aboutsummaryrefslogtreecommitdiffstats
path: root/src/core
Commit message (Collapse)AuthorAgeFilesLines
...
* | new edgelist builder that chops segments into clip-sized coordinates, to ↵Mike Reed2009-11-2012-161/+1159
| | | | | | | | | | | | | | | | | | | | | | | | | | | | avoid overflows The change is conditional at the moment inside SkScan_Path.cpp, USE_NEW_BUILDER. This chopping is meant to only pass in reasonable floats to the edgelist, so that we never overflow when converting to fixed-point. It also has the side-effect of speeding up edges to the left/right of the clip, since those become vertical lines, which are faster to walk over than the original curve segment. Future optimizations: throw away segments to the right of the clip, and tweak the edgelist walker to not require an explicit matching right-edge for the current run.
* | refresh from skia/trunkMike Reed2009-11-107-146/+33
| | | | | | | | | | | | support for SSE2 blitrow optimizations fix flattening in 2-point gradient fix rotation in colormatrix
* | add shaderproc32 for index bitmaps, neon version of 32->16+ditherMike Reed2009-10-272-0/+31
| |
* | filter preserves opaqueness, so remove work-around code in bitmap shaderMike Reed2009-10-262-23/+9
| | | | | | | | | | The work-around code forced us to take a slower path, which is unnecessary now that filter is precise for opaque inputs.
* | fix neon version of S32A_D565_BlendMike Reed2009-10-222-83/+93
| | | | | | | | remove unneeded file from mac utils
* | Merge change I15b269f4 into eclair-mr2Android (Google) Code Review2009-10-221-10/+31
|\ \ | | | | | | | | | | | | * changes: pull clipper fix from skia/trunk rev.404
| * | pull clipper fix from skia/trunk rev.404Mike Reed2009-10-221-10/+31
| | |
* | | clear xfermode in paint before calling rgb16 blitter chooser, so it knows ↵Mike Reed2009-10-221-11/+24
|/ / | | | | | | about the null mode
* | motorola neon patch for blitMask, used by text drawing.Mike Reed2009-10-201-1/+96
| |
* | moto neon opts for bitmap tilingMike Reed2009-10-192-61/+145
| |
* | refresh from skia/trunkMike Reed2009-10-165-17/+42
| |
* | refresh from skia/trunkMike Reed2009-10-133-194/+197
|/
* don't call the inherited saveLayer in record, but just call save.Mike Reed2009-10-011-1/+7
| | | | | | | | | During record, our clip is wide-open (in the bug's case, 40K pixels tall). If we call saveLayer with no bounds (as webkit does), we try to allocate an offscreen the size of the document, which can be arbitrarily big. This cause us to run out of memory. In reality, I think there is not benefit to calling through to saveLayer during record. On playback, we will of course call saveLayer, but then we are on a real device (screen) with finite bounds, and the offscreen allocation is reasonable.
* disable neon for perspective nofilter indicesMike Reed2009-09-291-1/+3
| | | | It is computing out-of-range indices. When I debug it, I'll renable it with the fix.
* refresh from skia, with blitrow_d32 bottlenecks for neon optimizationsMike Reed2009-09-237-139/+247
|
* uses doubles in selective places to improve matrix-inverse precision, to ↵Mike Reed2009-09-171-49/+33
| | | | | | help with subtle bitmap cracking when we're scaled but not filtered.
* roll in moto neon patchMike Reed2009-09-153-5/+1506
|
* don't convert to hairlines unless the matrix-applied width is < 1 in both ↵Mike Reed2009-09-081-15/+40
| | | | dimensions
* add decoding bitmaps to bench toolMike Reed2009-09-021-3/+32
| | | | refresh misc. skia files
* elegant fix for gaps in antialiased hairlines by preda. Don't reduce the ↵Mike Reed2009-08-281-1/+1
| | | | | | full-span count unless we know we're going to run a non-zero final proc for the endcap
* distinguish 32 and 16 bit const-in-Y flags for shaders, so we can still take ↵Mike Reed2009-08-273-5/+8
| | | | | | advantage of gradients' ability to pre-dither their cache.
* refresh from trunkMike Reed2009-08-262-3/+44
| | | | | - edge case fix in qsort compare proc (overflowing subtract) - special case 1x1 bitmaps to draw as a color (much faster)
* force no-hinting when the caller wants linear-textMike Reed2009-08-241-1/+11
|
* refresh from skia trunkMike Reed2009-08-246-17/+53
| | | | | | | - correctly respect dither for index bitmaps - fix copyTo to respect isOpaque() - sanity check in antipath (need to investigate root cause) - warning fix in bitmap sampler
* fix type that send affine matrix procs to the perspective case (much slower)Mike Reed2009-08-141-1/+1
|
* fix overflow in fill_sequential quad caseMike Reed2009-08-145-10/+53
| | | | | compute max count from buffer size for bitmap shader zero-init paint in constructor to account for padding so that our == works correctly using memcmp
* special case no scale in the matrixprocs for tiled bitmapsMike Reed2009-08-137-93/+347
| | | | | yields ~10% overall speedup also, refresh misc fixes in freetype and antipath from trunk
* merge from open-source masterJean-Baptiste Queru2009-08-071-0/+2
|\
| * Make/add virtual destructors to classes with virtual functionsJean-Baptiste Queru2009-08-051-1/+3
| |
* | conditionalize including cpu-features.h, since at the moment it is only ↵Mike Reed2009-08-071-1/+3
| | | | | | | | available for arm.
* | rearrange the neon filter code to be sharable among all of the "portable" ↵Mike Reed2009-08-054-89/+254
| | | | | | | | | | | | functions remove remaining special-case neon functions that are (no longer) faster than the portable ones
* | add arm/neon optimizations for bitmap samplingMike Reed2009-08-032-10/+30
| | | | | | | | original code submitted by ARM LIMITED 2009
* | add GetFileName api to fonthost, to return (optionally) the file backing a ↵Mike Reed2009-07-311-1/+1
| | | | | | | | | | | | given font minor refresh on freetype and blitrow.h from skia/trunk
* | add opts directory to hold arm/neon specific functionsMike Reed2009-07-302-17/+40
| | | | | | | | vector out to those from our blitrow factory
* | refresh from trunkMike Reed2009-07-279-73/+236
| | | | | | | | check-point for lcd text rendering
* | remove build conditional for mipmapsMike Reed2009-07-176-925/+5
| |
* | add fast-case for bitmap shadersMike Reed2009-07-165-9/+141
| | | | | | | | first stage of neon optimizations
* | respect cap-style for drawPointsMike Reed2009-07-141-3/+5
| |
* | handle intrinsic 16bit shaders smarter in blitRectMike Reed2009-07-104-24/+84
| | | | | | | | | | use expand16 format to speed up blend fix typo bug in loops for 4444
* | am ee38d710: catch underflow in restore()Mike Reed2009-07-091-9/+11
|\ \ | |/ | | | | | | | | | | Merge commit 'ee38d710c05d4a6a8d8a649f5a05b8380ba748bc' * commit 'ee38d710c05d4a6a8d8a649f5a05b8380ba748bc': catch underflow in restore()
| * catch underflow in restore()Mike Reed2009-07-091-9/+11
| |
* | conditionally set SK_SOFTWARE_FLOAT only if we're not building armv7-a, so ↵Mike Reed2009-07-082-4/+5
| | | | | | | | | | | | we can take full advantage of hw floats on v7 class devices.
* | fix null-shape crash in picturesMike Reed2009-07-074-13/+77
| | | | | | | | | | fix gradient interpolation w/ alpha (do it in nonpremul space) optimize index->16bit sprite blit
* | refresh from trunk: shape flattening, bzeroMike Reed2009-06-2913-76/+168
| |
* | am 323fc5f6: Fix a broken sim-eng build because of lacking of arm arch check.Wei-Ta Chen2009-06-241-5/+6
|\ \ | |/ | | | | | | | | | | Merge commit '323fc5f6b797daedeb80cf137de1f7abb1ce6ff3' * commit '323fc5f6b797daedeb80cf137de1f7abb1ce6ff3': Fix a broken sim-eng build because of lacking of arm arch check.
| * Fix a broken sim-eng build because of lacking of arm arch check.Wei-Ta Chen2009-06-251-5/+6
| |
* | am 9255d4e0: Bypass source-overing if the alpha value of a source color ↵Wei-Ta Chen2009-06-242-20/+142
|\ \ | |/ | | | | | | | | | | | | | | equals to 255, i.e., no need to add previous destination color values. Merge commit '9255d4e019df455c662b01c7030b5d9ec2376695' * commit '9255d4e019df455c662b01c7030b5d9ec2376695': Bypass source-overing if the alpha value of a source color equals to
| * Bypass source-overing if the alpha value of a source color equals toWei-Ta Chen2009-06-242-20/+142
| | | | | | | | | | | | | | | | | | | | | | | | 255, i.e., no need to add previous destination color values. My experimental results showed that the percentage of calls to SkSrcOver32To16() falling into the bypass path in the following processes are around: 1. system_server: 72% 2. acore: 80% 3. camera: 57% 4. browser: 66%
| * Automated import from //branches/donutburger/...@142335,142335Mike Reed2009-03-241-1/+2
| |
| * Automated import from //branches/donutburger/...@141451,141451Feng Qian2009-03-241-0/+8
| |
generated by cgit v1.1 at 2025-03-15 21:09:08 +0000