stable 1000fps hl engine without pb possible?

[Monk]

You can get ultra high FPS without touching gettimeofday/usleep or modifying the host_framerate or using any other plugins.

Regardless, touching usleep and modying it is altering the (physics, rcon stats, other side effects) of the engine, so it is lying to the engine.

I hope one day VALVe will remove the need to have 1 fps per tick in a stats so it will kindly make people just enjoy the game instead of blame the 'fps' of the server.

[BehaartesEtwas]

(04-24-2010, 08:08 AM)Monk Wrote:  Regardless, touching usleep and modying it is altering the (physics, rcon stats, other side effects) of the engine

nope, it doesn't. at least not for srcds.

Quote:so it is lying to the engine.

yes. it's lying to the engine in a sense. but since when is the engine a sentient being?

Quote:I hope one day VALVe will remove the need to have 1 fps per tick in a stats so it will kindly make people just enjoy the game instead of blame the 'fps' of the server.

wth? after reading this sentence I think you simply didn't understand what a "frame" is in hlds/srcds... it's not as simple as eliminating frames and the world is happy...

[war]

(04-08-2010, 02:58 PM)Peter_Pan123 Wrote:  mh

ok but how can i reach 1000fps without pingboost? with a special pre-lib like srcds?
Is it better so use a gameserver without?

-p

just put fps_max 0 and ull reach 1000fps

[Monk]

(04-26-2010, 01:06 AM)BehaartesEtwas Wrote:  

(04-24-2010, 08:08 AM)Monk Wrote:  Regardless, touching usleep and modying it is altering the (physics, rcon stats, other side effects) of the engine

nope, it doesn't. at least not for srcds.

Quote:so it is lying to the engine.

yes. it's lying to the engine in a sense. but since when is the engine a sentient being?

Quote:I hope one day VALVe will remove the need to have 1 fps per tick in a stats so it will kindly make people just enjoy the game instead of blame the 'fps' of the server.

wth? after reading this sentence I think you simply didn't understand what a "frame" is in hlds/srcds... it's not as simple as eliminating frames and the world is happy...

Eliminating it and going back to a single method like a HZ steps physics similar to the quake engine games is the best way to get rid of this horrible FPS crap that everyone is obsessed about. There is no evidence at all that stable FPS does anything worthwhile except suck major amounts of CPU.

-M

[BehaartesEtwas]

ehm, the concept of server fps has been "invented" in the QuakeWorld engine. And if you call it HZ step or frame does not change anything. As far as I know there is no better concept than doing all calculations in regular time intervals and sleep in between to give other processes, drivers, etc. cpu time.

If I could change the engine I would change something else: currently the engine estimates the latency and puts the client commands into the command queue shifted by that amount of time into the past, to correct for the latency. much better would be to use a global tick number and let the clients tag their commands with the tick number to which they belong. That way the precise latency would no longer play any role, the server could put the commands into the queue at the right point in time without any uncertainties from latency estimation. Maybe this would open some additional doors for cheaters, but there are so many ways to cheat that this simply would not matter (at least if some cross checks are done with the estimated latency). the concept of fps would remain but it would no longer be important to have high fps. fps equal the tickrate would be perfectly fine. (at the moment it's not, because it introduces an additional uncertainty of 10ms for the ping - even if you, Monk, still neglect this, it's there!)

[Peter_Pan123]

sooo now i have found some time to try it.

951 fps -- host( 1) sv( 0) cl( 0) gfx( 0) snd( 0) ents(112)
948 fps -- host( 1) sv( 0) cl( 0) gfx( 0) snd( 0) ents(112)
941 fps -- host( 1) sv( 0) cl( 0) gfx( 0) snd( 0) ents(112)
939 fps -- host( 1) sv( 0) cl( 0) gfx( 0) snd( 0) ents(112)
924 fps -- host( 1) sv( 0) cl( 0) gfx( 0) snd( 0) ents(112)
920 fps -- host( 1) sv( 0) cl( 0) gfx( 0) snd( 0) ents(112)
927 fps -- host( 1) sv( 0) cl( 0) gfx( 0) snd( 0) ents(112)
940 fps -- host( 1) sv( 0) cl( 0) gfx( 0) snd( 0) ents(112)
926 fps -- host( 1) sv( 0) cl( 0) gfx( 0) snd( 0) ents(112)
926 fps -- host( 1) sv( 0) cl( 0) gfx( 0) snd( 0) ents(112)

i got these result with 2.6.33 kernel on a quad core

no changes when i use sys_ticrate 1000 or 10000