Re: Poll-splitting

> Okay, now _this_ is a little odd....Further to my email about splitting
> poll() structures into groups according to the likelyhood of getting a
> ready-response, I thought..well..hey, tcp connections are buffered in
> the kernel. We can probably get _some_ performance improvement by only
> calling poll on (say) a third to a quarter of the descriptors per pass,
> and letting the kernel buffers take up the slack for the ones we pass
> over until a later poll.

FWIW, I looked into something similar to spread the load over
multiple processors on SMP systems.

-cw
Received on Mon Nov 29 1999 - 01:34:20 MST