[karnute]

Frankie V, on 13 February 2014 - 10:29 PM, said:

Want a better ping find a server with the least number of hops. ;)

I concede that too many hops in route can lead to instability in ping etc, but that is not my point. My ping with server is almost stable and low (25ms) with 14 smooth hops (including my router, the server, and 5 very fast hops internal to my ISP with FTTH), but I feel the problem (worst no-hits, etc) increased in the moments when there are other players with much different pings (between them) even if my ping remains the same (low and stable).

So, in other words, if in the same server all players have a ping similar to mine, the problem is less noticeable, but in the moment there are other player with much more or much less ping than the mean of the other players, then the problem raises in magnitude.

This could be related to the code dealing with anti-lag or network compensation, etc. And during game, in the screen it is perceived as if the other players position were wrong predicted by the client (more than real movement) and after rendering were corrected by server, so some times players seem to strafe too fast side to side during play (so the no-hit or micro-warping feeling) but the demo only register the server true position (not the prediction by client).

This post has been edited by karnute: 13 February 2014 - 11:21 PM

[FS Frankie V]

karnute, on 13 February 2014 - 11:17 PM, said:

Stuff

Gotcha

Personally I can not see a solution to this as a problem using the current Netcode system available in idtech3 with out creating a much bigger gap between the have and have nots. To be honest I'm the last person to talk about Netcode but people smarter than me in this area have stated many times it's something not to be messed with but replaced.

The solution to the problem though is to get into the 21st century and put all in the client but then cheating in Urban Terror would go through the roof.

[Mod ObScUrE]

I have the same problem as Karnute and it has nothing to do with the netcode.
I get this stuttering even offline on stable 125 fps (same with vsync enabled or disabled).
The best fix so far is to set s_mixprestep and s_mixahead to pretty low values. (this has the most impact on the stuttering for me)

s_mixahead "0.04"
s_mixPreStep "0"

with dmaHD it feels like the input is handled faster then the visual feedback were on the default sound engine it feels more like they are in sync but I'm still not happy with the end result.


-
CPU: i5 3570k
GPU: intel HD4000
RAM: 8GB
SSD: Samsung 256GB
Monitor: Benq @ 120Hz
OS: Win7

[karnute]

ObScUrE, on 14 February 2014 - 03:51 AM, said:

I have the same problem as Karnute and it has nothing to do with the netcode.
I get this stuttering even offline on stable 125 fps (same with vsync enabled or disabled).
The best fix so far is to set s_mixprestep and s_mixahead to pretty low values. (this has the most impact on the stuttering for me)
s_mixahead "0.04"
s_mixPreStep "0"

I have tested again dmaHD enabled with s_mixahead "0.04" and s_mixPreStep "0". The result was very good only with all players in server around or below 50ms ping (in scoreboard list), but just in the moment a player with ~100ms entered, then my hits (against any player) were worse and more random, and I start perceiving all players like stuttering/micro-warping/random-fast-strafing, etc. When there are any player with higher ping (130~150ms), the problem is even worse.

I checked again outside-game my ping (in 50 tries) to server has mean value 24.47ms solid stable with mean-deviation 0.2ms. The ping for server shown in-game-browser was 25ms. My ping in game (upper-right corner) was almost fixed in 24ms, but the scorelist shows my ping as 48 (by the way, is there an explanation to this discrepancy?).

All the time I had solid 125fps (no noticeable drops even in uptown).
I use Linux (Debian-jessie-testing) 64-bits and I run UrT official client (Quake3-UrT.x86_64) in a dedicated X-server (no desktop, no Gnome, no KDE).
CPU=Core-2-Duo(E8500 @3,16GHz), RAM=4GB, HD=S-ATA-7200rpm.
GPU=GeForce9800-GTX, Monitor=1920x1200@60Hz.

This post has been edited by karnute: 15 February 2014 - 02:59 PM

[phd]

karnute, on 13 February 2014 - 11:17 PM, said:

...
So, in other words, if in the same server all players have a ping similar to mine, the problem is less noticeable, but in the moment there are other player with much more or much less ping than the mean of the other players, then the problem raises in magnitude.
...

Is it possible that the server as more to process when multiple players have noticeably different pings?
Has the part of the server code been touched since 4.2?
Has the server CPU usage increased since 4.2?

[l3on]

deleted

This post has been edited by l3on: 05 March 2014 - 12:21 PM