From 7c0925e3959b7d54b49e8b63cd11105de572ea4b Mon Sep 17 00:00:00 2001 From: Mike Reed Date: Thu, 17 Sep 2009 15:57:58 -0400 Subject: bump up the RLE cutoff, since RLE is *only* supported in drawBitmapRect(), and therefore we should only trigger it when there is no other way to support the image. Now that we have ashmem (yay!!!) memory pressure is not so bad. This happens to also fix google reader site, which sends down large (but not giant) index images but doesn't trigger our drawBitmapRect() code. --- WebCore/platform/graphics/android/ImageSourceAndroid.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'WebCore') diff --git a/WebCore/platform/graphics/android/ImageSourceAndroid.cpp b/WebCore/platform/graphics/android/ImageSourceAndroid.cpp index d33f94f..5c75a15 100644 --- a/WebCore/platform/graphics/android/ImageSourceAndroid.cpp +++ b/WebCore/platform/graphics/android/ImageSourceAndroid.cpp @@ -52,7 +52,7 @@ SkPixelRef* SkCreateRLEPixelRef(const SkBitmap& src); // don't use RLE for images smaller than this, since they incur a drawing cost // (and don't work as patterns yet) we only want to use RLE when we must -#define MIN_RLE_ALLOC_SIZE (512*1024) +#define MIN_RLE_ALLOC_SIZE (2*1024*1024) /* Images larger than this should be subsampled. Using ashmem, the decoded pixels will be purged as needed, so this value can be pretty large. Making -- cgit v1.1