>
> So why not make the code rate-adaptive, and drop out of polled mode if the
> traffic is low enough?
>
> Like a 5 min sliding window of served calls, and then if the long baseline
> rate is sub-optimal..
Actually the way to do this (I was thinking about it last night) is
to set the select/poll timeout quite low if there's stuff waiting in the
async queue. The other way would be to actually make it event driven
rather than polling for finished tasks. This could probably be done by
opening up a dummy FD (say a unnamed pipe) and have each sub-thread write
a single byte to it when they're done. Then include that FD into the
regular poll/select set of FD's and we have event driven async-IO.
What do you think?
Cheers,
Stew.
-- Stewart Forster (Snr. Development Engineer) connect.com.au pty ltd, Level 9, 114 Albert Rd, Sth Melbourne, VIC 3205, Aust. Email: slf@connect.com.au Phone: +61 3 9251-3684 Fax: +61 3 9251-3666Received on Thu May 28 1998 - 17:01:35 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:40:29 MST