Question on Tickrate and Packets/sec for UT Server Admin G..

[Guest]

Archived from groups: alt.games.unreal.tournament (More info?)

This is a strange one, so I'm looking for some ideas.

I have a UT 451 version server only program setup on an old 550 MHz
k62+ MVP3 chipset based system, runs just fine.

I copied the entire UT server folder onto a new box, a 1.45Gig Tbird
based system using an ECS K7S5A mobo, should run the same right? It
doesn't. The 1.45Gig box only has 128 meg of ram, but with the UT
server running, there is no swap file activity, and there is about 25
meg of free unused ram. The 550 MHz box has 196 Meg of ram.

The old slow machine delivers a steady 25 packets/sec (tickrate is set
to 25) to any UT client machine on my LAN; the new 1.45G machine only
delivers 21-22 packets/sec even though tickrate is identical (in fact,
all files are identical). Both have 100 Mb NICs connected to my local
8-port 10/100 switch. I tried connecting to both UT servers from 2
other different machines running UT (my "gaming" machines), and got
the same results from either location connecting to the 2 UT servers.
Both UT servers are running win2k, sp4 with all patches and updates.
The new 1.45G Tbird box is a pretty much clean install, only a couple
of programs installed, such as Powerdesk and System Suite (from vcom).
In fact, the old 550 MHz machine is running a firewall also as it's my
ICS machine to share the broadband connection. The 550 MHz box also
has Powerdesk installed, and a few other programs, not many. In fact,
the old 550 MHz box also runs VNC in the background looking for a
connection over the local LAN. I just don't see any reason why there
should be any difference at all in packets/sec between the 2 servers.
The servers are using the exact same INI file settings.

BTW, a LAN speed test to the faster 1.45G server (using Sandra's
simple LAN speed test) yields a faster throughput rate than the old
550 MHz box.

Any insights are appreciated.

--
Best regards,
Kyle

 

[Grant]

Archived from groups: alt.games.unreal.tournament (More info?)

"Kylesb" <me@privacy.net> wrote in message news:2m1ahfFhrqnaU1@uni-berlin.de...
> This is a strange one, so I'm looking for some ideas.
>
> I have a UT 451 version server only program setup on an old 550 MHz
> k62+ MVP3 chipset based system, runs just fine.
>
> I copied the entire UT server folder onto a new box, a 1.45Gig Tbird
> based system using an ECS K7S5A mobo, should run the same right? It
> doesn't. The 1.45Gig box only has 128 meg of ram, but with the UT
> server running, there is no swap file activity, and there is about 25
> meg of free unused ram. The 550 MHz box has 196 Meg of ram.
>
> The old slow machine delivers a steady 25 packets/sec (tickrate is set
> to 25) to any UT client machine on my LAN; the new 1.45G machine only
> delivers 21-22 packets/sec even though tickrate is identical (in fact,
> all files are identical). Both have 100 Mb NICs connected to my local
> 8-port 10/100 switch. I tried connecting to both UT servers from 2
> other different machines running UT (my "gaming" machines), and got
> the same results from either location connecting to the 2 UT servers.
> Both UT servers are running win2k, sp4 with all patches and updates.
> The new 1.45G Tbird box is a pretty much clean install, only a couple
> of programs installed, such as Powerdesk and System Suite (from vcom).
> In fact, the old 550 MHz machine is running a firewall also as it's my
> ICS machine to share the broadband connection. The 550 MHz box also
> has Powerdesk installed, and a few other programs, not many. In fact,
> the old 550 MHz box also runs VNC in the background looking for a
> connection over the local LAN. I just don't see any reason why there
> should be any difference at all in packets/sec between the 2 servers.
> The servers are using the exact same INI file settings.
>
> BTW, a LAN speed test to the faster 1.45G server (using Sandra's
> simple LAN speed test) yields a faster throughput rate than the old
> 550 MHz box.
>
> Any insights are appreciated.
>

You said the network throughput on the new system was faster. But does your
test check for packet loss?

Here are a few things I would check:
- were there any good tweaks you did to the old system to make it run smoother?
If the new one is a fresh install it might need some tweaking
- swap network cables
- make sure the exact right network drivers are installed (similar drivers may
work but have some problems)
- make sure the network card isn't sharing an IRQ with anything else, that could
very easily cause this problem
- Use the same port on the switch you were using for the old system. Usually on
an 8 port switch the first 4 get higher priority if you have QoS enabled on it,
so a lot of network traffic during the game could cause that
- if your motherboard has chipset drivers for it, install them


There's most of the hardware related problems I can think of.

 

[Grant]

Archived from groups: alt.games.unreal.tournament (More info?)

"Kylesb" <me@privacy.net> wrote in message news:2m34liFih2v3U1@uni-berlin.de...
> About the only item you suggest that I've not tried is swap NIC cables
> and try a different port on the switch. All other items are properly
> addressed, all the correct drivers are being used, the fast box runs
> nicely AFAI can tell, thanks for the suggestions/tips.
>
> --
> Best regards,
> Kyle
>

Let me know how it goes. If it doesn't help maybe I can think of some more
things to try for you.

 

[Grant]

Archived from groups: alt.games.unreal.tournament (More info?)

"Kylesb" <me@privacy.net> wrote in message news:2m5c0cFjbevqU1@uni-berlin.de...

> Ok, here's an interesting update. I have 3 machines running CPUs
> faster than 1.4 Gig, and running a UT server on any of them gives the
> same results, the packets/sec does not match the tickrate, seems to be
> a 2-3 packets/sec difference (lower) than the set value. Do ya
> suppose it has something to do with the CPU speed?
>
> --
> Best regards,
> Kyle
>

I suppose it's possible, but somehow the machine being too fast doesn't quite
sound right. The old machine was only 550Mhz, right? Regardless of the number
of packets, does the game play smoothly and properly?

If you change the tickrate, do you still get the same number of packets/sec
missing?