ORBIT-USER: Change CWMAX

[Haris Kremo]

Sachin,

what's the metric you've been using to determine whether CWmax works?

The theory (you know who...) claims that changing CWmax does not
affect much saturation throughput - which sounds reasonable. However,
the average number of retries per packet should change a lot,
especially for a large number of nodes.

H.

On 3/28/07, Johannes Berg <johannes at sipsolutions.net> wrote:
> On Wed, 2007-03-28 at 10:52 -0400, Sachin wrote:
>
> > CWMin seemed to work (in powers of two only)
>
> That "powers of two" can be expected, I've seen implementations of the
> random generation that do CWMin = 2^N-1, cw = (cwmin & random)
>
> johannes
>
>