Well, first I would like to apologize for starting The Thread That Never Dies.
<;-)> I am astonished to discover I can run a 500 tic(k) server. Probably needs whopping processor power to run it, but simulation processing would be so wickedly fast, that response packets might even be sent back to the clients before the client commands are actually received. Finally, the ultimate solution to lag: a psychic server! </;)> I think there may be two confusions at play here: my original post was about SRCDS and not HLDS. The second confusion seems to be between server FPS and tickrate, which as I understand it, are fundamentally different things. If I am wrong, please correct me, but I thought tickrate represents the number of simulation tasks that the server should perform in a single "cycle". That's not "ticks per second" - that's "tasks per simulation cycle" - which I guess to be the interval between when all the client inputs are received until the responses are dispatched. A high tickrate results in more precision in the simulation (e.g. collisions, clipping, physics, directional sound generation, object motion and location) but causes more calculations to be applied before the cycle is deemed complete and hence, makes the "cycles" longer. If the server processor is not up the task, this injects lag, because clients will be waiting for the simulation cycle to complete, before response packets are dispatched. A high tickrate server is not necessarily a performant one. FPS, if I understand correctly, is essentially the "read rate" for I/O operations. How frequently does the server check for inputs? Since this is expressed as fps_max, I am led to guess that the the speed of the simulation engine drives I/O demand. Hence, a high fps_max will not overwhelm a low tickrate server, however, a low fps_max setting will certainly choke a high tickrate server and result in an I/O bound process. Tickrate is a fixed server parameter that can be set only once - when the process is started, as an argument provided on the command line that starts the process. _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds
