> On Fri, Apr 27, 2012 at 8:09 AM, <[email protected]> wrote: >>> On Thu, 26 Apr 2012 19:42:52 -0700 (PDT) >>> [email protected] wrote: >>> >>>> > http://cgit.collabora.com/git/user/pq/wayland-demos.git/log/?h=simple-fixes >>> ... >>>> About "simple-shm: render meaningful diagnostics", Because >>>> paint_pixels >>>> process pixels one by one, the following may be better: >>>> time/32 -> time>>5 >>>> time/64 -> time>>6 >>>> time/16 -> time>>4 >>> >>> You're saying the compiler does not automatically do that? Dang. >>> I thought it would if given the chance by using nice constant divisors. >>> >>> Right, I didn't pay too much attention to performance, it is a debug >>> app after all. Can be optimised if someone has trouble with it. >>> >>> >>> Thanks, >>> pq >>> >> Yes. Current popular compilers may do this optimization. But ">>" may be >> better. Just suggestion. > > IINM the compiler will produce identical output using bitshift and > division when it's trivial to optimize away the division as in the > above case. IMHO using bitshift in these situations just adds > confusion if the purpose of the expression actually was an arithmetic > division. > > Jonas > Could you please explain what are IINM and IMHO? Very thanks.
Yan Wang _______________________________________________ wayland-devel mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/wayland-devel
