Min, Can you please share the impact that this sequence of 3 patches has on performance for our presets. What machine did you do the test on?
Pradeep Ramachandran, PhD Solution Architect, pradeeprama.info/ www.multicorewareinc.com/ Ph: +91 99627 82018 On Tue, Dec 22, 2015 at 7:19 AM, Min Chen <[email protected]> wrote: > # HG changeset patch > # User Min Chen <[email protected]> > # Date 1450728715 21600 > # Node ID 3e9e45fd692d79806e53d69c2ed7a20b3e24e671 > # Parent d8c3eded1440670bde63e2fb5bec0e80ff6e2d67 > reduce pointer operators on I400 > --- > source/encoder/framefilter.cpp | 7 ++++--- > 1 files changed, 4 insertions(+), 3 deletions(-) > > diff -r d8c3eded1440 -r 3e9e45fd692d source/encoder/framefilter.cpp > --- a/source/encoder/framefilter.cpp Mon Dec 21 13:46:28 2015 -0600 > +++ b/source/encoder/framefilter.cpp Mon Dec 21 14:11:55 2015 -0600 > @@ -507,7 +507,7 @@ > const intptr_t stride = reconPic->m_stride; > const intptr_t strideC = reconPic->m_strideC; > pixel *pixY = reconPic->getLumaAddr(lineStartCUAddr); > - // // MUST BE check I400 since m_picOrg uninitialize in that case > + // MUST BE check I400 since m_picOrg uninitialize in that case > pixel *pixU = (m_param->internalCsp != X265_CSP_I400) ? > reconPic->getCbAddr(lineStartCUAddr) : NULL; > pixel *pixV = (m_param->internalCsp != X265_CSP_I400) ? > reconPic->getCrAddr(lineStartCUAddr) : NULL; > int copySizeY = realW; > @@ -560,13 +560,14 @@ > if (row == FrameFilter::ParallelFilter::numRows - 1) > { > pixY += (realH - 1) * stride; > - pixU += ((realH >> vChromaShift) - 1) * strideC; > - pixV += ((realH >> vChromaShift) - 1) * strideC; > for (uint32_t y = 0; y < lumaMarginY; y++) > memcpy(pixY + (y + 1) * stride, pixY, copySizeY * > sizeof(pixel)); > > if (m_param->internalCsp != X265_CSP_I400) > { > + pixU += ((realH >> vChromaShift) - 1) * strideC; > + pixV += ((realH >> vChromaShift) - 1) * strideC; > + > for (uint32_t y = 0; y < chromaMarginY; y++) > { > memcpy(pixU + (y + 1) * strideC, pixU, copySizeC * > sizeof(pixel)); > > _______________________________________________ > x265-devel mailing list > [email protected] > https://mailman.videolan.org/listinfo/x265-devel >
_______________________________________________ x265-devel mailing list [email protected] https://mailman.videolan.org/listinfo/x265-devel
