we can't control any threads count that could drop the performance.
remove async() and will come back with fine-tuned threading-pool.
Change-Id: I17c39792234acfce6db334abc0ce12da23978a9a
these conversions are intented, we explicity use casting so that compiler doesn't
catch them anymore.
Change-Id: I9d905c7562c43929b040d034e5ee0d14c6750a80
Some user have no idea of premultiplied alpha concept,
We suggest more user-friendly interfaces so that they don't confuse it.
Now, this pre-multipying is acommplished by backend engines.
Change-Id: Ifd84d56361cb56a8b98240bbd16690accf370bad
we can adjust cell size if the cell memory is out.
the main rle logic missed the exception handling.
Change-Id: I4419eefefccafd788729111eafeb65aa4e6a20e9
previous fast track logic is useless,
it actually doesn't helpful for performance, just increase the code complexity.
Change-Id: Ib6ad204edfb241d74c41413dfec7ab42fb02af81
if the transform scale factor for x/y is not identical,
it keeps its both xy scale factor then apply them
for stroking calculation.
Change-Id: I519dfce3ce7b4a12c13da1801d6a00e139e7400f
Basically, stroke width size is linear,
engine couldn't apply scale factor from the matrix which contains 2 dimensional values.
Thus, we can apply it if the scale factor of x/y is identical.
Otherwise, we should transform every stroke points in the stroking process.
That scenario can be improved with another patch.
Change-Id: I070dcf29d2e42f21e182bdf4239781464158ef73
sw_engine simulates floating point by integer bit shifting,
it loses the accuracy while converting the number data.
This occurs the inacculated curve points result if it scales up very large size.
So we transform points before converting data in order to avoid losing the numbers less decimal point.
Change-Id: I0172e83f06b1a19143a2f65f667dc193e9a4396a
avx is the cutting edge method for intel & amd cpus simd instruction.
We are going to support this feature for the desktop environment (instead of sse)
You can turn on this with configuration something like this:
$meson . build -Dvectors=avx
Current patch supports only for raster solid color
Change-Id: I068ba30a1f63d480415e2762f8021fc8d6d28a39
we don't need to apply anti-aliasing if stroke is there.
here turns off anti-alias if stroke width is more than 2.
magic number 2 is experimentally confirmed.
Change-Id: I09031dc2a0a84f31c5904651ed1e62004645ba9a
if the rectangle is not transformed, we don't need to use rle method.
we can directly raster pixels onto the bounding box.
Change-Id: I4e8b57149c0bcd78124d09388bf5115093a43bee
this matrix data is volatile since it's coming from stack memory.
thus engine should record its own memory space for keeping it.
Change-Id: I664dd56412f4d236ad04c312220c67da226274e5
there are a logical wrong that memory increase size,
it correct that also tune the number to reduce memory allocation time.
Now scene transform test works smoothly.
Change-Id: If0674e33426d794546801a96ad9de711b5de0dcd
there are unnecessary partial spans generated in orthogonal rectangle.
we can merge those partial spans to others if they are on the same scanline...
Change-Id: I35a437a4f2eec106bd50f46f0390c652e617311d
memset() is more than 10x faster than manual loop.
Thus we replace it to manipulate buffer pixels.
Change-Id: If0f255578f7d49ff6704c4f15e2eefe435cc3c15
this bounding box max indicates absolute positions,
not width/height size.
because of this, shape couldn't rendered properly,
when it's clipped out.
Change-Id: I0afb0d6e63d1b511b83716c55f55e3fd5370fdb8
this interface is designed for user-defined arbitrary affine-transform paints.
required by svg loader.
Change-Id: I7f08023605d224e36cef3770098d3757aee81848
replace dynamic_cast to static_cast,
This can be a controversial choice between optmization and clean code.
Obviously we know the converting types,
try avoiding a bit heavier casting some cases.
Change-Id: Id763e6d1db449a229a492ab5b1a901a195936152
We prefer to build up a tiny compact engine at memory rather than compatibility,
this engine is not considerd for end-users but designed for middle-level framework
and some low-level users.
Thus, we won't consider 64bits data size,
use explicit 32 bits data until coming next upgrade...
Change-Id: I0704d5f1e0eb909cccc10922bc5972e115fbbcc0
Introduce Result type for notifying caller more detailed info.
We should implement the result values for each apis, with practical values.
Change-Id: Ia47abcb56a8efca7094ac3eed0178aeac8aa2910
