[Libreoffice-commits] core.git: vcl/source

[Mike Kaganski (via logerrit)]

 vcl/source/bitmap/BitmapFilterStackBlur.cxx |   61 ++++++++--------------------
 1 file changed, 19 insertions(+), 42 deletions(-)

New commits:
commit 72795f68df49cfb58aee70daa848a1889a117c28
Author:     Mike Kaganski <mike.kaganski at collabora.com>
AuthorDate: Mon May 4 16:20:21 2020 +0300
Commit:     Mike Kaganski <mike.kaganski at collabora.com>
CommitDate: Mon May 4 19:04:20 2020 +0200

    StackBlur filter: Don't create threads for small bitmap
    
    This makes the last (or the only) strip to run im main thread, thus
    removing threading overhead for small (width/height <= 16) objects.
    
    Change-Id: I6da050a62b00a013838f16eb94a32e2639d93aec
    Reviewed-on: https://gerrit.libreoffice.org/c/core/+/93423
    Tested-by: Jenkins
    Reviewed-by: Mike Kaganski <mike.kaganski at collabora.com>

diff --git a/vcl/source/bitmap/BitmapFilterStackBlur.cxx b/vcl/source/bitmap/BitmapFilterStackBlur.cxx
index fde6ab089a3b..9b3c37127fcf 100644
--- a/vcl/source/bitmap/BitmapFilterStackBlur.cxx
+++ b/vcl/source/bitmap/BitmapFilterStackBlur.cxx
@@ -48,9 +48,8 @@ static const sal_Int16 constShiftTable[255]
         24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24,
         24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24 };
 
-class BlurSharedData
+struct BlurSharedData
 {
-public:
     BitmapReadAccess* mpReadAccess;
     BitmapWriteAccess* mpWriteAccess;
     long mnRadius;
@@ -70,9 +69,8 @@ public:
     }
 };
 
-class BlurArrays
+struct BlurArrays
 {
-public:
     BlurSharedData maShared;
 
     std::vector<sal_uInt8> maStackBuffer;
@@ -110,8 +108,6 @@ public:
 
 typedef void (*BlurRangeFn)(BlurSharedData const& rShared, long nStartY, long nEndY);
 
-constexpr long constBlurThreadStrip = 16;
-
 class BlurTask : public comphelper::ThreadTask
 {
     BlurRangeFn mpBlurFunction;
@@ -454,6 +450,8 @@ void stackBlurVertical(BlurSharedData const& rShared, long nStart, long nEnd)
     }
 }
 
+constexpr long nThreadStrip = 16;
+
 void runStackBlur(Bitmap& rBitmap, const long nRadius, const long nComponentWidth,
                   const long nColorChannels, BlurRangeFn pBlurHorizontalFn,
                   BlurRangeFn pBlurVerticalFn, const bool bParallel)
@@ -463,65 +461,44 @@ void runStackBlur(Bitmap& rBitmap, const long nRadius, const long nComponentWidt
         try
         {
             comphelper::ThreadPool& rShared = comphelper::ThreadPool::getSharedOptimalPool();
-            std::shared_ptr<comphelper::ThreadTaskTag> pTag
-                = comphelper::ThreadPool::createThreadTaskTag();
+            auto pTag = comphelper::ThreadPool::createThreadTaskTag();
 
             {
                 Bitmap::ScopedReadAccess pReadAccess(rBitmap);
                 BitmapScopedWriteAccess pWriteAccess(rBitmap);
-
                 BlurSharedData aSharedData(pReadAccess.get(), pWriteAccess.get(), nRadius,
                                            nComponentWidth, nColorChannels);
 
-                const Size aSize = rBitmap.GetSizePixel();
-                long nEnd = aSize.Height() - 1;
-
+                const long nLastIndex = pReadAccess->Height() - 1;
                 long nStripStart = 0;
-                long nStripEnd = nStripStart + constBlurThreadStrip - 1;
-
-                while (nStripEnd < nEnd)
-                {
-                    std::unique_ptr<BlurTask> pTask(
-                        new BlurTask(pTag, pBlurHorizontalFn, aSharedData, nStripStart, nStripEnd));
-                    rShared.pushTask(std::move(pTask));
-                    nStripStart += constBlurThreadStrip;
-                    nStripEnd += constBlurThreadStrip;
-                }
-                if (nStripStart <= nEnd)
+                for (; nStripStart < nLastIndex - nThreadStrip; nStripStart += nThreadStrip)
                 {
-                    std::unique_ptr<BlurTask> pTask(
-                        new BlurTask(pTag, pBlurHorizontalFn, aSharedData, nStripStart, nEnd));
+                    long nStripEnd = nStripStart + nThreadStrip - 1;
+                    auto pTask(std::make_unique<BlurTask>(pTag, pBlurHorizontalFn, aSharedData,
+                                                          nStripStart, nStripEnd));
                     rShared.pushTask(std::move(pTask));
                 }
+                // Do the last (or the only) strip in main thread without threading overhead
+                pBlurHorizontalFn(aSharedData, nStripStart, nLastIndex);
                 rShared.waitUntilDone(pTag);
             }
             {
                 Bitmap::ScopedReadAccess pReadAccess(rBitmap);
                 BitmapScopedWriteAccess pWriteAccess(rBitmap);
-
                 BlurSharedData aSharedData(pReadAccess.get(), pWriteAccess.get(), nRadius,
                                            nComponentWidth, nColorChannels);
 
-                const Size aSize = rBitmap.GetSizePixel();
-                long nEnd = aSize.Width() - 1;
-
+                const long nLastIndex = pReadAccess->Width() - 1;
                 long nStripStart = 0;
-                long nStripEnd = nStripStart + constBlurThreadStrip - 1;
-
-                while (nStripEnd < nEnd)
-                {
-                    std::unique_ptr<BlurTask> pTask(
-                        new BlurTask(pTag, pBlurVerticalFn, aSharedData, nStripStart, nStripEnd));
-                    rShared.pushTask(std::move(pTask));
-                    nStripStart += constBlurThreadStrip;
-                    nStripEnd += constBlurThreadStrip;
-                }
-                if (nStripStart <= nEnd)
+                for (; nStripStart < nLastIndex - nThreadStrip; nStripStart += nThreadStrip)
                 {
-                    std::unique_ptr<BlurTask> pTask(
-                        new BlurTask(pTag, pBlurVerticalFn, aSharedData, nStripStart, nEnd));
+                    long nStripEnd = nStripStart + nThreadStrip - 1;
+                    auto pTask(std::make_unique<BlurTask>(pTag, pBlurVerticalFn, aSharedData,
+                                                          nStripStart, nStripEnd));
                     rShared.pushTask(std::move(pTask));
                 }
+                // Do the last (or the only) strip in main thread without threading overhead
+                pBlurVerticalFn(aSharedData, nStripStart, nLastIndex);
                 rShared.waitUntilDone(pTag);
             }
         }