[x265] [PATCH] log: make qTreeCnt as stack arrays to avoid non determinism in 2 pass

[Steve Borho]

On 04/28, aarthi at multicorewareinc.com wrote:
> # HG changeset patch
> # User Aarthi Thirumalai
> # Date 1430225684 -19800
> #      Tue Apr 28 18:24:44 2015 +0530
> # Node ID 9a66b31df0d07a7146694763de119cb2e0ca2cfa
> # Parent  13290abce29209be5206cce66e436c8ff3682540
> log: make qTreeCnt as stack arrays to avoid non determinism in 2 pass
> 
> currently,qTreeCount arrays collected in log stats for each sliceType is not protected and may cause
> inconsistency in 2 pass encodes when multiple rows finish simultaneously.
> 
> diff -r 13290abce292 -r 9a66b31df0d0 source/encoder/frameencoder.cpp
> --- a/source/encoder/frameencoder.cpp	Mon Apr 27 14:15:28 2015 -0500
> +++ b/source/encoder/frameencoder.cpp	Tue Apr 28 18:24:44 2015 +0530
> @@ -823,7 +823,12 @@
>      const uint32_t numCols = m_numCols;
>      const uint32_t lineStartCUAddr = row * numCols;
>      bool bIsVbv = m_param->rc.vbvBufferSize > 0 && m_param->rc.vbvMaxBitrate > 0;
> -
> +    /* These states store the count of inter,intra and skip ctus within quad tree structure of each CU */
> +    uint32_t qTreeInterCnt[4];
> +    uint32_t qTreeIntraCnt[4];
> +    uint32_t qTreeSkipCnt[4];

queued, using an enum here instead of 4

> +    for (uint32_t depth = 0; depth <= g_maxCUDepth; depth++)
> +        qTreeIntraCnt[depth] = qTreeInterCnt[depth] = qTreeSkipCnt[depth] = 0;
>      while (curRow.completed < numCols)
>      {
>          ProfileScopeEvent(encodeCTU);
> @@ -896,7 +901,7 @@
>          curRow.completed++;
>  
>          if (m_param->bLogCuStats || m_param->rc.bStatWrite)
> -            curEncData.m_rowStat[row].sumQpAq += collectCTUStatistics(*ctu);
> +            curEncData.m_rowStat[row].sumQpAq += collectCTUStatistics(*ctu, qTreeInterCnt, qTreeIntraCnt, qTreeSkipCnt);
>          else if (m_param->rc.aqMode)
>              curEncData.m_rowStat[row].sumQpAq += calcCTUQP(*ctu);
>  
> @@ -906,18 +911,17 @@
>              curRow.rowStats.mvBits += best.mvBits;
>              curRow.rowStats.coeffBits += best.coeffBits;
>              curRow.rowStats.miscBits += best.totalBits - (best.mvBits + best.coeffBits);
> -            StatisticLog* log = &m_sliceTypeLog[slice->m_sliceType];
>  
>              for (uint32_t depth = 0; depth <= g_maxCUDepth; depth++)
>              {
>                  /* 1 << shift == number of 8x8 blocks at current depth */
>                  int shift = 2 * (g_maxCUDepth - depth);
> -                curRow.rowStats.iCuCnt += log->qTreeIntraCnt[depth] << shift;
> -                curRow.rowStats.pCuCnt += log->qTreeInterCnt[depth] << shift;
> -                curRow.rowStats.skipCuCnt += log->qTreeSkipCnt[depth] << shift;
> +                curRow.rowStats.iCuCnt += qTreeIntraCnt[depth] << shift;
> +                curRow.rowStats.pCuCnt += qTreeInterCnt[depth] << shift;
> +                curRow.rowStats.skipCuCnt += qTreeSkipCnt[depth] << shift;
>  
>                  // clear the row cu data from thread local object
> -                log->qTreeIntraCnt[depth] = log->qTreeInterCnt[depth] = log->qTreeSkipCnt[depth] = 0;
> +                qTreeIntraCnt[depth] = qTreeInterCnt[depth] = qTreeSkipCnt[depth] = 0;
>              }
>          }
>  
> @@ -1094,7 +1098,7 @@
>  }
>  
>  /* collect statistics about CU coding decisions, return total QP */
> -int FrameEncoder::collectCTUStatistics(const CUData& ctu)
> +int FrameEncoder::collectCTUStatistics(const CUData& ctu, uint32_t* qtreeInterCnt, uint32_t* qtreeIntraCnt, uint32_t* qtreeSkipCnt)
>  {
>      StatisticLog* log = &m_sliceTypeLog[ctu.m_slice->m_sliceType];
>      int totQP = 0;
> @@ -1108,14 +1112,14 @@
>  
>              log->totalCu++;
>              log->cntIntra[depth]++;
> -            log->qTreeIntraCnt[depth]++;
> +            qtreeIntraCnt[depth]++;
>              totQP += ctu.m_qp[absPartIdx] * (ctu.m_numPartitions >> (depth * 2));
>  
>              if (ctu.m_predMode[absPartIdx] == MODE_NONE)
>              {
>                  log->totalCu--;
>                  log->cntIntra[depth]--;
> -                log->qTreeIntraCnt[depth]--;
> +                qtreeIntraCnt[depth]--;
>              }
>              else if (ctu.m_partSize[absPartIdx] != SIZE_2Nx2N)
>              {
> @@ -1150,12 +1154,12 @@
>              {
>                  log->totalCu--;
>                  log->cntSkipCu[depth]++;
> -                log->qTreeSkipCnt[depth]++;
> +                qtreeSkipCnt[depth]++;
>              }
>              else if (ctu.isInter(absPartIdx))
>              {
>                  log->cntInter[depth]++;
> -                log->qTreeInterCnt[depth]++;
> +                qtreeInterCnt[depth]++;
>  
>                  if (ctu.m_partSize[absPartIdx] < AMP_ID)
>                      log->cuInterDistribution[depth][ctu.m_partSize[absPartIdx]]++;
> @@ -1165,7 +1169,7 @@
>              else if (ctu.isIntra(absPartIdx))
>              {
>                  log->cntIntra[depth]++;
> -                log->qTreeIntraCnt[depth]++;
> +                qtreeIntraCnt[depth]++;
>  
>                  if (ctu.m_partSize[absPartIdx] != SIZE_2Nx2N)
>                  {
> diff -r 13290abce292 -r 9a66b31df0d0 source/encoder/frameencoder.h
> --- a/source/encoder/frameencoder.h	Mon Apr 27 14:15:28 2015 -0500
> +++ b/source/encoder/frameencoder.h	Tue Apr 28 18:24:44 2015 +0530
> @@ -63,11 +63,6 @@
>      uint64_t cntTotalCu[4];
>      uint64_t totalCu;
>  
> -    /* These states store the count of inter,intra and skip ctus within quad tree structure of each CU */
> -    uint32_t qTreeInterCnt[4];
> -    uint32_t qTreeIntraCnt[4];
> -    uint32_t qTreeSkipCnt[4];
> -
>      StatisticLog()
>      {
>          memset(this, 0, sizeof(StatisticLog));
> @@ -226,7 +221,7 @@
>      void encodeSlice();
>  
>      void threadMain();
> -    int  collectCTUStatistics(const CUData& ctu);
> +    int  collectCTUStatistics(const CUData& ctu, uint32_t* qtreeInterCnt, uint32_t* qtreeIntraCnt, uint32_t* qtreeSkipCnt);
>      int  calcCTUQP(const CUData& ctu);
>      void noiseReductionUpdate();
>  
> _______________________________________________
> x265-devel mailing list
> x265-devel at videolan.org
> https://mailman.videolan.org/listinfo/x265-devel

-- 
Steve Borho