stephenbrooks.orgForumMuon1GeneralStats generator status
Username: Password:
Search site:
Subscribe to thread via RSS
Stephen Brooks
2004-04-19 09:52:59
You may have noticed the Muon1 stats were off today.  That was because I was moving into a new office and they didn't get ethernet working until the afternoon.
Today I also made some changes to the result-reader code, so the freshly-compiled stats generator doesn't work, but I still have the old functional EXE that I'll be running until I get the bugs in the new code sorted out.  The main object was to raise the limitation of 5000 chars on the length of the 'design' line in a Muon1 result, because results in v4.4 could have more like 7700 chars.
I'm also going to try out a more compact binary format for sending - should be at least 5x smaller than text - which should save everyone some bandwidth.
Stephen Brooks
2004-04-20 09:12:35
Fixed it!  The stats are now running on the newly compiled code.

I did some preliminary tests on the binary results format, yielding:

Original (44`089`816 bytes)
+ NTFS compression (26`983`424 bytes)
+ RAR compression (6`895`699 bytes)

Binary (8`197`758 bytes)
+ NTFS compression (7`046`656 bytes)
+ RAR compression (4`749`347 bytes)

Binary/16bit (12`121`501 bytes)
+ NTFS compression (10`263`040 bytes)
+ RAR compression (5`326`776 bytes)

On someone's ~40MB result file.  The Binary/16bit was one where I byte-packed instead of bit-packed the genome parameters, wondering if the alignment on 2-byte edges would make RAR or NTFS compress it better, but it didn't. The NTFS compression is relevant since it's activated on the 160GB drive where the results are stored.  If you believe these figures, then that alone makes the drive effectively 260GB and if I managed to convert the whole database to the binary format the capacity would be effectively 1000GB.  We've used under 10GB so far but it won't get any smaller, so it is good to have some options.
2004-04-20 11:32:14

I've been blitzed from the stats... seems others have been as well, example: [SG] Slipstream
2004-04-20 11:37:09
and now I'm back... Confused
2004-04-20 14:03:41
It's something with 'fenix' and 'ashes'. No?
: contact : - - -
E-mail: sbstrudel characterstephenbrooks.orgTwitter: stephenjbrooksMastodon: strudel charactersjbstrudel

Site has had 17822834 accesses.