Re: (PM) Retrains causing ~13 second 'hang'? (fwd)
Alexander Blauvelt (online@olg.com)
Sun, 9 Nov 1997 03:06:00 -0500
> Now that you mention it, we've seen this too. Drives our Quake players
> nuts--this includes most of our technical staff, FWIW :)
>
> Every k56flex modem we've tried has done this (Diamond, Zoltrix and
> Motorola), from every location (3 residences and 1 analog line at the
> office). We see an excellent connection for a few minutes, followed by a
> long retrain (5-20 sec), repeating every few minutes. We've never
exactly
> timed it but 2-3min sounds consistent with our experience. For now we
are
> telling Quakers to revert to 33.6 until we figure it out. :-(
This is the nature of the server <-> client structure in quake. Quake
pretty much maxes out a 28.8/33.6 connection. Give it more bandwith coming
down, and it starts taking it -- making it faster. However, quake then
starts sending stuff back to the server, at 28.8 speeds, when its taking in
at 44.0 (or whatever the connect). In a nutshell it backs up. Solution:
33.6 or don't play quake. Solution 2: Call ID and bitch at them for
writing terrible netcode.
Alexander Blauvelt
Online Gateway
efnet #quake op
-
To unsubscribe, email 'majordomo@livingston.com' with
'unsubscribe portmaster-users' in the body of the message.