On Tue, 19 Apr 2005 14:47:18 -0500, Curtis wrote in message <[EMAIL PROTECTED]>:
> Arnt, > > You should be aware that scenery building is much more of a data > shuffling job, and much less of a cpu intensive job. The big > bottlenecks will be your network bandwidth and server disk IO. ..ah. Server disk OI I can shot down doing everything on a 25 GB ramdisk, but that too takes network bandwidth. Question is how much do I need of each. ..and you forgot to tell me what kinda machines you used, so I assumed you used 2 of "my" Celeron 850's. ;o) > Curt. > > > Arnt Karlsen wrote: > > >On Tue, 19 Apr 2005 13:20:13 -0500, Curtis wrote in message > ><[EMAIL PROTECTED]>: > > > > > > > >>Arnt Karlsen wrote: > >> > >> > >> > >>>..Curt, I need an idea of how much cpu work, building the scenery, > >>>is. What kinda machine(s) did you use, and how long did it take > >to >>build the scenery? > >>> > >>> > >>> > >>> > >>I haven't timed the latest builds real close, but figure if you > >throw >a couple machines at it in parallel, it's going to take you > >at least > > >> > > > >..these are 2 recent machines? Specs? > > > > > > > >>a full 7 days (x 24 hours) to do the final assembly and crunching. > >>This doesn't include any of the data prep work (which could take > >>weeks if you start from scratch), nor does it include the airport > >>model generation which takes a day or so. > >> > >> > > > >..can do ;o), assuming you used 2 Celeron 850's, that makes it 2 > >hours. [EMAIL PROTECTED]:~ $ qalc 2*8*24*1700/320000 > >(1700 * 24 * 8 * 2) / 320000 = 2.04 > >1700 is a NTWAG BigoMips figure for a Celeron, 320,000 is for the > >cluster estimate, I will have to use some machines as switches too. > > > >..to build 12.6GB of scenery, I assume I simply do a rebuild of our > >last version, se we _can_ shoot for a 25GB target size, if we want it > >any bigger. Here I WAG the same cpu work per Gig of scenery, which > >probably is dead wrong. ;o) > > > > > > > > > >>And of course this doesn't count any of the time you need to spend > >>sitting down and sorting through tile build problems (or other > >>bugs/missing features) that you haven't gotten around to looking at > >>yet. > >> > >> > > > >..very true, and for a 4 hour stunt run, there will be _no_ such bug > >fix time, it will all have to be scripted, end to end. Watching my > >gasifier rig remains my top priority, a good 1/3 of my gas is CO, so > >I dont want _any_ leaks. > > > > > > > >>Regards, > >> > >>Curt. > >> > >> > >> > > > > > > > > > > -- ..med vennlig hilsen = with Kind Regards from Arnt... ;o) ..with a number of polar bear hunters in his ancestry... Scenarios always come in sets of three: best case, worst case, and just in case. _______________________________________________ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
