[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: MDC1 Tests
Yes, I'm starting with 2000 now...
--Alex
David M. Malon wrote:
>
> >
> > These files are 600 MB!!!
> > couldn't even stage two at a time...
> > I started with 800 MB limit...
> > I may have to increase it to 1500...
> > what do you think?
> > --Alex
> >
> > Number of files already cached: 0
> > Will start thread to stage file psc079_01_46evts_dst.xdf.STAR.DB with
> > DBID 11
> > and size 616136704 bytes.
> > Stage request for file with FID 11 passed to thread.
> > Thread 8 will transfer file psc079_01_46evts_dst.xdf.STAR.DB
> > of size 616136704 bytes.
> > Multinode is Disabled.
> > Not enough space to stage file with FID 114.
> >
> > --
> > Alex Sim
>
> There was a proposal to use 2 GB as a cache size (see Arie's note),
> and I think this is a good starting point (I pointed out the problem
> with the 650+MB files in a reply). This may turn out to be too large
> for some of the queries, though, since some of the files are quite small,
> and some queries requiring 75+ files will be able to cache everything.
>
> To David Zimmerman: Are the queries that specify glb_trk_tot<10 likely
> to return events in small files? (I suspect that there may be a
> significant correlation between small numbers of tracks and small events,
> and this may in turn (here's the leap) yield files that are more likely
> than most to be small.)
>
> David Malon
--
Alex Sim
MS 50B-3238 http://sim.lbl.gov
DM, NERSC, LBNL ASim@lbl.gov
1 Cyclotron Road office: (510) 495-2290
Berkeley, CA 94720 FAX: (510) 486-4004