Martin Spott writes: > Gene Buckle <[EMAIL PROTECTED]> wrote: > > > flightgear.distributed.net. :) > > If someone provides a portable distribution mechanism for distributed > scenery generation, then I think I can provide some horsepower for this,
The big problem is that scenery building is much more slanted towards data shuffling (i.e. reading and writing files is typically the largest component of the task.) There is a computational component but it is generally small in comparison. When you think about distributing the tasks, the bottle necks are almost all in the file loading and saving. Locally I have about 220Gb of HD space dedicated towards storing the original raw data. The intermediate preprocessed form of the data. The shared edge data. And the final scenery. If we get SRTM data for the whole world, that will have to jump up substantially. For scenery building I'd love to have at least an 8-16 node cluster with really high bandwidth/ low latency net between them, a terrabyte of scsi disk space, and probably a big air conditioner to keep the room cool. :-) (Seriously, when you start thinking about building large clusters of PC's, you quickly get to the point where the largest cost of the entire system is the cost of keeping it cool.) But, in the absence of winning the lottery, I have to live with whatever hardware I can scrape together. :-) Curt. -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Cities [EMAIL PROTECTED] [EMAIL PROTECTED] Minnesota http://www.menet.umn.edu/~curt http://www.flightgear.org _______________________________________________ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
