Jump to content


sv_fps, snaps, max_Rates & unknown fates


Recommended Posts

I use the same provider as Amidala from Chop Shop (www.jediservers.com). When the word came down that we were allowed to raise our sv_maxRate from 5000 to 6000, I jumped at the chance. I was under the impression maxRate would only benefit those who could take advantage of it and not penalize those who couldn't.


Much to my dismay, my personal performance went into the toilet (I'm a dialup, 56k user). My pings didn't radically increase, but my packet losses sure enough appeared to. The game became virtually unplayable.


I spoke with my service rep. about it, and he informed me that by increasing sv_maxRate to 6000, my client/server discrepancy was now 2k over the default modem setting of 4000.


I've since lowered the server's maxRate to 4000. My connections have never been more stable or consistent.


As mentioned in other posts, I have experimented with sv_fps (settings of 60, 40, 25 & 20) and did not see a marked difference in collision detection (not that I was personally experiencing much of a problem in the first place). I didn't realize at the time (been informed since) that altering my snaps was necessary to take advantage of it. That said, my initial research implies that raising snaps above the default (20) could cause problems for modem users. :(


I'd appreciate more input from someone with a comprehensive understanding of networking where to go from here. Any knowledgeable advice would be most appreciated.




BTW, if you're new to this discussion, some articles that may be worth looking over can be found here...



TechSpot Tweak Guide for Jedi Outcast

Link to comment
Share on other sites

My new host www.escapedturkey.com let me monitor CPU usage while I varied sv_fps with real players playing on the server. Increasing sv_fps does increase CPU usage. It's hard to say how much because there are so many other factors affecting CPU usage at the same time, but it did go up when I increased sv_fps from 20 to 40, and went down when I decreased it back down to 20. Most clients were probably running snaps at default (20) so they probably didn't notice any difference, but it didn't increase pings or cause any other obvious problems.


I think in an ideal situation (enough CPU reserve, good bandwidth from server to client) sv_fps 40 and snaps 40 (twice the default for each) would result in more solid hit detection. Because the "world" would be updated twice as often (analogous to a client having a twice-as-fast framerate [FPS]) , instead of hitting someone where they used to be, or where they are predicted (extrapolated) to be, you hit them where they really "are" in the server's virtual "world". I am guessing that Raven made the defaults 20 to give a cushion for less-than-current hardware and not-so-great Internet connections. But if you have modern server hardware, are not sharing the box with too many other servers, have a host with a good connection to the Internet, and are serving mostly clients with broadband connections and good hardware, sv_fps 40 and snaps 40 are probably the optimum settings.

Link to comment
Share on other sites

Amidala, bit off topic but... What kind of service have you been getting with escapedturkey? I'm using jediservers at the moment and while it's usually fine, they do have a few details here and there that I don't like. For example I've been asking them time and time again to help me with server log files but I think they disable them on purpose and just plead ignorance.


The server also lags terribly at times for no discernible reason.


I was thinking of switching over to escapedturkey's services but I wanted some opinions from other administrators first... Do they provide some way to control command line arguments or something along those lines?

Link to comment
Share on other sites

Well, yeah, Raven has been against changing the sv_fps from the beginning. The reason why is because it's basically a dirty hack (that works) for a broader problem of the server needing to push more fps than the default 20.


Doom 3 is going to running at 60 fps for that exact reason.


So, it's either the dirty hack that works or poorer hit detection.


Had Raven been going for the optimal solution without just boosting the sv_fps requirements (they probably didn't have time), they would have created a seperate server code loop just for saber detection that runs at a higher fps. This is normally how other systems have accurate hit detection for hit detection is super critical (like for vehicle tires to correctly touch/react to the ground).

Link to comment
Share on other sites

I was there with Kurgan (since it was a test server of mine) and when I told him I had changed it to like 60 fps, he went all out against how adjusting it causes more problems than it fixes. so I set it back to the default(20 or 25 , whatever it is) and every know and then, I do adjust the maxrate depending upon who I am expecting on the server. if its all broadband, I jump it up to 6,000, but if I expect ISDN/56k users, I kick it have it no higher than 4,000. sometimes I even kick it down to 0, which I see no difference in performance, but others love it.

Link to comment
Share on other sites


This topic is now archived and is closed to further replies.

  • Create New...