On 10/01/14 12:20 AM, Brett Mahar wrote:
On Thu, 9 Jan 2014 14:36:47 -0500
Brad Smith <b...@comstyle.com> wrote:
| Make use of Clang's integrated assembler to build the x86 ASM support
| within x264 so as to use MMX / MMXEXT / SSE / SSE2 / SSSE3 and if there
| is kernel support added also AVX / AVX2. This can have quite an impact
| on performance improving encoding speed by 2 - 3 times.
|
|
Just tried this out (with the newer mplayer diff that Brad sent a few days ago)
using mencoder to encode a video:
without this diff: 20m15.29s real 74m21.30s user 1m55.73s system
with this diff: 3m54.91s real 14m3.69s user 0m32.83s system
Brett.
Wow, I said 2 - 3 times based on the old crappy laptop I was using to
test this before. With not so busted hardware or anything more modern
it is obvious the improvement is that much more noticeable.
--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
