-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hello,
Thanks for quick answer, it provides me with some "ammunition" for the further discussions. Don Burns wrote: > Hi Jan, > <snip> > These trade-offs are quite application dependent. For a flight simulator, > for example, the scene itself stays quite static and the data > distributed is > really down to a few matrices, time, frame number, and the eyepoint. Each > node runs an independent viewer, but all are synchronized in "data sync". > Even if a paged database is used, a copy of the pager can be run on each > node. The database data can be paged from either a copy of the database on > each node, the network, or a SAN. Yes, for this case I could imagine how it could work. This is the "easy" case. > Now, if you want to be able to write a single program and running on one > node, update the scene graph dynamically from that one node and have all > nodes synchronously update the scene graph, then, no, this capability is > not > in OSG or Producer today. I don't know the OpenSG approach for doing this > although I do know that they have "cluster support". I am not familiar with OpenSG myself, however their web site claims to have support for data serialization and synchronization over the cluster machines, so it is probably supposed to support this type of applications. > So, the question needs to be asked more specifically. Rather than a > general, "cluster support", what level of data cohesion is really needed in > the application. It may only be a half a day of programming to make it > work. > What we are looking at is a simulator for psychological evaluation/treatment of soldiers returning from the battlefield. The typical scenario is something like a humvee being driven somewhere in Iraq, suddenly an insurgent jumps out and fires an RPG into the vehicle. Or an IED explodes destroying another humvee in front. Or something else of similar kind happens. The point is to simulate the traumatic situations the soldiers were in before and which are causing them psych. problems in order for the therapist to be able to treat them. However, it will not be a battle simulator/trainer, it should be realistic but the requirements are too extreme. The system should run in a CAVE where the clustering could be required (could - I am not familiar with their system yet). The problems then will be how to synchronize the movement of vehicles/insurgents or projectile trajectories between the machines. This should be possible to do on the scene graph level, however that takes more than just viewpoint synchronization. We could probably do an ad-hoc solution synchronizing only the relevant nodes in the scene graph, but I would prefer a more generic solution if there is one. Regards, Jan - -- Jan Ciger GPG public key: http://www.keyserver.net/ -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.5 (GNU/Linux) Comment: Using GnuPG with Mandriva - http://enigmail.mozdev.org iD8DBQFFKsdWn11XseNj94gRAt8tAJ9OgKMkVZXV2GGNZ925TfZjP6ysQgCgo/yh KsHRWUFRDen41NKtLyGJEM0= =+Zgr -----END PGP SIGNATURE----- _______________________________________________ osg-users mailing list [email protected] http://openscenegraph.net/mailman/listinfo/osg-users http://www.openscenegraph.org/
