[x265] [PATCH 2 of 6] optimize SAO statistics initialize

[Deepthi Nandakumar]

On Wed, Nov 18, 2015 at 10:44 PM, Min Chen <chenm003 at 163.com> wrote:

> # HG changeset patch
> # User Min Chen <chenm003 at 163.com>
> # Date 1447865926 21600
> # Node ID 600706db4e14bd95a5d933418186f86403e897f4
> # Parent  e160de7083971f21802517691f9a3a0934765952
> optimize SAO statistics initialize
> ---
>  source/encoder/sao.cpp |   52
> ++++++++++++++++++++++++-----------------------
>  1 files changed, 27 insertions(+), 25 deletions(-)
>
> diff -r e160de708397 -r 600706db4e14 source/encoder/sao.cpp
> --- a/source/encoder/sao.cpp    Wed Nov 18 10:58:43 2015 -0600
> +++ b/source/encoder/sao.cpp    Wed Nov 18 10:58:46 2015 -0600
> @@ -1156,32 +1156,34 @@
>          if (allowMerge[1])
>              m_entropyCoder.codeSaoMerge(0);
>          m_entropyCoder.store(m_rdContexts.temp);
> +
>          // reset stats Y, Cb, Cr
> -        for (int plane = 0; plane < 3; plane++)
> +        X265_CHECK(sizeof(PerPlane) == (sizeof(int32_t) * (NUM_PLANE *
> MAX_NUM_SAO_TYPE * MAX_NUM_SAO_CLASS)), "Found Padding space in struct
> PerPlane");
> +
> +        // TODO: Confirm the address space is continuous
> +        memset(m_count, 0, 3 * sizeof(m_count[0]));
>

Now that we're cleaning this up, m_count is confusing. PerClass is a 2D int
array, and we're allocating one per plane? Can we clean this up once and
for all in the malloc part itself?

+        if (m_param->bSaoNonDeblocked)
>          {
> -            for (int j = 0; j < MAX_NUM_SAO_TYPE; j++)
> -            {
> -                for (int k = 0; k < MAX_NUM_SAO_CLASS; k++)
> -                {
> -                    m_offset[plane][j][k] = 0;
> -                    if (m_param->bSaoNonDeblocked)
> -                    {
> -                        m_count[plane][j][k] =
> m_countPreDblk[addr][plane][j][k];
> -                        m_offsetOrg[plane][j][k] =
> m_offsetOrgPreDblk[addr][plane][j][k];
> -                    }
> -                    else
> -                    {
> -                        m_count[plane][j][k] = 0;
> -                        m_offsetOrg[plane][j][k] = 0;
> -                    }
> -                }
> -            }
> +            memcpy(m_count, m_countPreDblk[addr], 3 * sizeof(m_count[0]));
> +            memcpy(m_offsetOrg, m_offsetOrgPreDblk[addr], 3 *
> sizeof(m_offsetOrg[0]));
> +        }
> +        else
> +        {
> +            memset(m_count, 0, 3 * sizeof(m_count[0]));
> +            memset(m_offsetOrg, 0, 3 * sizeof(m_offsetOrg[0]));
> +        }
> +        const struct SaoCtuParam dummy_initSaoParam = { SAO_MERGE_NONE,
> -1, 0, {0, 0, 0, 0} };
> +        saoParam->ctuParam[0][addr] = dummy_initSaoParam;
> +        saoParam->ctuParam[1][addr] = dummy_initSaoParam;
> +        saoParam->ctuParam[2][addr] = dummy_initSaoParam;
>
> -            saoParam->ctuParam[plane][addr].mergeMode = SAO_MERGE_NONE;
> -            saoParam->ctuParam[plane][addr].typeIdx   = -1;
> -            saoParam->ctuParam[plane][addr].bandPos   = 0;
> -            if (saoParam->bSaoFlag[plane > 0])
> -                calcSaoStatsCu(addr, plane);
> +        if (saoParam->bSaoFlag[0])
> +            calcSaoStatsCu(addr, 0);
> +
> +        if (saoParam->bSaoFlag[1])
> +        {
> +            calcSaoStatsCu(addr, 1);
> +            calcSaoStatsCu(addr, 2);
>          }
>
>          saoComponentParamDist(saoParam, addr, addrUp, addrLeft,
> &mergeSaoParam[0][0], mergeDist);
>
> _______________________________________________
> x265-devel mailing list
> x265-devel at videolan.org
> https://mailman.videolan.org/listinfo/x265-devel
>



-- 
Deepthi Nandakumar
Engineering Manager, x265
Multicoreware, Inc
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/x265-devel/attachments/20151120/c3d53f28/attachment.html>