On Mon, 2 Jul 2012 20:54:18 +0200 Nemanja Lukic <nlu...@mips.com> wrote:
> Added optimizations for several bilinear fast paths: > - src_8888_8_8888 > - src_8888_8_0565 > - src_0565_8_x888 > - src_0565_8_0565 > - add_8888_8_8888 > - src_8888_8888 > - src_8888_0565 > - src_0565_8888 > - src_0565_0565 > - over_8888_8888 > - add_8888_8888 > Benchmark results (using lowlevel-blt-bench) on Malta board (@1Ghz) > are included in the log messages. Per previous code review: > - Assembly code now respects the interpolation precision settings > defined by BILINEAR_INTERPOLATION_BITS. > - "make check" pass when this flag is set to 7 (default) and 8. > Any comments to this patch are welcome. Thanks, pushed to pixman git. I just trimmed the summary strings, because you were about to set the new record for the commit summary length. The current record belongs to: http://cgit.freedesktop.org/pixman/commit/?id=5bcde57da704000f0fbb52291d01c7f3c0dc8655 While there is no special policy enforced (or I just don't know about it), I think it is reasonable to keep these strings short, so that their text properly fits: http://cgit.freedesktop.org/pixman/log/ Regarding the code itself. It is quite likely that these optimizations may become obsolete soon with the introduction of separable horizontal and vertical interpolation for bilinear scaling (which should allow even better performance). But there is no reason not to use the currently available optimizations until then. -- Best regards, Siarhei Siamashka _______________________________________________ Pixman mailing list Pixman@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/pixman