stephenbrooks.orgForumMuon1Bug ReportsQuestion on DPAD behavior
Username: Password:
Search site:
Subscribe to thread via RSS
[TA]BlackMountainCow
2004-03-23 07:46:25
I just set up a laptop with 4.34. Samplefile and latice are downloaded.  Now the following happens:
I've got 500 results from that machine so far, but non has a higher transfer than 0.6 and none has more mpts than 30. Did I miss something??  I did the same setup on 4 other machines and they all behave normal (=results from 1 to 310 mpt with variations)

Thx

Christian
slayre
2004-03-23 19:55:33
First off, be careful that you don't burn out that laptop.  Laptops are made just to be portable, not good computers.  I found this out that hard way when I burned out my laptop from running SETI's client for a few weeks.
Secondly, I'm sure that Stephen needs a variety of results for this project, not just ones with higher yields.  There probably isn't anything that you missed, its just chance that your yields are lower on that computer so I wouldn't worry about it.


Garrett
[TA]BlackMountainCow
2004-03-24 00:36:12
I know slyre!  Wink That laptop is only running DPAD when it's on use anyway, like when I sit in the library or so.  And even then I only use one of the HT processes.  It stays at 49-51 degrees and the fans just work on lowest speed. 

I know what u want to say about the variety of the project.  But we also got a member in [TA] that has over 60.000 results but a max yield of 5.6 which I think is not normal.  He uses the samplefiles of course and 4.34.
Stephen Brooks
2004-03-24 00:42:17
Maybe the sample files are not getting merged into the results.dat properly.  Check (i) that the URL used in config.txt does actually point to a sample file and if it does, (ii) that the checksum of a particular result in the sample file you downloaded also appears in results.dat.
[TA]BlackMountainCow
2004-03-24 05:55:11
I had the same idea Stephen.  So I downloaded the sample manually and merged it manually.  But it didn't help.

It just puzzles me that the very same DPAD folder on my 3200+ prodcues 300+mpts runs while on my laptop the very same folder only gives me 30mpts runs ...

Is there any CPU dependency with DPAD??  Like it uses some randomized numbers for new experiments and those CPU generated random numbers are different from AMD to Intel or even within certain Athlons?
Stephen Brooks
2004-03-24 08:16:46
Just to check something hasn't gone severely wrong - there isn't anything unusual about the graphical display on that Muon1 installation is there?  (Nothing like particles hitting an invisible wall that shouldn't be there and hence being stopped?) I guess you should also diff the l2y0.txt etc. data files with a working copy to make sure they're not corrupt.
[TA]BlackMountainCow
2004-03-24 09:47:30
I always ran the CLI version and never the GUI.  But the problem somewhat has solved itself as I had to give away that laptop again (was only borrowed) Frown When I get my hand on it again I'll try some more things.  Smile

[This message was edited by [TA]BlackMountainCow on 2004-Mar-24 at 09:47.]
[TA]JonB
2004-04-23 12:11:22
How much of the Muon program's "mutation" relies on random number generation?  Different systems will generate different series of random numbers, but some systems will always generate similar random numbers (not very random).  This can be because of a bad Real Time Clock or some other setting or component that the "seed" depends upon.
Stephen Brooks
2004-04-25 09:02:17
I seed the random number generator at the beginning of the program with the clock.  So if your clock is REALLY messed-up, your random numbers won't be good (or at least, the same every time).  I haven't seen a computer do that before, though.

Also on subsequent runs, the random number generator will be out of phase due to having been called differing numbers of times for things like particle decays and so on, plus some more differences coming from the fact more results will be in results.dat, so even if the first mutation of a run is the same in these cases, the later ones ought to be better.
: contact : - - -
E-mail: sbstrudel characterstephenbrooks.orgTwitter: stephenjbrooksMastodon: strudel charactersjbstrudel charactermstdn.io RSS feed

Site has had 25168068 accesses.