Qualify the static variables in pixman_f_transform_invert() with the const keyword.
d4aa82f
Use a compile-time constant for the "K" constraint in the MMX detection.
f9c91ee
In fast_composite_tiled_repeat() don't clone images with a palette
cd7ecf5
test: Make stress-test more likely to actually composite something
7b20ad3
sse2: add missing ABGR entires for bilinear src_8888_8888
4cdf8e9
loongson: optimize _mm_set_pi* functions with shuffle instructions
ef99f9e
mmx: optimize bilinear function when using 7-bit precision
9aa8e3a
mmx: add scaled bilinear over_8888_8_8888
1ad6ae6
mmx: add scaled bilinear over_8888_8888
c43de36
mmx: add scaled bilinear src_8888_8888
9209cd7
mmx: Use expand_alpha instead of mask/shift
51f27d7
Change default bilinear interpolation precision to 7 bits
b0855f0
sse2: _mm_madd_epi16 for faster bilinear scaling with 7-bit precision
c430b1d
Bilinear interpolation precision is now configurable at compile time
ccd3189
Fix distcheck due to custom iwMMXt rules
ad9f1d0
sse2: faster bilinear scaling (use _mm_loadl_epi64)
ff5d041
test: support nearest/bilinear scaling in lowlevel-blt-bench
fc162ba
test: Fix for strict aliasing issue in 'get_random_seed'
387e9bc
build: Fix compilation on win32
4cbeb0a
sse2: add src_x888_0565
21077e1
sse2: enable over_n_0565 for b5g6r5
7db07cb
.gitignore: add test/glyph-test
45946c5
test: Add missing break in stress-test.c
eadb442
test: fix bisecting issue in fuzzer-find-diff.pl
492dac7
test: OpenMP 2.5 requires signed loop iteration variables
40a0d10
test: Make glyph test pass on big endian
619a60d
bits-image: Turn all the fetchers into iterator getters
f80e7ad
Faster unorm_to_unorm for wide processing.
fd175f9
configure.ac: add iwmmxt2 configure flag
367b78f
autotools: use custom build rule to build iwMMXt code
31a6563