On Thu, May 20, 2010 at 5:40 PM, bugzilla-daemon@freedesktop.org wrote:
https://bugs.freedesktop.org/show_bug.cgi?id=27901
--- Comment #8 from Alain Perrot alain.perrot@gmail.com 2010-05-20 17:40:21 PDT --- Created an attachment (id=35777) View: https://bugs.freedesktop.org/attachment.cgi?id=35777 Review: https://bugs.freedesktop.org/review?bug=27901&attachment=35777
Alternative assemble_TRIG fix
I can confirm that your patch seems to work for me too.
By the way, you beat me at posting a working patch here :-)
I also figured out that the 0.5 special constant was an issue in your patch, and I managed to get a working assemble_TRIG function which implements the following instruction sequence (lightly different of yours) to normalize the angle:
MULADD tmp.x, angle, 1/(2*PI), 0.5 FRACT tmp.x, tmp.x ADD tmp.y, tmp.x, 1 CNDGE tmp.x, tmp.x, tmp.x, tmp.y MULADD tmp.x, tmp.x, 2*PI, -PI
I don't known if it is better or worse than yours beside the fact that it use only one helper variable.
I attached my patch (updated to use the same extended value of PI than yours) which fix the assemble_TRIG function, but not the assemble_SCS one.
-- Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug. _______________________________________________ dri-devel mailing list dri-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/dri-devel
Alain,
Its a tough call on who's is the better solution. Yours uses one less temp reg and mine will allow for a couple of operations to be done in parallel in the future. I guess we both deserve a pat on the back and leave it to someone more experienced to make the call on which one to choose.
Good job
Conn