sv_maxrate 25000 is indeed not enough. Unless you have a too slow
uplink (server-side) we have found no disadvantages when using
sv_maxrate 0, letting the client choose their own rate limit without
arbitrary limits imposed from our side.

2011/2/24 Björn Rohlén <[email protected]>:
> Hello Saul.
>
> I believe you are in error. The reason the clients choke, is because the
> CLIENT RATE of 25k is _not enough_ to accommodate all the delta- and full
> worldstate- packets sent from the server in "intense situations". You can
> stand in spawn on most servers and be perfectly fine, but as soon as you
> start to move into areas with action going on (verify this with netgraph 3
> or 4), you will notice the client getting chocked. This is everything from a
> chokevalue of 10 up to 100, which obviously is not good.
>
> Now, if you read my post carefully, you'll notice that when _I_ did it, I
> made the sv_minrate _FOLLOW_ the sv_maxrate, hence FORCING the clients above
> the default 25k value and also keeping the "highraters" to the same rate.
> This gave MUCH better performance on a 20 and 24 player server. As for 32
> player servers, I would believe the amount of bandwidth would be quite hefty
> to make it "acceptable", but oh well.
>
> My suggestion for REAL WORLD implementation was to NOT do it in this manner,
> but keep sv_minrate and sv_maxrate apart -- and let the clients who actually
> can DEAL with the increased number of packets, do so, while also making sure
> the "casuals" who could not care less about how "things work", would get a
> "fair" performance as well.
>
> 60-75k is about what a client on a 512kbit connection can handle without
> getting loss/choke due to get too many packets (mind you, the sv_minrate
> above 25k is to ensure the client gets ENOUGH packets, since the default is
> not enough).
>
> Go ahead, try it. Go on a well configured server, run around with 25k rate
> and notice the choke. Set your rate above 50k and notice how the choke
> dissipates, models move smoother, rockets are not fired "behind doors" and
> other totally whack stuff -- magic!
>
> I think that about covers it? This has been up on this list about 2 years
> ago, I would recommend a search in the archives if you want to know more
> in-depth details.
>
> -TheG
>
> On Thu, Feb 24, 2011 at 10:25 AM, Saul Rennison 
> <[email protected]>wrote:
>
>> How can a high maxrate value choke clients? :/ if they're choking they
>> just need to reduce their rate!
>>
>> On Thursday, 24 February 2011, Björn Rohlén <[email protected]>
>> wrote:
>> > Hello.
>> >
>> > Having fiddled with this particular issue since the beta-days, sv_maxrate
>> > 50000 is sufficient for a 20 player server, unless you want to experience
>> > choke and weirdness on the clients. I would presume 24 and even 32 player
>> > servers would require even more.
>> >
>> > I'd watch out for going over 80k though, unless you want to choke players
>> > with bad internet-connections like 512kbit downstream to death. We
>> actually
>> > tested this as well, since we had a 512kbit-player (hi snowie) and 75k
>> was
>> > the value we settled on.
>> >
>> > This, of course, is because you make sv_minrate follow sv_maxrate as to
>> > avoid having the default rate of the client (25k) being set.
>> >
>> > If you are cheap, you could set minrate to 30k and maxrate to 60k and
>> watch
>> > the awe of the server population when they suddenly notice a DRASTIC
>> > improvement in performance on your server.
>> >
>> > To put them into awe but still keep within acceptable limits, sv_minrate
>> > 45000 and sv_maxrate 75000 would probably do the trick.
>> >
>> > -TheG
>> >
>> >
>> > On Wed, Feb 23, 2011 at 11:14 PM, Vathral <[email protected]> wrote:
>> >
>> >> Well, I'd like to work on an issue that members says they've been
>> having.
>> >> There have been complaints of lag and such that I'd like to look into.
>> What
>> >> are the default rates to work off? If possible I'd like to see what
>> other
>> >> admins out there are running as I do not fully understand how to tweak
>> >> settings like fps, sv_maxrate, maxupdaterate, etc.
>> >>
>> >> Running a Xeon 3230 with 4gb of ram, CentOS5, 100 Mbps link. Running
>> >> Sourcemod/MM with as few plugins as possible. No services like apache,
>> >> mysql, logging disabled where possible.
>> >>
>> >> This is what I have so far on one server
>> >>
>> >> fps_max 300                   // Frame rate limiter
>> >>
>> >> sv_maxreplay 2
>> >>
>> >> sv_minrate 10000              // Min bandwidth rate allowed on server, 0
>> ==
>> >> unlimited
>> >> sv_maxrate 25000              // Max bandwidth rate allowed on server, 0
>> ==
>> >> unlimited
>> >>
>> >> sv_mincmdrate 33
>> >> sv_maxcmdrate 66
>> >>
>> >> sv_minupdaterate 33           // Minimum updates per second that the
>> server
>> >> will allow
>> >> sv_maxupdaterate 66           // Maximum updates per second that the
>> server
>> >> will allow
>> >>
>> >> Any suggestions? Yes theres Google but I trust what I've read on here
>> such
>> >> as with the whole FPS issue.
>> >>
>> >> _______________________________________________
>> >> To unsubscribe, edit your list preferences, or view the list archives,
>> >> please visit:
>> >> http://list.valvesoftware.com/mailman/listinfo/hlds_linux
>> >>
>> > _______________________________________________
>> > To unsubscribe, edit your list preferences, or view the list archives,
>> please visit:
>> > http://list.valvesoftware.com/mailman/listinfo/hlds_linux
>> >
>>
>> --
>>
>> Thanks,
>>  - Saul.
>>
>> _______________________________________________
>> To unsubscribe, edit your list preferences, or view the list archives,
>> please visit:
>> http://list.valvesoftware.com/mailman/listinfo/hlds_linux
>>
> _______________________________________________
> To unsubscribe, edit your list preferences, or view the list archives, please 
> visit:
> http://list.valvesoftware.com/mailman/listinfo/hlds_linux
>

_______________________________________________
To unsubscribe, edit your list preferences, or view the list archives, please 
visit:
http://list.valvesoftware.com/mailman/listinfo/hlds_linux

Reply via email to