Well, I finally started to copy the 800 K database N times onto itself in a new database.
Up to 2.49 M games sorting and compaction runs fine. That is, acceptable, I did not do real benchmarking. At the next attempt, with 3.32 M games, the time is exploding, I guess somewhere in between those stages, that is. Currently, after running for 11 minutes, the estimated completion is at 54:54 minutes. I guess I am not willing to wait for that. Who bids more (or rather, less)... Cheers, Joost. Joost 't Hart wrote: > Puzzled indeed! > > I have tried several things > > 1) I noticed that - I think after installing eclipse - my system > 'suddenly' supports zlib. Forcing usage of scid's zlib makes no difference > > 2) I do transitions from Y,M,S,E,D,R sorting (top default) to E,R > sorting (bottom default) and vice versa. The sorting itself works > equally well in both directions on both linux and PC. > > 3) The compaction thereafter is a nightmare on linux (it will take > hours, if it ever completes, no matter what sorting has been applied). > On windows the compaction after ER sorting is painful in the beginning, > but speeds up after a struggle for 2 minutes, to complete in about 4 > minutes. > > 4) Compaction without re-sorting, but after only deleting a few games, > goes fine on Linux. > > 5) Hence, for some reason physically re-ordering the games on linux is a > nightmare. > > Unix aficionados, please check the above as well and report your findings. > > Some factor of importance must the database size, which is in my case > about 3.8 M games (sg3 file size about 418 MB). > The same stuff works fine on linux for a database with 800 K games (sg3 > ~ 69 MB). > > Cheers, > Joost. > > Joost 't Hart wrote: > > >> [cvs - Ubuntu Linux] >> >> Hi, >> >> Today I noticed that gamefile compaction of my big dbase (after a >> re-sort) does not work - in practice. >> >> Immediate throughput time estimates run into something like 225 minutes >> and after 45 minutes of 'crunching' the estimation goes up in line with >> the progress. Nothing to be done but to abort this fruitless operation. >> >> From the system monitor I see that scid receives only 3-4 % of the CPU, >> and that it is continuously in sync_page state; that is, waiting for a >> disk write operation to complete. >> >> I copied the database to my windoze box and the same compaction goes in >> 3 minutes and a bit, using the 3.7.1 release.. >> >> I fear that there is something rotten in my Linux build, in the tcl >> packages that are being used or even in the kernel. Although the latter >> does not seem very likely, since I do not see trouble like this with >> other software I use. >> >> Anyone a clue? >> >> Cheers, >> Joost. >> >> ------------------------------------------------------------------------------ >> _______________________________________________ >> Scid-users mailing list >> Scid-users@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/scid-users >> >> >> > > > ------------------------------------------------------------------------------ > _______________________________________________ > Scid-users mailing list > Scid-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/scid-users > > ------------------------------------------------------------------------------ _______________________________________________ Scid-users mailing list Scid-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/scid-users
