Hi Soren, > - The commit adding support to lowlevel-blt-bench should come before the > commit that adds the fast path. That makes it easy to get before and > after numbers.
I'll reorder commits. Since commit that adds support to lowlevel-blt benchmark only touches lowlevel-blt-bench.c file, that is only up to reordering of the pushes. > - Why are the pixbuf test cases only available as a compile time option? I decided to add them as a compile time option, since they don't cover big range of color formats (actually only couple) and operations. But, it can be implemented in different way. bench_composite function can check for pixbuf string in testname, and if that is detected, use same bits for src and mask images. Than, pixbuf testcases will not be only compile time option. Do you think that approach is better? Thanks, Nemanja Lukic -----Original Message----- From: "Søren" Sandmann [mailto:sandm...@cs.au.dk] Sent: Tuesday, March 19, 2013 1:50 AM To: Nemanja Lukic Cc: pixman@lists.freedesktop.org Subject: Re: [Pixman] [PATCH 9/9] test: add "pixbuf" and "rpixbuf" to lowlevel-blt-bench Nemanja Lukic <nemanja.lu...@rt-rk.com> writes: > Add necessary support to lowlevel-blt benchmark for benchmarking pixbuf and > rpixbuf fast paths. Trick is to force benchmark to use same bits for src and > mask images. This feature can be turned on using TEST_PIXBUF_FAST_PATHS build > flag (set to 1). By default it is set to 0. Two comments: - The commit adding support to lowlevel-blt-bench should come before the commit that adds the fast path. That makes it easy to get before and after numbers. - Why are the pixbuf test cases only available as a compile time option? Søren _______________________________________________ Pixman mailing list Pixman@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/pixman