[mythtv] Audio-upmixing questions regarding #5900

[foo bar]

2008/11/18 Yeechang Lee <ylee at pobox.com>:
> I was misusing terminology, I guess. I was referring to recreating 5.1
> audio from native 5.1 audio when timestretching a HD recording to
>>1.0X. Whether Passive or either Active mode, CPU hits >90% usage and
> the playback stutters.

The upmix setting won't make any difference in this scenario - it's
not turned on.

The vast majority of the additional cpu usage when timestretching is
by the video decoder and (perhaps) deinterlacer - especially true for
HD. Worth checking that your fallback deinterlacer for HD is cheap
(linear blend, say). With that set my 2.2Ghz can do 1.2x 1080i with
re-encode of 5.1 at around 70% usage. The re-encode of the audio is
verging on negligible.

FYI, for me cpu usage by myth for that scenario breaks down (very) roughly as:

65% mpeg2 decode
12% linear blend
10% misc
7% audio reencode
6% audio stretch