hamachi interrupt coalescing ioctl

[Eric Kasten]

The ioctl change and the inclusion of Pete's example ioctl program
are now in the hamachi-0.11 source available from:

      http://www.nscl.msu.edu/~kasten/perf/hamachi/ 
   or http://www.nscl.msu.edu/~kasten/perf/hamachi/src

      (if you're having trouble with frames)

...Eric
 

> Chris said:
> > Pete said:
> >> I've added a few lines to the hamachi driver to enable changing the
> >> parameters regarding interrupt coalescing by an ioctl.
> > What's "interrupt coalescing by an ioctl"?
> 
> A good example of poor placement of subordinate clauses in a sentence.
> 
> Check http://cesdis.gsfc.nasa.gov/linux/drivers/hamachi.html for a short
> description of the registers I'm playing with.  They control interrupt
> coalescing.  The way I play with the registers is using an ioctl().
> 
> Interrupt coalescing is a means by which the NIC can "save up" a bunch of
> pending interrupts and deliver them to the host all at once, avoiding the
> overhead of processing each one individually.  This comes at the expense
> of increased per-packet latency, though.
> 
> The tradeoff is CPU usage against packet latency.  Adding this ioctl()
> makes it easier to control that tradeoff during a connection, as opposed
> to just at module insertion time.
> 
> 		-- Pete
> ---------------------------------------------
> Pete Wyckoff          | wyckoff@ca.sandia.gov
> Sandia National Labs  | 925 294 3503 (voice)
> MS 9011, P.O. Box 969 | 925 294 1225 (fax)
> Livermore, CA  94551  |
>  | To unsubscribe, send mail to Majordomo@cesdis.gsfc.nasa.gov, and within the
>  |  body of the mail, include only the text:
>  |   unsubscribe this-list-name youraddress@wherever.org
>  | You will be unsubscribed as speedily as possible.
> 
> 

Eric Kasten
kasten@nscl.msu.edu
National Superconducting Cyclotron Lab
(517) 333-6412
 | To unsubscribe, send mail to Majordomo@cesdis.gsfc.nasa.gov, and within the
 |  body of the mail, include only the text:
 |   unsubscribe this-list-name youraddress@wherever.org
 | You will be unsubscribed as speedily as possible.