On Sat, Oct 10, 2015 at 11:32:06PM +0200, Henrik Gramner wrote: > On Sat, Oct 10, 2015 at 11:06 PM, Ganesh Ajjanagadde > <gajjanaga...@gmail.com> wrote: > > This uses Stein's binary GCD algorithm: > > https://en.wikipedia.org/wiki/Binary_GCD_algorithm > > to get a reported 1.7-2.1x speedup over Euclidean GCD on standard > > architectures. > > Have not benchmarked, so I can't comment > > Before you submit a patch that's supposed to make something faster, > you should benchmark it to verify that it is in fact faster. Do this > with inputs of various sizes on both 32- and 64-bit architectures and > both with and without compilers that support __builtin_ctzll(v).
without __builtin_ctzll() the old code seems faster in a simple test
like:
make -j12 && ./ffmpeg -i matrixbench_mpeg2.mpg test.mov
also it seems a simple
while (u != v) {
if (u > v)
FFSWAP(int64_t, v, u);
v -= u;
do {
v >>= 1;
} while(!(v & 1));
}
is faster than the non builtin ctzll in its place but still not as
fast as the current code
[...]
--
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
What does censorship reveal? It reveals fear. -- Julian Assange
signature.asc
Description: Digital signature
_______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
