[mythtv] Suggestion for improving SLOW channel
ltd at interlink.com.au
Sun Sep 11 00:46:54 UTC 2005
Daniel Kristjansson wrote:
> On Sat, 2005-09-10 at 22:38 +1000, Lincoln Dale wrote:
>> i ran 0.17 for a while with setsockopt TCP_NODELAY to disable nagling.
>> i didn't verify whether it was a benefit but based on my perception of
>> delay, it shaved 100msec of various things.
> I think this would be safe for the RingBuffer data socket...
the TCP socket for frontend/backend communication could also do with this.
without this, any frontend's commands to the backend are automatically
delayed 100msec as typically those aren't a full TCP segment worth.
if this uncovers any race-conditions then those need to be fixed - one
shouldn't rely on nagling to mask a race.
i'll open a ticket & patch for these if folks would like ..
pretty easy 100msec shaving of time!
More information about the mythtv-dev