Improve GPU blur performance by clearing only when necessary. This gives a

1.8X speedup on the Blurs sample, and 2.3X on the BigBlur sample.

We don't need to clear while downsampling, since each step reads only the pixels
written in the previous step.  We can avoid destination clears before
convolution by disabling blending.  We also don't need to clear when upsampling,
since the upsample step also only reads pixels written by the convolution.  The
only clears we then need to do are on each side of the srcRect used for
convolution, and a 1-pixel border for bilinear upsampling.  Since our srcRect is
always offset to (0, 0), we only need to clear on the right and bottom.

Review URL:  http://codereview.appspot.com/4803048/



git-svn-id: http://skia.googlecode.com/svn/trunk@1941 2bbb7eff-a529-9590-31e7-b0007b416f81

2 files changed, 22 insertions, 10 deletions

diff --git a/gpu/src/GrContext.cpp b/gpu/src/GrContext.cpp
index ef5ad1157d..b829323b2e 100644
--- a/gpu/src/GrContext.cpp
+++ b/gpu/src/GrContext.cpp
@@ -1736,5 +1736,6 @@ void GrContext::convolve(GrTexture* texture,
     fGpu->setSamplerState(0, sampler);
     fGpu->setViewMatrix(GrMatrix::I());
     fGpu->setTexture(0, texture);
+    fGpu->setBlendFunc(kOne_BlendCoeff, kZero_BlendCoeff);
     fGpu->drawSimpleRect(rect, NULL, 1 << 0);
 }
diff --git a/src/gpu/SkGpuDevice.cpp b/src/gpu/SkGpuDevice.cpp
index c4e0ebecd1..45dac336ea 100644
--- a/src/gpu/SkGpuDevice.cpp
+++ b/src/gpu/SkGpuDevice.cpp
@@ -913,9 +913,7 @@ static bool drawWithGPUMaskFilter(GrContext* context, const SkPath& path,
     GrClip oldClip = context->getClip();
     context->setRenderTarget(dstTexture->asRenderTarget());
     context->setClip(srcRect);
-    // FIXME:  could just clear bounds
     context->clear(NULL, 0);
-    GrMatrix transM;
     GrPaint tempPaint;
     tempPaint.reset();
 
@@ -959,9 +957,6 @@ static bool drawWithGPUMaskFilter(GrContext* context, const SkPath& path,
         context->setRenderTarget(dstTexture->asRenderTarget());
         SkRect dstRect(srcRect);
         scaleRect(&dstRect, 0.5f);
-        // Clear out 1 pixel border for linear filtering.
-        // FIXME:  for now, clear everything
-        context->clear(NULL, 0);
         paint.setTexture(0, srcTexture);
         context->drawRectToRect(paint, dstRect, srcRect);
         srcRect = dstRect;
@@ -972,25 +967,41 @@ static bool drawWithGPUMaskFilter(GrContext* context, const SkPath& path,
     float* kernel = kernelStorage.get();
     buildKernel(sigma, kernel, kernelWidth);
 
+    // Clear out a halfWidth to the right of the srcRect to prevent the
+    // X convolution from reading garbage.
+    SkIRect clearRect = SkIRect::MakeXYWH(
+        srcRect.fRight, srcRect.fTop, halfWidth, srcRect.height());
+    context->clear(&clearRect, 0x0);
+
     context->setRenderTarget(dstTexture->asRenderTarget());
-    context->clear(NULL, 0);
     context->convolveInX(srcTexture, srcRect, kernel, kernelWidth);
     SkTSwap(srcTexture, dstTexture);
 
+    // Clear out a halfWidth below the srcRect to prevent the Y
+    // convolution from reading garbage.
+    clearRect = SkIRect::MakeXYWH(
+        srcRect.fLeft, srcRect.fBottom, srcRect.width(), halfWidth);
+    context->clear(&clearRect, 0x0);
+
     context->setRenderTarget(dstTexture->asRenderTarget());
-    context->clear(NULL, 0);
     context->convolveInY(srcTexture, srcRect, kernel, kernelWidth);
     SkTSwap(srcTexture, dstTexture);
 
+    // Clear one pixel to the right and below, to accommodate bilinear
+    // upsampling.
+    clearRect = SkIRect::MakeXYWH(
+        srcRect.fLeft, srcRect.fBottom, srcRect.width() + 1, 1);
+    context->clear(&clearRect, 0x0);
+    clearRect = SkIRect::MakeXYWH(
+        srcRect.fRight, srcRect.fTop, 1, srcRect.height());
+    context->clear(&clearRect, 0x0);
+
     if (scaleFactor > 1) {
         // FIXME:  This should be mitchell, not bilinear.
         paint.getTextureSampler(0)->setFilter(GrSamplerState::kBilinear_Filter);
         sampleM.setIDiv(srcTexture->width(), srcTexture->height());
         paint.getTextureSampler(0)->setMatrix(sampleM);
         context->setRenderTarget(dstTexture->asRenderTarget());
-        // Clear out 2 pixel border for bicubic filtering.
-        // FIXME:  for now, clear everything
-        context->clear(NULL, 0);
         paint.setTexture(0, srcTexture);
         SkRect dstRect(srcRect);
         scaleRect(&dstRect, scaleFactor);