The Render X extension can draw triangles as well as trapezoids, but
the implementation has always converted them to trapezoids. This patch
moves the X server's triangle conversion code into pixman, where we
can reuse the pixman_composite_trapezoid() code.
This function is an implementation of the X server request
Trapezoids. That request is what the X backend of cairo is using all
the time; by moving it into pixman we can hopefully make it faster.
This separates the test suite from the random gtk+ using test
programs. "demos" is somewhat misleading because the programs there
are not particularly exciting (with the possible exception of
composite-test which shows off all the compositing operators).
Scaling function now gets an extra boolean argument, which is set
to TRUE when we are fetching padding pixels for NONE repeat. This
allows to make a decision whether to interpret alpha as 0xFF or 0x00
for such pixels when working with formats which don't have alpha
channel (for example x8r8g8b8 and r5g6b5).
In addition to the most common case of not having any mask at all, two
variants of scaling with mask show up in cairo traces:
1. non-scaled a8 mask with SAMPLES_COVER_CLIP flag
2. solid mask
This patch extends the nearest scaling main loop template to also
support these cases.
Depending on CPU architecture, performance is in the range of 1.5 to 4 times
slower than simple nonrotated copy (which would be an ideal case, perfectly
utilizing memory bandwidth), but still is more than 7 times faster if
compared to general path.
This implementation sets a performance baseline for rotation. The use
of SIMD instructions may further improve memory bandwidth utilization.
Split this function into two, one that has a mask, and one that
doesn't. This is a fairly substantial speed-up in many cases.
New output of lowlevel-blt-bench over_x888_8_0565:
over_x888_8_0565 = L1: 63.76 L2: 62.75 M: 59.37 ( 21.55%) HT: 45.89 VT: 43.55 R: 34.51 RT: 16.80 ( 201Kops/s)
New output of lowlevel-blt-bench over_x888_8_0565:
over_x888_8_0565 = L1: 57.85 L2: 56.80 M: 54.14 ( 19.50%) HT: 42.64 VT: 40.56 R: 32.67 RT: 16.22 ( 195Kops/s)
Based in part on code by Steve Snyder from
https://bugs.freedesktop.org/show_bug.cgi?id=21173
New output of lowlevel-blt-bench over_x888_8_0565:
over_x888_8_0565 = L1: 55.68 L2: 55.11 M: 52.83 ( 19.04%) HT: 39.62 VT: 37.70 R: 30.88 RT: 14.62 ( 174Kops/s)
The fetcher is looked up in a table, so that other fetchers can easily
be added.
See also https://bugs.freedesktop.org/show_bug.cgi?id=20709
The next few commits will speed this up quite a bit.
Current output:
---
reference memcpy speed = 2217.5MB/s (554.4MP/s for 32bpp fills)
---
over_x888_8_0565 = L1: 54.67 L2: 54.01 M: 52.33 ( 18.88%) HT: 37.19 VT: 35.54 R: 29.40 RT: 13.63 ( 162Kops/s)
Instead of having each individual implementation decide which fallback
to use, move it into pixman-cpu.c, where a more global decision can be
made.
This is accomplished by adding a "fallback" argument to all the
pixman_implementation_create_*() implementations, and then in
_pixman_choose_implementation() pass in the desired fallback.
It seems to be relatively common for people to use development
snapshots of pixman thinking they are ordinary releases. This patch
makes it such that if the current minor version is odd, configure will
print a banner explaining the version number scheme plus information
about where to report bugs.
Removes useless variable declarations. This can only result in more
efficient code, as these variables where sometimes assigned, but
their values were never used.
Green Hills Software MULTI compiler was producing a number
of warnings due to incorrect uses of int instead of the correct
corresponding pixman_*_t type.
The mask_bits variable is only declared in a limited scope, so the
pointer to it becomes invalid instantly. Somehow this didn't actually
trigger any bugs, but Brent Fulgham reported that Bounds Checker was
complaining about it.
Fix the bug by moving mask_bits to the function scope.
radial-test is a port of the radial-gradient test from the cairo test
suite. It has been modified so that some pixels have 0 in both the a
and b coefficients of the quadratic equation solved by the rasterizer,
to expose a division by zero in the original implementation.
When fetching from destinations, we need to ignore transformations,
repeat and filtering. Currently we don't ignore them, which means all
kinds of bad things can happen.
This bug fixes this problem by directly calling the scanline fetchers
for destinations instead of going through the full
get_scanline_32/64().
Add two new iterator flags, ITER_IGNORE_ALPHA and ITER_IGNORE_RGB that
are set when the alpha and rgb values are not needed. If both are set,
then we can skip fetching entirely and just use
_pixman_iter_get_scanline_noop.
Introduce a new ITER_LOCALIZED_ALPHA flag that indicates that the
alpha value computed is used only for the alpha channel of the output;
it doesn't affect the RGB channels.
Then in pixman-bits-image.c, if a destination is either a8r8g8b8 or
x8r8g8b8 with localized alpha, the iterator will return a pointer
directly into the image.
The separate get_scanline_32() functions in solid, linear, radial and
conical images are no longer necessary because all access to these
images now go through iterators.
At this point these functions are basically a cache that the bits
image uses for its fetchers, so they can be moved to the bits image.
With the scanline getters only being initialized in the bits image,
the _pixman_image_get_scanline_generic_64 can be moved to
pixman-bits-image.c. That gets rid of the final user of
_pixman_image_get_scanline_32/64, so these can be deleted.
