These are the test patches for switching to 7-bit bilinear interpolation precisions. The first patch makes bilinear precision configurable. The second patch tweaks SSE2 bilinear scaler for better performance using PMADDWD instruction. Both should be applied after: http://lists.freedesktop.org/archives/pixman/2012-June/002074.html
Compared to pixman-0.26.0, bilinear src_8888_8888 operation becomes almost 1.8x faster on Core i7. The firefox-fishtank trace gets ~30% speedup, which should help a bit with: http://lists.freedesktop.org/archives/pixman/2012-May/001942.html Also 7-bit bilinear interpolation precision still allows a nice SSSE3 optimization, which can be implemented later to get even more speed. But the patches need a lot more testing on different hardware. Siarhei Siamashka (2): Bilinear interpolation precision is now configurable at compile time sse2: faster bilinear scaling with 7-bit precision (use _mm_madd_epi16) pixman/pixman-arm-neon-asm-bilinear.S | 119 ++++++++++++------------ pixman/pixman-arm-neon-asm.S | 159 +++++++++++++++++---------------- pixman/pixman-bits-image.c | 16 ++-- pixman/pixman-inlines.h | 37 ++++++--- pixman/pixman-mips-dspr2-asm.S | 9 ++- pixman/pixman-private.h | 8 ++ pixman/pixman-sse2.c | 42 ++++++--- test/affine-test.c | 10 ++- test/scaling-test.c | 10 ++- 9 files changed, 236 insertions(+), 174 deletions(-) -- 1.7.3.4 _______________________________________________ Pixman mailing list Pixman@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/pixman