[FFmpeg-devel] [PATCH 1/2] add support for generic seeking by?reading timestamps in nuv

Joakim Plate elupus at ecce.se
Mon Sep 12 19:51:08 CEST 2011


On Mon, 12 Sep 2011 19:04:20 +0200, Michael Niedermayer wrote:

> On Mon, Sep 12, 2011 at 06:33:45PM +0200, Reimar Döffinger wrote:
>> [33 quoted lines suppressed]
> 
> its a resync function, it might come in handy to support damaged
> streams
> either way this is a minor thing, iam happy with any improvment
> 
>> [2 quoted lines suppressed]
> 
> of course
> 
> [...]

Here's the patch from that thread updated to new API's. My test file is
about 50 megs, and seeking to the end of it shows a marginal reduction in
cpu usage. I suspect the differense would be much more noticable on a
500meg file, but can't seem to find anything in that size.

Seekbacks seem like a bad idea to me. But maybe searching for jjjjjjjj as a
64bit value, remembering last 32bits of last read. When found check if last
32bit contain RT somewhere. If it does depending on position do a few byte
reads to check for given amount of trailing 'j'.. Would be more efficient,
but way more complex to implement.

/Joakim
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-add-support-for-generic-seeking-by-reading-timestamp.patch
Type: text/x-patch
Size: 3306 bytes
Desc: Attached file: 0001-add-support-for-generic-seeking-by-reading-timestamp.patch
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20110912/5a4e0841/attachment.bin>


More information about the ffmpeg-devel mailing list