<div style="line-height:1.7;color:#000000;font-size:14px;font-family:arial"><div>If we split these common data into one class, we have to pass pointer of FrameFilter to ParallelFilter,</div>
<div>it made slower access in process (see below), so I make a copy in local class member.</div>
<div> </div>
<div>old: class -> offset -> numRows</div>
<div>new: pointer to FrameFilter -> class -> offset -> numRows</div>
<div></div>
<div id="divNeteaseMailCard"></div>
<div><br></div>At 2016-01-14 16:01:19,"Deepthi Nandakumar" <deepthi@multicorewareinc.com> wrote:<br>
<blockquote id="isReplyContent" style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: #ccc 1px solid">
<div dir="ltr">Min,
<div><br></div>
<div>Sure - what I mean is, can we have a class that looks like: </div>
<div><br></div>
<div>
<div>class FrameFilter</div>
<div>{</div>
<div>public:</div>
<div><br></div>
<div> x265_param* m_param;</div>
<div> Frame* m_frame;</div>
<div> FrameEncoder* m_frameEncoder;</div>
<div> FrameData* m_encData;</div>
<div> int m_hChromaShift;</div>
<div> int m_vChromaShift;</div>
<div> int m_pad[2];</div>
<div><br></div>
<div> uint32_t m_numRows;</div>
<div> uint32_t m_numCols;</div>
<div> uint32_t m_lastWidth;</div>
<div> .................</div>
<div>..................</div>
<div>}</div>
<div><br></div>
<div>class ParallelFilter</div>
<div>{</div>
<div>public:</div>
<div> uint32_t m_rowHeight;</div>
<div> uint32_t m_row;</div>
<div> uint32_t m_rowAddr;</div>
<div><br></div>
<div> ParallelFilter* m_prevRow;</div>
<div> SAO m_sao;</div>
<div> ThreadSafeInteger m_lastCol; /* The column that next to process */</div>
<div> ThreadSafeInteger m_allowedCol; /* The column that processed from Encode pipeline */</div>
<div> ThreadSafeInteger m_lastDeblocked; /* The column that finished all of Deblock stages */</div>
<div><br></div>
<div>}</div></div>
<div><br></div>
<div>The common variables are in FrameFilter, and ParallelFilter contains only the row specific data. </div></div>
<div class="gmail_extra"><br>
<div class="gmail_quote">On Wed, Jan 13, 2016 at 12:15 AM, chen <span dir="ltr"><<a href="mailto:chenm003@163.com" target="_blank">chenm003@163.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: #ccc 1px solid">
<div style="FONT-SIZE: 14px; COLOR: #000000; LINE-HEIGHT: 1.7; FONT-FAMILY: arial">
<div>I fixed the crash and modify/remove FrameFilter::processPostCu, I spending ~64 bytes every row.</div>
<div><br></div>
<div>In my design, FrameFilter just one instance every frame and ParallelFilter instance for every row.</div>
<div>
<div class="h5">
<div><br></div>
<div></div>
<div></div>
<div><br></div>At 2016-01-12 13:37:36,"Deepthi Nandakumar" <<a href="mailto:deepthi@multicorewareinc.com" target="_blank">deepthi@multicorewareinc.com</a>> wrote:<br>
<blockquote style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: rgb(204,204,204) 1px solid">
<div dir="ltr">
<div>Min, thanks. Queuing this.<br><br></div>I still think the ParallelFilter class needs a refactor - not nested, but inherited from FrameFilter. And FrameFilter contains all the constants for a single frame - numCols, numRows etc. This will also clean up FrameFilter::processPostCu (which is causing the crash in no-deblock, no-sao, since now you'll only refer to FrameFilter class members and not ParallelFilter (which is null if sao and deblock are disabled). <br><br></div>
<div class="gmail_extra"><br>
<div class="gmail_quote">On Tue, Jan 12, 2016 at 2:36 AM, Min Chen <span dir="ltr"><<a href="mailto:chenm003@163.com" target="_blank">chenm003@163.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: rgb(204,204,204) 1px solid"># HG changeset patch<br># User Min Chen <<a href="mailto:chenm003@163.com" target="_blank">chenm003@163.com</a>><br># Date 1452545402 21600<br># Node ID 1c273c83a4478ed39b0e6734eec1ba1cfccd07d6<br># Parent 19cfada7162147f293e37302e4c7f9c1760928a0<br>fix deadlock and output change on new ParallelFilter framework. (Issue #225)<br>The bug from two part:<br>1. we old sync system allow latest column execute parallelism<br>2. new ParallelFilter logic will distribute threads and waitting the status change, there have a race conditions<br>---<br> source/encoder/frameencoder.cpp | 108 +++++++++++++++++++--------------------<br> 1 files changed, 53 insertions(+), 55 deletions(-)<br><br>diff -r 19cfada71621 -r 1c273c83a447 source/encoder/frameencoder.cpp<br>--- a/source/encoder/frameencoder.cpp Mon Jan 11 16:28:39 2016 +0530<br>+++ b/source/encoder/frameencoder.cpp Mon Jan 11 14:50:02 2016 -0600<br>@@ -962,6 +962,58 @@<br> // Save CABAC state for next row<br> curRow.bufferedEntropy.loadContexts(rowCoder);<br><br>+ /* SAO parameter estimation using non-deblocked pixels for CTU bottom and right boundary areas */<br>+ if (m_param->bEnableSAO && m_param->bSaoNonDeblocked)<br>+ m_frameFilter.m_parallelFilter[row].m_sao.calcSaoStatsCu_BeforeDblk(m_frame, col, row);<br>+<br>+ /* Deblock with idle threading */<br>+ if (m_param->bEnableLoopFilter | m_param->bEnableSAO)<br>+ {<br>+ // TODO: Multiple Threading<br>+ // Delay ONE row to avoid Intra Prediction Conflict<br>+ if (m_pool && (row >= 1))<br>+ {<br>+ // Waitting last threading finish<br>+ m_frameFilter.m_parallelFilter[row - 1].waitForExit();<br>+<br>+ // Processing new group<br>+ int allowCol = col;<br>+<br>+ // avoid race condition on last column<br>+ if (row >= 2)<br>+ {<br>+ allowCol = X265_MIN(((col == numCols - 1) ? m_frameFilter.m_parallelFilter[row - 2].m_lastDeblocked.get()<br>+ : m_frameFilter.m_parallelFilter[row - 2].m_lastCol.get()), (int)col);<br>+ }<br>+ m_frameFilter.m_parallelFilter[row - 1].m_allowedCol.set(allowCol);<br>+ m_frameFilter.m_parallelFilter[row - 1].tryBondPeers(*this, 1);<br>+ }<br>+<br>+ // Last Row may start early<br>+ if (m_pool && (row == m_numRows - 1))<br>+ {<br>+ // Waiting for the last thread to finish<br>+ m_frameFilter.m_parallelFilter[row].waitForExit();<br>+<br>+ // Deblocking last row<br>+ int allowCol = col;<br>+<br>+ // avoid race condition on last column<br>+ if (row >= 2)<br>+ {<br>+ allowCol = X265_MIN(((col == numCols - 1) ? m_frameFilter.m_parallelFilter[row - 1].m_lastDeblocked.get()<br>+ : m_frameFilter.m_parallelFilter[row - 1].m_lastCol.get()), (int)col);<br>+ }<br>+ m_frameFilter.m_parallelFilter[row].m_allowedCol.set(allowCol);<br>+ m_frameFilter.m_parallelFilter[row].tryBondPeers(*this, 1);<br>+ }<br>+ }<br>+ // Both Loopfilter and SAO Disabled<br>+ else<br>+ {<br>+ m_frameFilter.processPostCu(row, col);<br>+ }<br>+<br> // Completed CU processing<br> curRow.completed++;<br><br>@@ -1091,60 +1143,6 @@<br> }<br> }<br><br>- // TODO: move Deblock and SAO to before VBV check<br>-<br>- /* SAO parameter estimation using non-deblocked pixels for CTU bottom and right boundary areas */<br>- if (m_param->bEnableSAO && m_param->bSaoNonDeblocked)<br>- m_frameFilter.m_parallelFilter[row].m_sao.calcSaoStatsCu_BeforeDblk(m_frame, col, row);<br>-<br>- /* Deblock with idle threading */<br>- if (m_param->bEnableLoopFilter | m_param->bEnableSAO)<br>- {<br>- // TODO: Multiple Threading<br>- // Delay ONE row to avoid Intra Prediction Conflict<br>- if (m_pool && (row >= 1))<br>- {<br>- // Waitting last threading finish<br>- m_frameFilter.m_parallelFilter[row - 1].waitForExit();<br>-<br>- // Processing new group<br>- int allowCol = col;<br>-<br>- // avoid race condition on last column<br>- if (row >= 2)<br>- {<br>- allowCol = X265_MIN(((col == numCols - 1) ? m_frameFilter.m_parallelFilter[row - 2].m_lastDeblocked.get()<br>- : m_frameFilter.m_parallelFilter[row - 2].m_lastCol.get()), (int)col);<br>- }<br>- m_frameFilter.m_parallelFilter[row - 1].m_allowedCol.set(allowCol);<br>- m_frameFilter.m_parallelFilter[row - 1].tryBondPeers(*this, 1);<br>- }<br>-<br>- // Last Row may start early<br>- if (m_pool && (row == m_numRows - 1))<br>- {<br>- // Waiting for the last thread to finish<br>- m_frameFilter.m_parallelFilter[row].waitForExit();<br>-<br>- // Deblocking last row<br>- int allowCol = col;<br>-<br>- // avoid race condition on last column<br>- if (row >= 2)<br>- {<br>- allowCol = X265_MIN(((col == numCols - 1) ? m_frameFilter.m_parallelFilter[row - 1].m_lastDeblocked.get()<br>- : m_frameFilter.m_parallelFilter[row - 1].m_lastCol.get()), (int)col);<br>- }<br>- m_frameFilter.m_parallelFilter[row].m_allowedCol.set(allowCol);<br>- m_frameFilter.m_parallelFilter[row].tryBondPeers(*this, 1);<br>- }<br>- }<br>- // Both Loopfilter and SAO Disabled<br>- else<br>- {<br>- m_frameFilter.processPostCu(row, col);<br>- }<br>-<br> if (m_param->bEnableWavefront && curRow.completed >= 2 && row < m_numRows - 1 &&<br> (!m_bAllRowsStop || intRow + 1 < m_vbvResetTriggerRow))<br> {<br>@@ -1161,7 +1159,7 @@<br><br> ScopedLock self(curRow.lock);<br> if ((m_bAllRowsStop && intRow > m_vbvResetTriggerRow) ||<br>- (row > 0 && curRow.completed < numCols - 1 && m_rows[row - 1].completed < m_rows[row].completed + 2))<br>+ (row > 0 && ((curRow.completed < numCols - 1) || (m_rows[row - 1].completed < numCols)) && m_rows[row - 1].completed < m_rows[row].completed + 2))<br> {<br> curRow.active = false;<br> curRow.busy = false;<br><br>_______________________________________________<br>x265-devel mailing list<br><a href="mailto:x265-devel@videolan.org" target="_blank">x265-devel@videolan.org</a><br><a href="https://mailman.videolan.org/listinfo/x265-devel" target="_blank" rel="noreferrer">https://mailman.videolan.org/listinfo/x265-devel</a><br></blockquote></div><br><br clear="all"><br>-- <br>
<div>
<div dir="ltr">
<div>
<div>Deepthi Nandakumar<br></div>Engineering Manager, x265<br></div>Multicoreware, Inc<br></div></div></div></blockquote></div></div></div><br>_______________________________________________<br>x265-devel mailing list<br><a href="mailto:x265-devel@videolan.org">x265-devel@videolan.org</a><br><a href="https://mailman.videolan.org/listinfo/x265-devel" target="_blank" rel="noreferrer">https://mailman.videolan.org/listinfo/x265-devel</a><br><br></blockquote></div><br><br clear="all">
<div><br></div>-- <br>
<div class="gmail_signature">
<div dir="ltr">
<div>
<div>Deepthi Nandakumar<br></div>Engineering Manager, x265<br></div>Multicoreware, Inc<br></div></div></div></blockquote></div>