Re: [Plib-users] Multithreading in FG
Brought to you by:
sjbaker
From: Hemalatha S. <hem...@re...> - 2006-08-31 04:36:05
|
=A0=0AHi,=0AThanks for U reply .=0AIt would be very kind of u if u tell ho= w to sync to vblank.=0A=0ARegards,=0AHemalatha=0A=0AOn Wed, 30 Aug 2006 Cur= tis L.Olson wrote :=0A>Hemalatha Sharma wrote:=0A>>=0A>> Hi,=0A>>Yes U are= correct!=0A>>There are 3 separate physical displays each with it's own pro= gram and each receiving network input from one system(each system has to pr= ocess and render independently)=0A>>=0A>>yes "send a done signal" mean that= the left and right programs send a=0A>>network message to the center syste= m( we are using here a flag=3D0/1 concept)=0A>>=0A>>The main problem if the= y are not synced is -All three views keep updating in different rates which= is not good to visualize)=0A>>By synchronization i mean all three views sh= ould update at same rate=0A>>[ie when u are rolling imagine how center ,lef= t ,right view will look updating at diff rate]=0A>>=0A>=0A>I've had pretty = good luck synchronizing 5 "slave" displays from a 6th "master" machine usin= g the standard FlightGear facilities. If you sync to vblank and have enoug= h horse power to sustain 60 fps on all machines, then this ends up working = really well. I've done it with monitors and LCD projectors. On one simula= tor we did this with 6 forward "monitor" type displays. Note that you can = get away with a bit of error with monitors because there is a gap between d= isplays that helps hide discrepancies. That's much harder with tiled proje= cted displays.=0A>=0A>Curt.=0A>=0A>-- Curtis Olson http://baron.flig= htgear.org/~curt=0A>HumanFIRST Program http://www.humanfirst.umn.edu/=0A>F= lightGear Project http://www.flightgear.org=0A>Unique text: 2f585ee= ea02e2c79d7b1d8c4963bae2d=0A>=0A |