Well well well, isn't it funny how amazingly
conflicting people's experiences of the same situation can be.
In total I`ve received about two dozen mails regarding running
Q3a demo servers, and another
forty odd along similar lines with UT
demo, none of which actually agree.
Got a mail from Cliffy
B and one following immediately after from
Polge regarding the networking
patch Steve put out a couple of
weeks ago for the demo. Now some friends and I played around
with that tonight, and although connections felt a lot better
we still had the same problems with CPU spikes and hard disk
accessing making game play fall over when several clients came
into view of each other. Now this may just be us, cos so far
all of the people that mailed in to say they were having
similar problems with the Unreal
Tourney demo have since said they were getting much
better results after applying the patch, but there's still
some issues there on our setup that I'll look into a little
deeper and hopefully we'll have some recommendations for you
soon. This patch I might add is in the final so I guess we'll
find out some time in the next week or so anyway.
J
As for the Quake 3
Arena demo I`ve had some seriously conflicting mails on
the subject. A lot of people seem to be having problems
getting the com_hunkmegs setting to stick, which appears to
depend on where the setting is on the command line. With that
in place a server with hunk size 16, (the minimum available in
the next demo) runs at around the 38Mb mark, which by itself
is all fine and dandy. But this is where things start to get
confusing, as a dozen or so sysops with similar spec boxes are
finding vastly different results when running multiple
servers. Some are reporting nasty bouts of paging which
usually starts the first time one of the servers changes map
and then ping pongs continuously after. Some like
onethumb who is running on a
release candidate Win2K box are finding they can run other
programs on top of the two servers without any compliants from
clients. And all these people are running boxes around the PX
450-550 mark with a minimum of 128Mb. Now this surely can't be
right... And I`m sure 99% of this is box configuration.
Carmack updated regarding a
high CPU usage problem that he's since fixed, but this
shouldn't effect memory usage a great deal under the
conditions we're imposing... Considered the possibility of
m$ vm handling being at fault
and whether or not the box was set up to let the OS handle
it's own virtual memory allocation, but until we've played
more we'll have to wait and see.
It is nice tho, to get some kind of response
regarding the subject from both games companies. It's rare
anyone in this industry listens to it`s punters are we're all
supposed to be too thick to understand, and the two giants
show a commitment to their communities that others would learn
a great deal from.
Got a lot of mails asking about
Zoid`s CTF maps. Not in a
position to share I`m a afraid but I`m sure like anything at
id more will be available when
it`s deemed ready.
Anyway, back to the grind, will keep you
informed on what comes out of this rather controversial
debate, but in the mean time, there`s always
Tetrinet.
J