xMalloc(int, BDOF_SUBPU_SIZE);
#endif
#if ENABLE_OBMC
m_tmpObmcBufL0.create(UnitArea(chromaFormatIDC, Area(0, 0, 4, MAX_CU_SIZE)));
@@ -1097,8 +1097,10 @@ void InterPrediction::xPredInterBiSubPuBDOF(PredictionUnit &pu, PelUnitBuf &pcYu
if (bioMv.hor == 0 && bioMv.ver == 0)
{
// only chroma MC
- if (!lumaOnly)
- xPredInterUni ( subPu, eRefPicList, pcMbBuf, true, bioApplied, false, chroma, false );
+ if( !lumaOnly )
+ {
+ xPredInterUni( subPu, eRefPicList, pcMbBuf, true, bioApplied, false, chroma, false );
+ }
}
else
{
@@ -1170,7 +1172,9 @@ void InterPrediction::xPredInterBiSubPuBDOF(PredictionUnit &pu, PelUnitBuf &pcYu
}
#else
if (!lumaOnly)
+ {
xWeightedAverage( false/*isBdofMvRefine*/, 0/*bdofBlockOffset*/, subPu, srcSubPred0, srcSubPred1, subYuvPredBuf, slice.getSPS()->getBitDepths(), slice.clpRngs(), false/*bioApplied*/, lumaOnly, true/*chromaOnly*/, NULL/*yuvPredTmp*/ );
+ }
#endif
}
else
@@ -3617,26 +3621,35 @@ void InterPrediction::applyBiOptFlow(const PredictionUnit &pu, const CPelUnitBuf
continue;
}
- int sumAbsGX_block = 0, sumAbsGY_block = 0, sumDIX_block = 0, sumDIY_block = 0, sumSignGY_GX_block = 0;
+ int sumAbsGxBlock = 0, sumAbsGyBlock = 0, sumDIXblock = 0, sumDIYblock = 0, sumSignGyGxBlock = 0;
g_pelBufOP.calcBIOParamSum4(m_absGx + bioBlockParamOffset, m_absGy + bioBlockParamOffset, m_dIx + bioBlockParamOffset,
m_dIy + bioBlockParamOffset, m_signGxGy + bioBlockParamOffset, bioDx + 4, bioDy + 4, widthG,
- &sumAbsGX_block, &sumAbsGY_block, &sumDIX_block, &sumDIY_block, &sumSignGY_GX_block);
+ &sumAbsGxBlock, &sumAbsGyBlock, &sumDIXblock, &sumDIYblock, &sumSignGyGxBlock);
- int tmpx_block = (sumAbsGX_block == 0 ? 0 : rightShiftMSB(sumDIX_block << 3, sumAbsGX_block));
- int tmpData_block = ((tmpx_block * sumSignGY_GX_block) >> 1);
- int tmpy_block = (sumAbsGY_block == 0 ? 0 : rightShiftMSB(((sumDIY_block << 3) - tmpData_block), sumAbsGY_block));
- tmpx_block = Clip3(-256, 256, tmpx_block);
- tmpy_block = Clip3(-256, 256, tmpy_block);
+ int tmpXblock = (sumAbsGxBlock == 0 ? 0 : rightShiftMSB(sumDIXblock << 3, sumAbsGxBlock));
+ int tmpDataBlock = ((tmpXblock * sumSignGyGxBlock) >> 1);
+ int tmpYblock = (sumAbsGyBlock == 0 ? 0 : rightShiftMSB(((sumDIYblock << 3) - tmpDataBlock), sumAbsGyBlock));
+ tmpXblock = Clip3(-256, 256, tmpXblock);
+ tmpYblock = Clip3(-256, 256, tmpYblock);
Mv bioMv;
- if (tmpx_block >= 0)
- bioMv.hor = ((tmpx_block + 4) >> 3);
+ if( tmpXblock >= 0 )
+ {
+ bioMv.hor = ((tmpXblock + 4) >> 3);
+ }
else
- bioMv.hor = (-1) * ((((-1) * tmpx_block) + 4) >> 3);
- if (tmpy_block >= 0)
- bioMv.ver = ((tmpy_block + 4) >> 3);
+ {
+ bioMv.hor = (-1) * ((((-1) * tmpXblock) + 4) >> 3);
+ }
+
+ if( tmpYblock >= 0 )
+ {
+ bioMv.ver = ((tmpYblock + 4) >> 3);
+ }
else
- bioMv.ver = (-1) * ((((-1) * tmpy_block) + 4) >> 3);
+ {
+ bioMv.ver = (-1) * ((((-1) * tmpYblock) + 4) >> 3);
+ }
m_bdofSubPuMvOffset[bdofBlockOffset + bioSubPuMvIndex] = bioMv;
if (bioMv.hor == 0 && bioMv.ver == 0)
@@ -3734,9 +3747,9 @@ void InterPrediction::applyBiOptFlow(const PredictionUnit &pu, const CPelUnitBuf
gradY0 = m_gradY0; gradY1 = m_gradY1;
Pel *pGradX0Tmp, *pGradX1Tmp, *pGradY0Tmp, *pGradY1Tmp;
- const Pel *SrcY0Tmp, *SrcY1Tmp;
+ const Pel *srcY0Tmp, *srcY1Tmp;
int tmpx = 0, tmpy = 0;
- int sumAbsGX = 0, sumAbsGY = 0, sumDIX = 0, sumDIY = 0, sumSignGY_GX = 0;
+ int sumAbsGX = 0, sumAbsGY = 0, sumDIX = 0, sumDIY = 0, sumSignGyGx = 0;
int gradOfst, srcOfst, dstOfst, gradLineOfst = 0, srcLineOfst = 0, dstLineOfst = 0;
for (int yu = 0; yu < yUnit; yu++)
@@ -3747,27 +3760,27 @@ void InterPrediction::applyBiOptFlow(const PredictionUnit &pu, const CPelUnitBuf
for (int xu = 0; xu < xUnit; xu++)
{
- sumAbsGX = 0; sumAbsGY = 0; sumDIX = 0; sumDIY = 0, sumSignGY_GX = 0;
+ sumAbsGX = 0; sumAbsGY = 0; sumDIX = 0; sumDIY = 0, sumSignGyGx = 0;
pGradX0Tmp = m_gradX0 + gradOfst;
pGradX1Tmp = m_gradX1 + gradOfst;
pGradY0Tmp = m_gradY0 + gradOfst;
pGradY1Tmp = m_gradY1 + gradOfst;
- SrcY1Tmp = srcY1 + srcOfst;
- SrcY0Tmp = srcY0 + srcOfst;
+ srcY1Tmp = srcY1 + srcOfst;
+ srcY0Tmp = srcY0 + srcOfst;
- g_pelBufOP.calcBIOSums(SrcY0Tmp, SrcY1Tmp, pGradX0Tmp, pGradX1Tmp, pGradY0Tmp, pGradY1Tmp, xu, yu, src0Stride, src1Stride, widthG, bitDepth, &sumAbsGX, &sumAbsGY, &sumDIX, &sumDIY, &sumSignGY_GX);
+ g_pelBufOP.calcBIOSums(srcY0Tmp, srcY1Tmp, pGradX0Tmp, pGradX1Tmp, pGradY0Tmp, pGradY1Tmp, xu, yu, src0Stride, src1Stride, widthG, bitDepth, &sumAbsGX, &sumAbsGY, &sumDIX, &sumDIY, &sumSignGyGx);
tmpx = (sumAbsGX == 0 ? 0 : rightShiftMSB(sumDIX << 2, sumAbsGX));
tmpx = Clip3(-limit, limit, tmpx);
- int mainsGxGy = sumSignGY_GX >> 12;
- int secsGxGy = sumSignGY_GX & ((1 << 12) - 1);
+ int mainsGxGy = sumSignGyGx >> 12;
+ int secsGxGy = sumSignGyGx & ((1 << 12) - 1);
int tmpData = tmpx * mainsGxGy;
tmpData = ((tmpData << 12) + tmpx*secsGxGy) >> 1;
tmpy = (sumAbsGY == 0 ? 0 : rightShiftMSB(((sumDIY << 2) - tmpData), sumAbsGY));
tmpy = Clip3(-limit, limit, tmpy);
- srcY0Temp = SrcY0Tmp + ( stridePredMC + 1 );
- srcY1Temp = SrcY1Tmp + ( stridePredMC + 1 );
+ srcY0Temp = srcY0Tmp + ( stridePredMC + 1 );
+ srcY1Temp = srcY1Tmp + ( stridePredMC + 1 );
gradX0 = pGradX0Tmp + offsetPos;
gradX1 = pGradX1Tmp + offsetPos;
gradY0 = pGradY0Tmp + offsetPos;
@@ -3797,32 +3810,32 @@ void InterPrediction::subBlockBiOptFlow(Pel* dstY, const int dstStride, const Pe
#if SAMPLE_BASED_BDOF
g_pelBufOP.calcBIOParamSum5(m_absGx + bioParamOffset, m_absGy + bioParamOffset, m_dIx + bioParamOffset,
m_dIy + bioParamOffset, m_signGxGy + bioParamOffset, bioParamStride, width, height,
- m_sumAbsGX_pixel_32bit, m_sumAbsGY_pixel_32bit, m_sumDIX_pixel_32bit, m_sumDIY_pixel_32bit, m_sumSignGY_GX_pixel_32bit);
+ m_sumAbsGxSample32bit, m_sumAbsGySample32bit, m_sumDIXSample32bit, m_sumDIYSample32bit, m_sumSignGyGxSample32bit);
// sumDIX and sumDIY left shift by 2 is calculated in previous step
const int bioSubblockSize = width * height;
- for (int pixel_index = 0; pixel_index < bioSubblockSize; pixel_index++)
+ for (int sampleIndex = 0; sampleIndex < bioSubblockSize; sampleIndex++)
{
- if (m_sumAbsGX_pixel_32bit[pixel_index] == 0)
+ if (m_sumAbsGxSample32bit[sampleIndex] == 0)
{
- m_sumDIX_pixel_32bit[pixel_index] = 0;
- m_sumAbsGX_pixel_32bit[pixel_index] = 32;
+ m_sumDIXSample32bit[sampleIndex] = 0;
+ m_sumAbsGxSample32bit[sampleIndex] = 32;
}
else
{
- m_sumAbsGX_pixel_32bit[pixel_index] = floorLog2(m_sumAbsGX_pixel_32bit[pixel_index]);
+ m_sumAbsGxSample32bit[sampleIndex] = floorLog2(m_sumAbsGxSample32bit[sampleIndex]);
}
- if (m_sumAbsGY_pixel_32bit[pixel_index] == 0)
+ if (m_sumAbsGySample32bit[sampleIndex] == 0)
{
- m_sumDIY_pixel_32bit[pixel_index] = 0;
- m_sumSignGY_GX_pixel_32bit[pixel_index] = 0;
- m_sumAbsGY_pixel_32bit[pixel_index] = 32;
+ m_sumDIYSample32bit[sampleIndex] = 0;
+ m_sumSignGyGxSample32bit[sampleIndex] = 0;
+ m_sumAbsGySample32bit[sampleIndex] = 32;
}
else
{
- m_sumAbsGY_pixel_32bit[pixel_index] = floorLog2(m_sumAbsGY_pixel_32bit[pixel_index]);
+ m_sumAbsGySample32bit[sampleIndex] = floorLog2(m_sumAbsGySample32bit[sampleIndex]);
}
}
- g_pelBufOP.calcBIOClippedVxVy(m_sumDIX_pixel_32bit, m_sumAbsGX_pixel_32bit, m_sumDIY_pixel_32bit, m_sumAbsGY_pixel_32bit, m_sumSignGY_GX_pixel_32bit, limit, bioSubblockSize, m_tmpx_pixel_32bit, m_tmpy_pixel_32bit);
+ g_pelBufOP.calcBIOClippedVxVy(m_sumDIXSample32bit, m_sumAbsGxSample32bit, m_sumDIYSample32bit, m_sumAbsGySample32bit, m_sumSignGyGxSample32bit, limit, bioSubblockSize, m_tmpxSample32bit, m_tmpySample32bit);
bioParamOffset += ((bioParamStride + 1) << 1);
#else
bioParamOffset += ((bioParamStride + 1) << 1);
@@ -3832,7 +3845,7 @@ void InterPrediction::subBlockBiOptFlow(Pel* dstY, const int dstStride, const Pe
for (int xUnit = 0; xUnit < width; xUnit += unitSize)
{
int subTmpx = 0, subTmpy = 0;
- int subSumGx = 0, subSumGy = 0, subSumDIX = 0, subSumDIY = 0, subSumSignGY_GX = 0;
+ int subSumGx = 0, subSumGy = 0, subSumDIX = 0, subSumDIY = 0, subSumSignGyGx = 0;
int subBioParamOffset = bioParamOffset + (yUnit - extendSize) * bioParamStride + xUnit;
for (int ySub = -extendSize; ySub < (extendSize + unitSize); ySub++)
{
@@ -3842,16 +3855,16 @@ void InterPrediction::subBlockBiOptFlow(Pel* dstY, const int dstStride, const Pe
subSumGy += m_absGy[subBioParamOffset + xSub];
subSumDIX += m_dIx[subBioParamOffset + xSub];
subSumDIY += m_dIy[subBioParamOffset + xSub];
- subSumSignGY_GX += m_signGxGy[subBioParamOffset + xSub];
+ subSumSignGyGx += m_signGxGy[subBioParamOffset + xSub];
}
subBioParamOffset += bioParamStride;
}
subTmpx = (subSumGx == 0 ? 0 : rightShiftMSB(subSumDIX << 2, subSumGx));
subTmpx = Clip3(-limit, limit, subTmpx);
- int mainsGxGy = subSumSignGY_GX >> 12;
- int secsGxGy = subSumSignGY_GX & ((1 << 12) - 1);
- int tmpData = subTmpx * mainsGxGy;
+ int mainsGxGy = subSumSignGyGx >> 12;
+ int secsGxGy = subSumSignGyGx & ((1 << 12) - 1);
+ int tmpData = subTmpx * mainsGxGy;
tmpData = ((tmpData << 12) + subTmpx*secsGxGy) >> 1;
subTmpy = (subSumGy == 0 ? 0 : rightShiftMSB(((subSumDIY << 2) - tmpData), subSumGy));
subTmpy = Clip3(-limit, limit, subTmpy);
@@ -3860,8 +3873,8 @@ void InterPrediction::subBlockBiOptFlow(Pel* dstY, const int dstStride, const Pe
{
for (int xSub = 0; xSub < unitSize; xSub++)
{
- m_tmpx_pixel_32bit[curSubIdx + xSub] = subTmpx;
- m_tmpy_pixel_32bit[curSubIdx + xSub] = subTmpy;
+ m_tmpxSample32bit[curSubIdx + xSub] = subTmpx;
+ m_tmpySample32bit[curSubIdx + xSub] = subTmpy;
}
curSubIdx += width;
}
@@ -3869,9 +3882,9 @@ void InterPrediction::subBlockBiOptFlow(Pel* dstY, const int dstStride, const Pe
}
#endif
#if JVET_Z0136_OOB
- g_pelBufOP.addBIOAvgN(src0, src0Stride, src1, src1Stride, dstY, dstStride, m_gradX0 + bioParamOffset, m_gradX1 + bioParamOffset, m_gradY0 + bioParamOffset, m_gradY1 + bioParamOffset, bioParamStride, width, height, m_tmpx_pixel_32bit, m_tmpy_pixel_32bit, shiftNum, offset, clpRng, mcMask, mcStride, isOOB);
+ g_pelBufOP.addBIOAvgN(src0, src0Stride, src1, src1Stride, dstY, dstStride, m_gradX0 + bioParamOffset, m_gradX1 + bioParamOffset, m_gradY0 + bioParamOffset, m_gradY1 + bioParamOffset, bioParamStride, width, height, m_tmpxSample32bit, m_tmpySample32bit, shiftNum, offset, clpRng, mcMask, mcStride, isOOB);
#else
- g_pelBufOP.addBIOAvgN(src0, src0Stride, src1, src1Stride, dstY, dstStride, m_gradX0 + bioParamOffset, m_gradX1 + bioParamOffset, m_gradY0 + bioParamOffset, m_gradY1 + bioParamOffset, bioParamStride, width, height, m_tmpx_pixel_32bit, m_tmpy_pixel_32bit, shiftNum, offset, clpRng);
+ g_pelBufOP.addBIOAvgN(src0, src0Stride, src1, src1Stride, dstY, dstStride, m_gradX0 + bioParamOffset, m_gradX1 + bioParamOffset, m_gradY0 + bioParamOffset, m_gradY1 + bioParamOffset, bioParamStride, width, height, m_tmpxSample32bit, m_tmpySample32bit, shiftNum, offset, clpRng);
#endif
}
#endif
diff --git a/source/Lib/CommonLib/InterPrediction.h b/source/Lib/CommonLib/InterPrediction.h
index b92d16c9f214ced4cbf288f1893deb0b56d315f4..42925e79b53d81b82ee95155b45755f71bb1307b 100644
--- a/source/Lib/CommonLib/InterPrediction.h
+++ b/source/Lib/CommonLib/InterPrediction.h
@@ -219,13 +219,13 @@ protected:
Pel* m_dIy;
Pel* m_dI;
Pel* m_signGxGy;
- int* m_tmpx_pixel_32bit;
- int* m_tmpy_pixel_32bit;
- int* m_sumAbsGX_pixel_32bit;
- int* m_sumAbsGY_pixel_32bit;
- int* m_sumDIX_pixel_32bit;
- int* m_sumDIY_pixel_32bit;
- int* m_sumSignGY_GX_pixel_32bit;
+ int* m_tmpxSample32bit;
+ int* m_tmpySample32bit;
+ int* m_sumAbsGxSample32bit;
+ int* m_sumAbsGySample32bit;
+ int* m_sumDIXSample32bit;
+ int* m_sumDIYSample32bit;
+ int* m_sumSignGyGxSample32bit;
bool m_bdofMvRefined;
Mv m_bdofSubPuMvOffset[BDOF_SUBPU_MAX_NUM];
#endif
diff --git a/source/Lib/CommonLib/IntraPrediction.cpp b/source/Lib/CommonLib/IntraPrediction.cpp
index 29ed7002264ce64d3e86462f48dd2c50d58cab10..28292146af4001c75a972b18c4fe81929013fa56 100644
--- a/source/Lib/CommonLib/IntraPrediction.cpp
+++ b/source/Lib/CommonLib/IntraPrediction.cpp
@@ -1841,7 +1841,7 @@ void IntraPrediction::predIntraAng( const ComponentID compId, PelBuf &piPred, co
}
#if JVET_AC0071_DBV
-void IntraPrediction::PredIntraDbv(const ComponentID compId, PelBuf &piPred, const PredictionUnit &pu)
+void IntraPrediction::predIntraDbv(const ComponentID compId, PelBuf &piPred, const PredictionUnit &pu)
{
const int shiftSampleHor = ::getComponentScaleX(compId, pu.chromaFormat);
const int shiftSampleVer = ::getComponentScaleY(compId, pu.chromaFormat);
diff --git a/source/Lib/CommonLib/IntraPrediction.h b/source/Lib/CommonLib/IntraPrediction.h
index 9f0e06831cf81416b951691ff5cd4296029067db..179517a9bb4038d37548db4bfec444f4b51b5acc 100644
--- a/source/Lib/CommonLib/IntraPrediction.h
+++ b/source/Lib/CommonLib/IntraPrediction.h
@@ -580,7 +580,7 @@ public:
#if JVET_AC0071_DBV
// Direct Block Vector
- void PredIntraDbv(const ComponentID compId, PelBuf &piPred, const PredictionUnit &pu);
+ void predIntraDbv(const ComponentID compId, PelBuf &piPred, const PredictionUnit &pu);
Mv refineChromaBv(const ComponentID compId, const PredictionUnit &pu);
#endif
diff --git a/source/Lib/CommonLib/Picture.cpp b/source/Lib/CommonLib/Picture.cpp
index d30510dad252454ceb84b315f5f01ac3167f37bd..7c109be0274589505d624ed96ab629f2c459af5c 100644
--- a/source/Lib/CommonLib/Picture.cpp
+++ b/source/Lib/CommonLib/Picture.cpp
@@ -297,6 +297,9 @@ void Picture::createTempBuffers( const unsigned _maxCUSize )
const Area a( Position{ 0, 0 }, lumaSize() );
#else
const Area a = m_ctuArea.Y();
+#if JVET_AC0162_ALF_RESIDUAL_SAMPLES_INPUT
+ const Area aOld( Position{ 0, 0 }, lumaSize() );
+#endif
#endif
#if ENABLE_SPLIT_PARALLELISM
@@ -306,9 +309,8 @@ void Picture::createTempBuffers( const unsigned _maxCUSize )
#endif
{
M_BUFS( jId, PIC_PREDICTION ).create( chromaFormat, a, _maxCUSize );
-#if JVET_AC0162_ALF_RESIDUAL_SAMPLES_INPUT
- const Area aOld(Position{ 0, 0 }, lumaSize());
- M_BUFS(jId, PIC_RESIDUAL).create(chromaFormat, aOld, _maxCUSize);
+#if JVET_AC0162_ALF_RESIDUAL_SAMPLES_INPUT && !KEEP_PRED_AND_RESI_SIGNALS
+ M_BUFS( jId, PIC_RESIDUAL ).create(chromaFormat, aOld, _maxCUSize );
#else
M_BUFS( jId, PIC_RESIDUAL ).create( chromaFormat, a, _maxCUSize );
#endif
@@ -1532,15 +1534,9 @@ PelBuf Picture::getBuf( const CompArea &blk, const PictureType &type )
#if !KEEP_PRED_AND_RESI_SIGNALS
#if JVET_AC0162_ALF_RESIDUAL_SAMPLES_INPUT
if (type == PIC_PREDICTION)
- {
- CompArea localBlk = blk;
- localBlk.x &= (cs->pcv->maxCUWidthMask >> getComponentScaleX(blk.compID, blk.chromaFormat));
- localBlk.y &= (cs->pcv->maxCUHeightMask >> getComponentScaleY(blk.compID, blk.chromaFormat));
-
- return M_BUFS(jId, type).getBuf(localBlk);
- }
#else
if( type == PIC_RESIDUAL || type == PIC_PREDICTION )
+#endif
{
CompArea localBlk = blk;
localBlk.x &= ( cs->pcv->maxCUWidthMask >> getComponentScaleX( blk.compID, blk.chromaFormat ) );
@@ -1548,7 +1544,6 @@ PelBuf Picture::getBuf( const CompArea &blk, const PictureType &type )
return M_BUFS( jId, type ).getBuf( localBlk );
}
-#endif
#endif
return M_BUFS( jId, type ).getBuf( blk );
@@ -1568,15 +1563,9 @@ const CPelBuf Picture::getBuf( const CompArea &blk, const PictureType &type ) co
#if !KEEP_PRED_AND_RESI_SIGNALS
#if JVET_AC0162_ALF_RESIDUAL_SAMPLES_INPUT
if (type == PIC_PREDICTION)
- {
- CompArea localBlk = blk;
- localBlk.x &= (cs->pcv->maxCUWidthMask >> getComponentScaleX(blk.compID, blk.chromaFormat));
- localBlk.y &= (cs->pcv->maxCUHeightMask >> getComponentScaleY(blk.compID, blk.chromaFormat));
-
- return M_BUFS(jId, type).getBuf(localBlk);
- }
#else
if( type == PIC_RESIDUAL || type == PIC_PREDICTION )
+#endif
{
CompArea localBlk = blk;
localBlk.x &= ( cs->pcv->maxCUWidthMask >> getComponentScaleX( blk.compID, blk.chromaFormat ) );
@@ -1584,7 +1573,6 @@ const CPelBuf Picture::getBuf( const CompArea &blk, const PictureType &type ) co
return M_BUFS( jId, type ).getBuf( localBlk );
}
-#endif
#endif
return M_BUFS( jId, type ).getBuf( blk );
diff --git a/source/Lib/CommonLib/x86/BufferX86.h b/source/Lib/CommonLib/x86/BufferX86.h
index 61d74d3618355a9df08dbcaf83e5097d2c794a94..ff94e11368f5769c88fdb3bf3c6b5bd08aaf8be9 100644
--- a/source/Lib/CommonLib/x86/BufferX86.h
+++ b/source/Lib/CommonLib/x86/BufferX86.h
@@ -579,7 +579,7 @@ void addBIOAvg4_SSE(const Pel* src0, int src0Stride, const Pel* src1, int src1St
#if MULTI_PASS_DMVR || SAMPLE_BASED_BDOF
template< X86_VEXT vext >
-void calcBIOParameter_SSE(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0, Pel* gradX1, Pel* gradY0, Pel* gradY1, int width, int height, const int src0Stride, const int src1Stride, const int widthG, const int bitDepth, Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGY_GX, Pel* dI)
+void calcBIOParameter_SSE(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0, Pel* gradX1, Pel* gradY0, Pel* gradY1, int width, int height, const int src0Stride, const int src1Stride, const int widthG, const int bitDepth, Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGyGx, Pel* dI)
{
width -= 2;
height -= 2;
@@ -590,7 +590,7 @@ void calcBIOParameter_SSE(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0,
gradY0 += bioParamOffset; gradY1 += bioParamOffset;
absGX += bioParamOffset; absGY += bioParamOffset;
dIX += bioParamOffset; dIY += bioParamOffset;
- signGY_GX += bioParamOffset;
+ signGyGx += bioParamOffset;
if (dI)
{
dI += bioParamOffset;
@@ -619,13 +619,13 @@ void calcBIOParameter_SSE(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0,
}
__m128i dIX_tmp = _mm_sign_epi16(subTemp1, packTempX );
__m128i dIY_tmp = _mm_sign_epi16(subTemp1, packTempY );
- __m128i signGY_GX_tmp = _mm_sign_epi16(packTempX, packTempY );
+ __m128i signGyGxTmp = _mm_sign_epi16(packTempX, packTempY );
_mm_storeu_si128( ( __m128i * )absGX, gX_tmp );
_mm_storeu_si128( ( __m128i * )absGY, gY_tmp );
_mm_storeu_si128( ( __m128i * )dIX, dIX_tmp );
_mm_storeu_si128( ( __m128i * )dIY, dIY_tmp );
- _mm_storeu_si128( ( __m128i * )signGY_GX, signGY_GX_tmp );
+ _mm_storeu_si128( ( __m128i * )signGyGx, signGyGxTmp );
srcY0Tmp += src0Stride;
srcY1Tmp += src1Stride;
gradX0 += widthG;
@@ -640,7 +640,7 @@ void calcBIOParameter_SSE(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0,
}
dIX += widthG;
dIY += widthG;
- signGY_GX += widthG;
+ signGyGx += widthG;
}
}
else // width = 12, 20, 36, 68, 132, 260
@@ -666,13 +666,13 @@ void calcBIOParameter_SSE(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0,
}
__m256i dIX_tmp = _mm256_sign_epi16(subTemp1, packTempX );
__m256i dIY_tmp = _mm256_sign_epi16(subTemp1, packTempY );
- __m256i signGY_GX_tmp = _mm256_sign_epi16(packTempX, packTempY );
+ __m256i signGyGxTmp = _mm256_sign_epi16(packTempX, packTempY );
_mm256_storeu_si256( ( __m256i * ) ( absGX + x ), gX_tmp );
_mm256_storeu_si256( ( __m256i * ) ( absGY + x ), gY_tmp );
_mm256_storeu_si256( ( __m256i * ) ( dIX + x ), dIX_tmp );
_mm256_storeu_si256( ( __m256i * ) ( dIY + x ), dIY_tmp );
- _mm256_storeu_si256( ( __m256i * ) ( signGY_GX + x ), signGY_GX_tmp );
+ _mm256_storeu_si256( ( __m256i * ) ( signGyGx + x ), signGyGxTmp );
}
srcY0Tmp += src0Stride;
srcY1Tmp += src1Stride;
@@ -688,7 +688,7 @@ void calcBIOParameter_SSE(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0,
}
dIX += widthG;
dIY += widthG;
- signGY_GX += widthG;
+ signGyGx += widthG;
}
}
#else
@@ -713,13 +713,13 @@ void calcBIOParameter_SSE(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0,
}
__m128i dIX_tmp = _mm_sign_epi16(subTemp1, packTempX );
__m128i dIY_tmp = _mm_sign_epi16(subTemp1, packTempY );
- __m128i signGY_GX_tmp = _mm_sign_epi16(packTempX, packTempY );
+ __m128i signGyGxTmp = _mm_sign_epi16(packTempX, packTempY );
_mm_storeu_si128( ( __m128i * ) ( absGX + x ), gX_tmp );
_mm_storeu_si128( ( __m128i * ) ( absGY + x ), gY_tmp );
_mm_storeu_si128( ( __m128i * ) ( dIX + x ), dIX_tmp );
_mm_storeu_si128( ( __m128i * ) ( dIY + x ), dIY_tmp );
- _mm_storeu_si128( ( __m128i * ) ( signGY_GX + x ), signGY_GX_tmp );
+ _mm_storeu_si128( ( __m128i * ) ( signGyGx + x ), signGyGxTmp );
}
srcY0Tmp += src0Stride;
srcY1Tmp += src1Stride;
@@ -735,12 +735,12 @@ void calcBIOParameter_SSE(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0,
{
dI += widthG;
}
- signGY_GX += widthG;
+ signGyGx += widthG;
}
#endif
}
template< X86_VEXT vext >
-void calcBIOParamSum5_SSE(Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGY_GX, const int widthG, const int width, const int height, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGY_GX)
+void calcBIOParamSum5_SSE(Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGyGx, const int widthG, const int width, const int height, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGyGx)
{
__m128i vzero = _mm_setzero_si128();
__m128i vmask = _mm_setr_epi16(1, 1, 1, 1, 1, 0, 0, 0);
@@ -782,9 +782,9 @@ void calcBIOParamSum5_SSE(Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signG
sumDIYTmp16 = _mm_add_epi16(_mm_loadu_si128((const __m128i*)dIY), _mm_loadu_si128((const __m128i*)(dIY + widthG)));
sumDIYTmp16 = _mm_add_epi16(sumDIYTmp16, _mm_loadu_si128((const __m128i*)(dIY + widthG_2)));
sumDIYTmp16 = _mm_add_epi16(sumDIYTmp16, _mm_loadu_si128((const __m128i*)(dIY + widthG_3)));
- sumSignGyGxTmp16 = _mm_add_epi16(_mm_loadu_si128((const __m128i*)signGY_GX), _mm_loadu_si128((const __m128i*)(signGY_GX + widthG)));
- sumSignGyGxTmp16 = _mm_add_epi16(sumSignGyGxTmp16, _mm_loadu_si128((const __m128i*)(signGY_GX + widthG_2)));
- sumSignGyGxTmp16 = _mm_add_epi16(sumSignGyGxTmp16, _mm_loadu_si128((const __m128i*)(signGY_GX + widthG_3)));
+ sumSignGyGxTmp16 = _mm_add_epi16(_mm_loadu_si128((const __m128i*)signGyGx), _mm_loadu_si128((const __m128i*)(signGyGx + widthG)));
+ sumSignGyGxTmp16 = _mm_add_epi16(sumSignGyGxTmp16, _mm_loadu_si128((const __m128i*)(signGyGx + widthG_2)));
+ sumSignGyGxTmp16 = _mm_add_epi16(sumSignGyGxTmp16, _mm_loadu_si128((const __m128i*)(signGyGx + widthG_3)));
__m128i absGXOneRow = vzero;
__m128i dIXOneRow = vzero;
@@ -806,8 +806,8 @@ void calcBIOParamSum5_SSE(Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signG
dIYOneRow = _mm_loadu_si128((const __m128i*)dIY);
sumDIYTmp16 = _mm_add_epi16(sumDIYTmp16, _mm_loadu_si128((const __m128i*)(dIY + widthG_4)));
sumSignGyGxTmp16 = _mm_sub_epi16(sumSignGyGxTmp16, signGyGxOneRow);
- signGyGxOneRow = _mm_loadu_si128((const __m128i*)signGY_GX);
- sumSignGyGxTmp16 = _mm_add_epi16(sumSignGyGxTmp16, _mm_loadu_si128((const __m128i*)(signGY_GX+ widthG_4)));
+ signGyGxOneRow = _mm_loadu_si128((const __m128i*)signGyGx);
+ sumSignGyGxTmp16 = _mm_add_epi16(sumSignGyGxTmp16, _mm_loadu_si128((const __m128i*)(signGyGx+ widthG_4)));
sumAbsGXTmp32 = _mm_madd_epi16(sumAbsGXTmp16, vmask);
sumAbsGYTmp32 = _mm_madd_epi16(sumAbsGYTmp16, vmask);
@@ -830,33 +830,33 @@ void calcBIOParamSum5_SSE(Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signG
sumSignGyGxTmp32 = _mm_madd_epi16(sumSignGyGxTmp16, vmask);
sumSignGyGxTmp32 = _mm_add_epi32(sumSignGyGxTmp32, _mm_shuffle_epi32(sumSignGyGxTmp32, 0x4e)); // 01001110
sumSignGyGxTmp32 = _mm_add_epi32(sumSignGyGxTmp32, _mm_shuffle_epi32(sumSignGyGxTmp32, 0xb1)); // 10110001
- *sumSignGY_GX = _mm_cvtsi128_si32(sumSignGyGxTmp32);
+ *sumSignGyGx = _mm_cvtsi128_si32(sumSignGyGxTmp32);
// bio parameter increment
absGX += widthG;
absGY += widthG;
dIX += widthG;
dIY += widthG;
- signGY_GX += widthG;
+ signGyGx += widthG;
// sum parameter increment
sumAbsGX += width;
sumAbsGY += width;
sumDIX += width;
sumDIY += width;
- sumSignGY_GX += width;
+ sumSignGyGx += width;
}
// bio parameter back to first row
absGX += widthG_N;
absGY += widthG_N;
dIX += widthG_N;
dIY += widthG_N;
- signGY_GX += widthG_N;
+ signGyGx += widthG_N;
// sum parameter back to first row
sumAbsGX += width_N;
sumAbsGY += width_N;
sumDIX += width_N;
sumDIY += width_N;
- sumSignGY_GX += width_N;
+ sumSignGyGx += width_N;
}
sumDIX -= width;
sumDIY -= width;
@@ -883,7 +883,7 @@ void calcBIOParamSum5_SSE(Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signG
#endif
}
template< X86_VEXT vext >
-void calcBIOParamSum4_SSE(Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGY_GX, int width, int height, const int widthG, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGY_GX)
+void calcBIOParamSum4_SSE(Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signGyGx, int width, int height, const int widthG, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGyGx)
{
__m128i vzero = _mm_setzero_si128();
@@ -900,13 +900,13 @@ void calcBIOParamSum4_SSE(Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signG
sumDIXTmp16 = _mm_add_epi16(sumDIXTmp16, _mm_loadu_si128((const __m128i*)dIX));
sumAbsGYTmp16 = _mm_add_epi16(sumAbsGYTmp16, _mm_loadu_si128((const __m128i*)absGY));
sumDIYTmp16 = _mm_add_epi16(sumDIYTmp16, _mm_loadu_si128((const __m128i*)dIY));
- sumSignGyGxTmp16 = _mm_add_epi16(sumSignGyGxTmp16, _mm_loadu_si128((const __m128i*)signGY_GX));
+ sumSignGyGxTmp16 = _mm_add_epi16(sumSignGyGxTmp16, _mm_loadu_si128((const __m128i*)signGyGx));
// bio parameter increment
absGX += widthG;
absGY += widthG;
dIX += widthG;
dIY += widthG;
- signGY_GX += widthG;
+ signGyGx += widthG;
}
}
else // (width == 12)
@@ -921,14 +921,14 @@ void calcBIOParamSum4_SSE(Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signG
sumAbsGYTmp16 = _mm_add_epi16(sumAbsGYTmp16, _mm_loadl_epi64((const __m128i*)(absGY + 8)));
sumDIYTmp16 = _mm_add_epi16(sumDIYTmp16, _mm_loadu_si128((const __m128i*)dIY));
sumDIYTmp16 = _mm_add_epi16(sumDIYTmp16, _mm_loadl_epi64((const __m128i*)(dIY + 8)));
- sumSignGyGxTmp16 = _mm_add_epi16(sumSignGyGxTmp16, _mm_loadu_si128((const __m128i*)signGY_GX));
- sumSignGyGxTmp16 = _mm_add_epi16(sumSignGyGxTmp16, _mm_loadl_epi64((const __m128i*)(signGY_GX + 8)));
+ sumSignGyGxTmp16 = _mm_add_epi16(sumSignGyGxTmp16, _mm_loadu_si128((const __m128i*)signGyGx));
+ sumSignGyGxTmp16 = _mm_add_epi16(sumSignGyGxTmp16, _mm_loadl_epi64((const __m128i*)(signGyGx + 8)));
// bio parameter increment
absGX += widthG;
absGY += widthG;
dIX += widthG;
dIY += widthG;
- signGY_GX += widthG;
+ signGyGx += widthG;
}
}
@@ -954,11 +954,11 @@ void calcBIOParamSum4_SSE(Pel* absGX, Pel* absGY, Pel* dIX, Pel* dIY, Pel* signG
__m128i sumSignGyGxTmp32 = _mm_madd_epi16(sumSignGyGxTmp16, _mm_set1_epi16(1));
sumSignGyGxTmp32 = _mm_add_epi32(sumSignGyGxTmp32, _mm_shuffle_epi32(sumSignGyGxTmp32, 0x4e)); // 01001110
sumSignGyGxTmp32 = _mm_add_epi32(sumSignGyGxTmp32, _mm_shuffle_epi32(sumSignGyGxTmp32, 0xb1)); // 10110001
- *sumSignGY_GX = _mm_cvtsi128_si32(sumSignGyGxTmp32);
+ *sumSignGyGx = _mm_cvtsi128_si32(sumSignGyGxTmp32);
}
template< X86_VEXT vext >
-void calcBIOClippedVxVy_SSE(int* sumDIX_pixel_32bit, int* sumAbsGX_pixel_32bit, int* sumDIY_pixel_32bit, int* sumAbsGY_pixel_32bit, int* sumSignGY_GX_pixel_32bit, const int limit, const int bioSubblockSize, int* tmpx_pixel_32bit, int* tmpy_pixel_32bit)
+void calcBIOClippedVxVy_SSE(int* sumDIXSample32bit, int* sumAbsGxSample32bit, int* sumDIYSample32bit, int* sumAbsGySample32bit, int* sumSignGyGxSample32bit, const int limit, const int bioSubblockSize, int* tmpxSample32bit, int* tmpySample32bit)
{
#ifdef USE_AVX2
__m256i vibdimin = _mm256_set1_epi32(-limit);
@@ -967,25 +967,25 @@ void calcBIOClippedVxVy_SSE(int* sumDIX_pixel_32bit, int* sumAbsGX_pixel_32bit,
for (int idx = 0; idx < bioSubblockSize; idx += 8)
{
- tmp256 = _mm256_loadu_si256((const __m256i*)sumDIX_pixel_32bit);
- tmp256 = _mm256_srav_epi32(tmp256, _mm256_loadu_si256((const __m256i*)sumAbsGX_pixel_32bit));
+ tmp256 = _mm256_loadu_si256((const __m256i*)sumDIXSample32bit);
+ tmp256 = _mm256_srav_epi32(tmp256, _mm256_loadu_si256((const __m256i*)sumAbsGxSample32bit));
tmp256 = _mm256_max_epi32(tmp256, vibdimin);
tmp256 = _mm256_min_epi32(tmp256, vibdimax);
- _mm256_storeu_si256( ( __m256i * )tmpx_pixel_32bit, tmp256);
- tmp256 = _mm256_mullo_epi32(tmp256, _mm256_loadu_si256((const __m256i*)sumSignGY_GX_pixel_32bit));
+ _mm256_storeu_si256( ( __m256i * )tmpxSample32bit, tmp256);
+ tmp256 = _mm256_mullo_epi32(tmp256, _mm256_loadu_si256((const __m256i*)sumSignGyGxSample32bit));
tmp256 = _mm256_srai_epi32(tmp256, 1);
- tmp256 = _mm256_sub_epi32(_mm256_loadu_si256((const __m256i*)sumDIY_pixel_32bit), tmp256);
- tmp256 = _mm256_srav_epi32(tmp256, _mm256_loadu_si256((const __m256i*)sumAbsGY_pixel_32bit));
+ tmp256 = _mm256_sub_epi32(_mm256_loadu_si256((const __m256i*)sumDIYSample32bit), tmp256);
+ tmp256 = _mm256_srav_epi32(tmp256, _mm256_loadu_si256((const __m256i*)sumAbsGySample32bit));
tmp256 = _mm256_max_epi32(tmp256, vibdimin);
tmp256 = _mm256_min_epi32(tmp256, vibdimax);
- _mm256_storeu_si256( ( __m256i * )tmpy_pixel_32bit, tmp256);
- sumDIX_pixel_32bit += 8;
- sumAbsGX_pixel_32bit += 8;
- sumDIY_pixel_32bit += 8;
- sumAbsGY_pixel_32bit += 8;
- sumSignGY_GX_pixel_32bit += 8;
- tmpx_pixel_32bit += 8;
- tmpy_pixel_32bit += 8;
+ _mm256_storeu_si256( ( __m256i * )tmpySample32bit, tmp256);
+ sumDIXSample32bit += 8;
+ sumAbsGxSample32bit += 8;
+ sumDIYSample32bit += 8;
+ sumAbsGySample32bit += 8;
+ sumSignGyGxSample32bit += 8;
+ tmpxSample32bit += 8;
+ tmpySample32bit += 8;
}
#else
__m128i vibdimin = _mm_set1_epi32(-limit);
@@ -994,33 +994,33 @@ void calcBIOClippedVxVy_SSE(int* sumDIX_pixel_32bit, int* sumAbsGX_pixel_32bit,
for (int idx = 0; idx < bioSubblockSize; idx += 4)
{
- *sumDIX_pixel_32bit = (*sumDIX_pixel_32bit) >> (*sumAbsGX_pixel_32bit);
- *(sumDIX_pixel_32bit + 1) = (*(sumDIX_pixel_32bit + 1)) >> (*(sumAbsGX_pixel_32bit + 1));
- *(sumDIX_pixel_32bit + 2) = (*(sumDIX_pixel_32bit + 2)) >> (*(sumAbsGX_pixel_32bit + 2));
- *(sumDIX_pixel_32bit + 3) = (*(sumDIX_pixel_32bit + 3)) >> (*(sumAbsGX_pixel_32bit + 3));
- tmp128 = _mm_loadu_si128((const __m128i*)sumDIX_pixel_32bit);
+ *sumDIXSample32bit = (*sumDIXSample32bit) >> (*sumAbsGxSample32bit);
+ *(sumDIXSample32bit + 1) = (*(sumDIXSample32bit + 1)) >> (*(sumAbsGxSample32bit + 1));
+ *(sumDIXSample32bit + 2) = (*(sumDIXSample32bit + 2)) >> (*(sumAbsGxSample32bit + 2));
+ *(sumDIXSample32bit + 3) = (*(sumDIXSample32bit + 3)) >> (*(sumAbsGxSample32bit + 3));
+ tmp128 = _mm_loadu_si128((const __m128i*)sumDIXSample32bit);
tmp128 = _mm_max_epi32(tmp128, vibdimin);
tmp128 = _mm_min_epi32(tmp128, vibdimax);
- _mm_storeu_si128( ( __m128i * )tmpx_pixel_32bit, tmp128);
- tmp128 = _mm_mullo_epi32(tmp128, _mm_loadu_si128((const __m128i*)sumSignGY_GX_pixel_32bit));
+ _mm_storeu_si128( ( __m128i * )tmpxSample32bit, tmp128);
+ tmp128 = _mm_mullo_epi32(tmp128, _mm_loadu_si128((const __m128i*)sumSignGyGxSample32bit));
tmp128 = _mm_srai_epi32(tmp128, 1);
- tmp128 = _mm_sub_epi32(_mm_loadu_si128((const __m128i*)sumDIY_pixel_32bit), tmp128);
- _mm_storeu_si128( ( __m128i * )sumDIY_pixel_32bit, tmp128);
- *sumDIY_pixel_32bit = (*sumDIY_pixel_32bit) >> (*sumAbsGY_pixel_32bit);
- *(sumDIY_pixel_32bit + 1) = (*(sumDIY_pixel_32bit + 1)) >> (*(sumAbsGY_pixel_32bit + 1));
- *(sumDIY_pixel_32bit + 2) = (*(sumDIY_pixel_32bit + 2)) >> (*(sumAbsGY_pixel_32bit + 2));
- *(sumDIY_pixel_32bit + 3) = (*(sumDIY_pixel_32bit + 3)) >> (*(sumAbsGY_pixel_32bit + 3));
- tmp128 = _mm_loadu_si128((const __m128i*)sumDIY_pixel_32bit);
+ tmp128 = _mm_sub_epi32(_mm_loadu_si128((const __m128i*)sumDIYSample32bit), tmp128);
+ _mm_storeu_si128( ( __m128i * )sumDIYSample32bit, tmp128);
+ *sumDIYSample32bit = (*sumDIYSample32bit) >> (*sumAbsGySample32bit);
+ *(sumDIYSample32bit + 1) = (*(sumDIYSample32bit + 1)) >> (*(sumAbsGySample32bit + 1));
+ *(sumDIYSample32bit + 2) = (*(sumDIYSample32bit + 2)) >> (*(sumAbsGySample32bit + 2));
+ *(sumDIYSample32bit + 3) = (*(sumDIYSample32bit + 3)) >> (*(sumAbsGySample32bit + 3));
+ tmp128 = _mm_loadu_si128((const __m128i*)sumDIYSample32bit);
tmp128 = _mm_max_epi32(tmp128, vibdimin);
tmp128 = _mm_min_epi32(tmp128, vibdimax);
- _mm_storeu_si128( ( __m128i * )tmpy_pixel_32bit, tmp128);
- sumDIX_pixel_32bit += 4;
- sumAbsGX_pixel_32bit += 4;
- sumDIY_pixel_32bit += 4;
- sumAbsGY_pixel_32bit += 4;
- sumSignGY_GX_pixel_32bit += 4;
- tmpx_pixel_32bit += 4;
- tmpy_pixel_32bit += 4;
+ _mm_storeu_si128( ( __m128i * )tmpySample32bit, tmp128);
+ sumDIXSample32bit += 4;
+ sumAbsGxSample32bit += 4;
+ sumDIYSample32bit += 4;
+ sumAbsGySample32bit += 4;
+ sumSignGyGxSample32bit += 4;
+ tmpxSample32bit += 4;
+ tmpySample32bit += 4;
}
#endif
}
@@ -1219,7 +1219,7 @@ void calAbsSum_SSE(const Pel* diff, int stride, int width, int height, int* absS
#endif
template< X86_VEXT vext >
-void calcBIOSums_SSE(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0, Pel* gradX1, Pel* gradY0, Pel* gradY1, int xu, int yu, const int src0Stride, const int src1Stride, const int widthG, const int bitDepth, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGY_GX)
+void calcBIOSums_SSE(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0, Pel* gradX1, Pel* gradY0, Pel* gradY1, int xu, int yu, const int src0Stride, const int src1Stride, const int widthG, const int bitDepth, int* sumAbsGX, int* sumAbsGY, int* sumDIX, int* sumDIY, int* sumSignGyGx)
{
int shift4 = 4;
@@ -1254,13 +1254,13 @@ void calcBIOSums_SSE(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0, Pel*
__m128i gY = _mm_abs_epi16(packTempY);
__m128i dIX = _mm_sign_epi16(subTemp1, packTempX );
__m128i dIY = _mm_sign_epi16(subTemp1, packTempY );
- __m128i signGY_GX = _mm_sign_epi16(packTempX, packTempY );
+ __m128i signGyGx = _mm_sign_epi16(packTempX, packTempY );
sumAbsGXTmp = _mm_add_epi16(sumAbsGXTmp, gX);
sumDIXTmp = _mm_add_epi16(sumDIXTmp, dIX);
sumAbsGYTmp = _mm_add_epi16(sumAbsGYTmp, gY);
sumDIYTmp = _mm_add_epi16(sumDIYTmp, dIY);
- sumSignGyGxTmp = _mm_add_epi16(sumSignGyGxTmp, signGY_GX);
+ sumSignGyGxTmp = _mm_add_epi16(sumSignGyGxTmp, signGyGx);
srcY0Tmp += src0Stride;
srcY1Tmp += src1Stride;
gradX0 += widthG;
@@ -1293,7 +1293,7 @@ void calcBIOSums_SSE(const Pel* srcY0Tmp, const Pel* srcY1Tmp, Pel* gradX0, Pel*
sumSignGyGxTmp = _mm_add_epi32(sumSignGyGxTmp, _mm_shuffle_epi32(sumSignGyGxTmp, 0x4e)); // 01001110
sumSignGyGxTmp = _mm_add_epi32(sumSignGyGxTmp, _mm_shuffle_epi32(sumSignGyGxTmp, 0xb1)); // 10110001
- *sumSignGY_GX = _mm_cvtsi128_si32(sumSignGyGxTmp);
+ *sumSignGyGx = _mm_cvtsi128_si32(sumSignGyGxTmp);
}
template< X86_VEXT vext >
diff --git a/source/Lib/DecoderLib/DecCu.cpp b/source/Lib/DecoderLib/DecCu.cpp
index 9e21ed60836dc13027a66354763810d054e4b919..35aeabfe502aff54fdecd49bdf4d13ba094e4d48 100644
--- a/source/Lib/DecoderLib/DecCu.cpp
+++ b/source/Lib/DecoderLib/DecCu.cpp
@@ -710,7 +710,7 @@ void DecCu::xIntraRecBlk( TransformUnit& tu, const ComponentID compID )
#if JVET_AC0071_DBV
if (compID != COMPONENT_Y && uiChFinalMode == DBV_CHROMA_IDX)
{
- m_pcIntraPred->PredIntraDbv(compID, piPred, pu);
+ m_pcIntraPred->predIntraDbv(compID, piPred, pu);
}
else
#endif
@@ -749,7 +749,7 @@ void DecCu::xIntraRecBlk( TransformUnit& tu, const ComponentID compID )
#if JVET_AC0071_DBV
if (uiChFinalMode == DBV_CHROMA_IDX)
{
- m_pcIntraPred->PredIntraDbv(COMPONENT_Cr, piPredCr, pu);
+ m_pcIntraPred->predIntraDbv(COMPONENT_Cr, piPredCr, pu);
}
else
#endif
diff --git a/source/Lib/EncoderLib/IntraSearch.cpp b/source/Lib/EncoderLib/IntraSearch.cpp
index 15508e55380b459907de00a280586a7e78007020..248da80da9a13f4f91cbebe6c4b48629751f58c8 100644
--- a/source/Lib/EncoderLib/IntraSearch.cpp
+++ b/source/Lib/EncoderLib/IntraSearch.cpp
@@ -7660,16 +7660,15 @@ bool IntraSearch::xRecurIntraCodingLumaQT( CodingStructure &cs, Partitioner &par
saveCS.getPredBuf( tu.Y() ).copyFrom( csFull->getPredBuf( tu.Y() ) );
saveCS.getRecoBuf( tu.Y() ).copyFrom( csFull->getRecoBuf( tu.Y() ) );
- if( KEEP_PRED_AND_RESI_SIGNALS )
+ if( KEEP_PRED_AND_RESI_SIGNALS || JVET_AC0162_ALF_RESIDUAL_SAMPLES_INPUT )
{
saveCS.getResiBuf ( tu.Y() ).copyFrom( csFull->getResiBuf ( tu.Y() ) );
+ }
+ if( KEEP_PRED_AND_RESI_SIGNALS )
+ {
saveCS.getOrgResiBuf( tu.Y() ).copyFrom( csFull->getOrgResiBuf( tu.Y() ) );
}
-#if JVET_AC0162_ALF_RESIDUAL_SAMPLES_INPUT
- saveCS.getResiBuf(tu.Y()).copyFrom(csFull->getResiBuf(tu.Y()));
-#endif
-
tmpTU->copyComponentFrom( tu, COMPONENT_Y );
ctxBest = m_CABACEstimator->getCtx();
@@ -7693,16 +7692,15 @@ bool IntraSearch::xRecurIntraCodingLumaQT( CodingStructure &cs, Partitioner &par
csFull->getPredBuf( tu.Y() ).copyFrom( saveCS.getPredBuf( tu.Y() ) );
csFull->getRecoBuf( tu.Y() ).copyFrom( saveCS.getRecoBuf( tu.Y() ) );
- if( KEEP_PRED_AND_RESI_SIGNALS )
+ if( KEEP_PRED_AND_RESI_SIGNALS || JVET_AC0162_ALF_RESIDUAL_SAMPLES_INPUT )
{
csFull->getResiBuf ( tu.Y() ).copyFrom( saveCS.getResiBuf ( tu.Y() ) );
+ }
+ if( KEEP_PRED_AND_RESI_SIGNALS )
+ {
csFull->getOrgResiBuf( tu.Y() ).copyFrom( saveCS.getOrgResiBuf( tu.Y() ) );
}
-#if JVET_AC0162_ALF_RESIDUAL_SAMPLES_INPUT
- csFull->getResiBuf(tu.Y()).copyFrom(saveCS.getResiBuf(tu.Y()));
-#endif
-
tu.copyComponentFrom( *tmpTU, COMPONENT_Y );
if( !bCheckSplit )
@@ -8833,8 +8831,8 @@ ChromaCbfs IntraSearch::xRecurIntraChromaCodingQT( CodingStructure &cs, Partitio
#if JVET_AC0071_DBV
if (predMode == DBV_CHROMA_IDX)
{
- PredIntraDbv(COMPONENT_Cb, piPredCb, pu);
- PredIntraDbv(COMPONENT_Cr, piPredCr, pu);
+ predIntraDbv(COMPONENT_Cb, piPredCb, pu);
+ predIntraDbv(COMPONENT_Cr, piPredCr, pu);
}
else
{