[x265] [PATCH] analysis: model the effectiveness of --limit-ref with RD 0/4

[Steve Borho]

On 05/19, ashok at multicorewareinc.com wrote:
> # HG changeset patch
> # User Ashok Kumar Mishra<ashok at multicorewareinc.com>
> # Date 1431935793 -19800
> #      Mon May 18 13:26:33 2015 +0530
> # Node ID f6c9a1e184fe7a8c17744117120828458be43106
> # Parent  5bc61c2bc0cec50dc33eda9638f215de21fb4bcf
> analysis: model the effectiveness of --limit-ref with RD 0/4

the series looks good, queued for testing

> diff -r 5bc61c2bc0ce -r f6c9a1e184fe source/encoder/analysis.cpp
> --- a/source/encoder/analysis.cpp	Mon May 18 13:24:42 2015 +0530
> +++ b/source/encoder/analysis.cpp	Mon May 18 13:26:33 2015 +0530
> @@ -850,6 +850,26 @@
>              md.pred[PRED_2Nx2N].cu.initSubCU(parentCTU, cuGeom, qp);
>              checkInter_rd0_4(md.pred[PRED_2Nx2N], cuGeom, SIZE_2Nx2N, refMasks);
>  
> +            if (m_param->limitReferences & X265_REF_LIMIT_CU)
> +            {
> +                CUData& cu = md.pred[PRED_2Nx2N].cu;
> +                int refMask;
> +                switch (cu.m_interDir[0])
> +                {
> +                case 1:
> +                    refMask = 1 << cu.m_refIdx[0][0];
> +                    break;
> +                case 2:
> +                    refMask = 1 << (cu.m_refIdx[1][0] + 16);
> +                    break;
> +                case 3:
> +                    refMask = 1 << cu.m_refIdx[0][0];
> +                    refMask |= 1 << (cu.m_refIdx[1][0] + 16);
> +                    break;
> +                }

this probably wants to eventually be a cudata helper function, and could
be micro-optimized to avoid the switch statement by using:

  refmask |= (cu.m_interDir[0] & (1 << list)) * listmask;

> +                allSplitRefs = splitRefs[0] = splitRefs[1] = splitRefs[2] = splitRefs[3] = refMask;
> +            }
> +
>              if (m_slice->m_sliceType == B_SLICE)
>              {
>                  md.pred[PRED_BIDIR].cu.initSubCU(parentCTU, cuGeom, qp);
> _______________________________________________
> x265-devel mailing list
> x265-devel at videolan.org
> https://mailman.videolan.org/listinfo/x265-devel

-- 
Steve Borho