On 08/10/2011 02:17 AM, Ronald S. Bultje wrote: > Hi, > > On Tue, Aug 9, 2011 at 5:16 PM, Luca Barbato <[email protected]> wrote: >> It just does that part in scalar form, I doubt using a vector store >> over 2 array would speed it up particularly. >> >> The function should be written to not use a scratch buffer. >> --- >> libswscale/ppc/swscale_altivec.c | 21 +++++++-------------- >> 1 files changed, 7 insertions(+), 14 deletions(-) > > Still LGTM, and still would be nice if you could add a FIXME to > rewrite the pieces that you just replaced to be vectorized again.
Depending on the width this way could be faster, removing the scratch buffer should make the whole thing faster no matter the width. lu -- Luca Barbato Gentoo/linux http://dev.gentoo.org/~lu_zero _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
