stephenbrooks.orgForumMuon1Generalv4.2 is released, nobody knows if this will work
Username: Password:
Search site:
Subscribe to thread via RSS
Stephen Brooks
2002-07-27 10:31:36
Have a look in the Version History entry for v4.2 and see what you think.  I've given the program quite a hard problem to solve this time, but the potential reward is a completely new solenoid channel design: ALL the lengths, gaps, radii and magnetic fields can be varied, so it might come up with something very much better than we've had before for 4.x (nothing over 2.3% on v4.1x).


"As every 11-year-old kid knows, if you concentrate enough Van-der-Graff generators and expensive special effects in one place, you create a spiral space-time whirly thing, AND an interesting plotline"
Pascal
2002-07-27 10:37:00
Let's test and have a look at the results...

BTW: Are we allowed to copy the old results or am I confused now?  tell me please.

___________________________
1: Athlon TB-C, 1.2 GC/s, 256 MB DDR-RAM, Erazor x², ADSL-Flatrate, NIC Intel, Win 98 SE Mainboard MSI-6380 Rev.  1
2: Pentium III, 600 MC/s, 256 MB RAM, NIC Intel, Win 98 SE
Frank
2002-07-27 13:47:06
Hi Stephan

No way to keep the Version 4.2 running on my Computer.

Muon1.exe needs app.  200 sec.  for a WU, then the Progrogram terminate itself.
a second programstart is not possible. 

The background client crasches too.  frown

Bug in Result-Filesize 10times to big on window.

I'm back to 4.12

greetings from Switzerland

My System:
DualAthlon MP 1900+, 1024 MB DDR-RAM, Gforce3 Ti500, ADSL-Flatrate, Win XP Pro, Mainboard ASUS A7M266D, Promise Raid 280GB (CPU's & GPU Watercooled cool )
Frank
2002-07-27 14:10:43
uups, sorry, my fault !!!!!  big grin

do NOT copy the old result.dat file from Version 4.12 in the 4.2 directory !!!!!!!!!!

the program is stable now.
Stephen Brooks
2002-07-27 15:08:59
Yes - DON'T copy results from previous versions into this one!


"As every 11-year-old kid knows, if you concentrate enough Van-der-Graff generators and expensive special effects in one place, you create a spiral space-time whirly thing, AND an interesting plotline"
mcclearyironman
2002-07-27 18:32:41
Stephen Brooks wasn't kidding when he said the results for v4.2 would be low.  I got a result that was 0.00187% efficiency!!!!
Pascal
2002-07-27 22:15:08
Yesterday I got a result with 0.0000 percent!
Another thing: Stephen, you said, a work unit will be computed in three minutes.  Machine (1) needs sometimes 13 minutes with the background client.
___________________________
1: Athlon TB-C, 1.2 GC/s, 256 MB DDR-RAM, Erazor x², ADSL-Flatrate, NIC Intel, Win 98 SE Mainboard MSI-6380 Rev.  1
2: Pentium III, 600 MC/s, 256 MB RAM, NIC Intel, Win 98 SE

[This message was edited by Pascal [www.rechenkraft.net] on 2002-Jul-28 at 5:40.]
Stephen Brooks
2002-07-28 09:16:12
quote:
Originally posted by Pascal [www.rechenkraft.net]:
Another thing: Stephen, you said, a work unit will be computed in three minutes.


Are you sure I said that?  These units will take about 40 minutes each on a mid-range PC.  They're quite slow, and that's why I put the intermediate-save feature in.  That might be where you got the "3 minute" figure from: the program saves its progress on a (partial) workunit every 3 minutes.

Actually I'm encouraged that someone has managed to get to 0.3% already.  Hopefully it'll keep on increasing steadily.  Maybe when it hits 1% I'll make a best100 results.dat or something.


"As every 11-year-old kid knows, if you concentrate enough Van-der-Graff generators and expensive special effects in one place, you create a spiral space-time whirly thing, AND an interesting plotline"
Pascal
2002-07-28 09:46:24
@Stephen, sorry, I haven't read everthing about the new version.

___________________________
1: Athlon TB-C, 1.2 GC/s, 256 MB DDR-RAM, Erazor x², ADSL-Flatrate, NIC Intel, Win 98 SE Mainboard MSI-6380 Rev.  1
2: Pentium III, 600 MC/s, 256 MB RAM, NIC Intel, Win 98 SE
Jwb52z
2002-07-28 17:13:28
Can someone refresh my memory as to how long results take to show up using the manual send option?  I had some good results with 4.1 and I want them shown.
Pascal
2002-07-28 22:27:21
The upload function of elder versions is still working.  Yesterday I uploaded some results from versions 4.1x
I hope, you meant that, did you?

___________________________
1: Athlon TB-C, 1.2 GC/s, 256 MB DDR-RAM, Erazor x², ADSL-Flatrate, NIC Intel, Win 98 SE Mainboard MSI-6380 Rev.  1
2: Pentium III, 600 MC/s, 256 MB RAM, NIC Intel, Win 98 SE
Frank
2002-07-28 23:47:11
Speed of the new Client 4,20:

380 WU's in 24h on my Computer
Time per WU from 3...13min.
Avg.  Time 7.5min per WU and CPU, app.  4 min.  faster than version 4.12
80% of the results between 0.0 and 0.1% muons. 



My System:
DualAthlon MP 1900+, 1024 MB DDR-RAM, Gforce3 Ti500, ADSL-Flatrate, Win XP Pro, Mainboard ASUS A7M266D, Promise Raid 280GB (CPU's & GPU Watercooled cool )
Stephen Brooks
2002-07-29 02:37:16
I think as the results get higher, they will take longer to simulate.  A run in which 1% or 2% of muons are retained should be slower than a similar run on v4.1x. They will also vary somewhat, as you have found out.

--[Can someone refresh my memory as to how long results take to show up using the manual send option?  I had some good results with 4.1 and I want them shown.]--

You uploaded those after the time and date shown on the stats page currently?  (UTC = GMT, i.e. UK time without summertime).  I think I should probably leave my stats-calculating computer running all the time now - I was just going on to check it myself, here from work, when I realised that of course it wasn't going to show anything new as I switched off my home PC last night!  At least I could leave it running during the day.  For now, the next stats upload will probably be around 1900UTC.


"As every 11-year-old kid knows, if you concentrate enough Van-der-Graff generators and expensive special effects in one place, you create a spiral space-time whirly thing, AND an interesting plotline"
Pascal
2002-07-29 04:51:38
Will there be any update for the result viewer with the new extensions?

___________________________
1: Athlon TB-C, 1.2 GC/s, 256 MB DDR-RAM, Erazor x², ADSL-Flatrate, NIC Intel, Win 98 SE Mainboard MSI-6380 Rev.  1
2: Pentium III, 600 MC/s, 256 MB RAM, NIC Intel, Win 98 SE
Jwb52z
2002-07-29 05:19:15
Speaking of the result viewer, and this may be my fault for unzipping the program wrongly with this new version, it doesn't report the correct muon retention.  It gives a report of 44% retained, but the graph shows it correctly as very small.  Was there an unzipping issue with this version that is different from previous versions that I missed.  BTW, Stephen, I uploaded the results I was talking about a few minutes before I wrote that last message.
Jwb52z
2002-07-29 05:22:55
After reading the "Muon Background will not run" thread I think I made the mistake of just overwriting the files into the same directory I have alway used for the project.  Is that a big mistake?
Pascal
2002-07-29 05:28:35
@Jwb53z: I always extracted the files into a new folder, and that worked in every time..

___________________________
1: Athlon TB-C, 1.2 GC/s, 256 MB DDR-RAM, Erazor x², ADSL-Flatrate, NIC Intel, Win 98 SE Mainboard MSI-6380 Rev.  1
2: Pentium III, 600 MC/s, 256 MB RAM, NIC Intel, Win 98 SE
Pascal
2002-07-30 02:00:08
quote:
Originally posted by Stephen Brooks:
For now, the next stats upload will probably be around 1900UTC.



Will the next update be today?

___________________________
1: Athlon TB-C, 1.2 GC/s, 256 MB DDR-RAM, Erazor x², ADSL-Flatrate, NIC Intel, Win 98 SE Mainboard MSI-6380 Rev.  1
2: Pentium III, 600 MC/s, 256 MB RAM, NIC Intel, Win 98 SE
Orbi-tel
2002-07-30 12:56:41
I appear to have a problem with the auto send of results when running the background version - i.e. it dosn't.

Just checked and found a 59k results.txt file which appears to be the original since I started v4.2 about 5 days ago.

I stopped the prog and ran manualsend and the results were uploaded without problem.

v4.2 was unzipped to a new directory so there are no other 'old' files that could interfere with it.


Anyone else have this problem?
scottsaxman
2002-07-30 13:43:37
Doesn't the program wait until you have 100K in results now, since the results file is bigger?
Orbi-tel
2002-07-30 15:37:23
Where does it say it waits for 100k before sending AUTOMATICALLY.  The ReadMe mentions about manualsend if you have more than 100k but nothing about the figure that the auto send uses
Stephen Brooks
2002-07-30 15:38:00
Yep, it now sends at 100KB by default.


"As every 11-year-old kid knows, if you concentrate enough Van-der-Graff generators and expensive special effects in one place, you create a spiral space-time whirly thing, AND an interesting plotline"


It doesn't say anywhere how big the results file gets before auto-send happens.  Therefore the value is at my discretion.  (If it still says 15KB anywhere, it's wrong).

Also you can see at http://www.stephenbrooks.org/temp/muonstats.png that there are a lot of 98.5KB (i.e. just over 100`000 bytes decimal) files on there.  You can tell the trigger-happy/stats-addicted manualsend types because their files get uploaded while smaller than this.  You can tell the people on intermittant/dialup connections because their files get much bigger than this.
Pascal
2002-07-31 01:10:10
Steohen, here are some questions:
I've got some results with a yield of 0.000000 percent.  Many parameters are also 000. Is that normal?
How do you think about creating a new TOP100?  When will it be and at what yield will it end downwards?
Is there anything new about creating stats for teams?  If someone wants them, can you send me an Excel or text file of the results.
And at last: Some people asked me about a Linux client, will there be any in that version?
I thank you very much for your answer...

___________________________
1: Athlon TB-C, 1.2 GC/s, 256 MB DDR-RAM, Erazor x², ADSL-Flatrate, NIC Intel, Win 98 SE Mainboard MSI-6380 Rev.  1
2: Pentium III, 600 MC/s, 256 MB RAM, NIC Intel, Win 98 SE
Phippi
2002-07-31 02:29:43
@Stephen

For me would it be intersting, if I can get the complete results.  The SwissTeam plans to create a statistic page for Muon.  Maybe it is intersting for you and we can do some work to create Team stats?

Pascal question:
And at last: Some people asked me about a Linux client, will there be any in that version?

We also have some people, they have linux on their computer.

Thanks and regards,
Phippi
Stephen Brooks
2002-07-31 08:41:11
quote:
Originally posted by Pascal [www.rechenkraft.net]:
Steohen, here are some questions:
I've got some results with a yield of 0.000000 percent.  Many parameters are also 000. Is that normal?


That is not normal - I suspect there is a bug somewhere in the genetic algorithm with this version, as a lot of my results here have many zeroes in as well.  I think the simulation is still OK, it's just the selection of parameters that could be better.

I'm in the middle of upgrading the code, but I'll upload a fix when possible.

quote:
Is there anything new about creating stats for teams?  If someone wants them, can you send me an Excel or text file of the results


If you use Internet Explorer, you can rightclick the stats table and select "Import to Microsoft Excel". Then you can easily find the team members and add up the stats.  If you have OfficeXP you can even make a dynamic web query that recalculates the spreadsheet when the website stats are updated.

The stats are available in text form at http://stephenbrooks.org/muon1rawstats.txt which is tab-separated and again may be imported into Excel or parsed with Perl or PHP.

quote:
For me would it be intersting, if I can get the complete results.  The SwissTeam plans to create a statistic page for Muon.  Maybe it is intersting for you and we can do some work to create Team stats?


I think a PHP program could be easily written to load that text-file version I've linked to above and add up all the columns for people with "swissteam" in their name.  I might have a go at that now.

There is a Linux computer here but right now I haven't got much time for porting.


"As every 11-year-old kid knows, if you concentrate enough Van-der-Graff generators and expensive special effects in one place, you create a spiral space-time whirly thing, AND an interesting plotline"
Pascal
2002-07-31 09:03:11
Stephen, do you want to have some results with that many zeros?  I can mail it to you?  That are not so many results, perhaps 5-10 percent.  But these results are often finished after a few minutes.

Ok I'll try it with Excel.  But I still have Office 97, I can't install Office XP on my computer. 

When do you have an update? 
What will be with the actual results?

___________________________
1: Athlon TB-C, 1.2 GC/s, 256 MB DDR-RAM, Erazor x², ADSL-Flatrate, NIC Intel, Win 98 SE Mainboard MSI-6380 Rev.  1
2: Pentium III, 600 MC/s, 256 MB RAM, NIC Intel, Win 98 SE
Stephen Brooks
2002-07-31 12:18:23
I've seen what's happening with the results: I have some zeroes of my own.


"As every 11-year-old kid knows, if you concentrate enough Van-der-Graff generators and expensive special effects in one place, you create a spiral space-time whirly thing, AND an interesting plotline"
Stephen Brooks
2002-08-01 10:08:43
Today I've located and solved the bug that was generating the zeroes, but still have a bit of work to do before releasing v4.21 (fixing SMP support, for example smile ).


"As every 11-year-old kid knows, if you concentrate enough Van-der-Graff generators and expensive special effects in one place, you create a spiral space-time whirly thing, AND an interesting plotline"
Pascal
2002-08-01 10:52:05
I hope, you won't reject again any results..

___________________________
1: Athlon TB-C, 1.2 GC/s, 256 MB DDR-RAM, Erazor x², ADSL-Flatrate, NIC Intel, Win 98 SE Mainboard MSI-6380 Rev.  1
2: Pentium III, 600 MC/s, 256 MB RAM, NIC Intel, Win 98 SE
Stephen Brooks
2002-08-01 11:56:51
The simulation results are all still valid, just with occasionally dubious choices of design parameters, so I will still count them all as v4.2


"As every 11-year-old kid knows, if you concentrate enough Van-der-Graff generators and expensive special effects in one place, you create a spiral space-time whirly thing, AND an interesting plotline"
HETTATLONGUN
2002-08-01 17:00:10
Heh!

I uploaded some old results and got the 'new' muon1v42.

Results seem to be about 1.1KB, each.

I watched the first one on the screen, and there was a loooooong time at the end when it didn't seem to be doing anything.  The numbers at the bottom of the screen were running but the only graphic was a green line down the center of my screen.  big grin

Ahh, morning....Nothing like a good cigar, a cup of coffee and the sound of hot shell casings hitting the deck....HETT[TEP]
Helix_Von_Smelix
2002-08-02 11:09:11
Hi people,

Stephen can you double check the results uploaded just before this date/time, i had connection problems and may have sent the same set twice.  Sorry about that!!

If you make instant coffee in to a microwave do you go back in time??
Jwb52z
2002-08-02 16:00:22
I think I found a bug.  Just now when I stopped the program right after a save and rebooted my computer, I restarted it and it now shows that there are no results in the result file even though the amount of results in the result.dat counted as new results is correct.  The results.txt file is still in the directory and shows the correct size.  What do you think is wrong, Stephen?
Jwb52z
2002-08-02 16:15:13
Here's an update to that last post.  After that simulation that was halted for a reboot had finished, the correct information about the results and the file size came back.  I still think that should not happen though.
Stephen Brooks
2002-08-03 12:07:15
Oh I see why that's happening.  When Muon1 restores an auto-save, it won't request a new set of parameters to be generated, and the function that generates the parameters also (I think) tells it how big one of the files is.

[later] OK now it won't display these figures if it hasn't yet called the function to calculate them.


"As every 11-year-old kid knows, if you concentrate enough Van-der-Graff generators and expensive special effects in one place, you create a spiral space-time whirly thing, AND an interesting plotline"
Orbi-tel
2002-08-04 01:39:59
Stephen,

My Autto.sav file is ranging between 5.5 and 7Megs in size.  Why is it so big?  I am sure this is effecting the speed at which I can process as I am using a slow k6-400Mhz processor and writing that amount of data can take time.

Also my recycle bin is filling up very quickly as the file gets deleted/re-written every 1 to 10 minutes (I have saves set to 10 mins - 600 secs).
Stephen Brooks
2002-08-04 13:06:36
quote:
Originally posted by Orbi-tel:
Stephen,

My Autto.sav file is ranging between 5.5 and 7Megs in size.  Why is it so big?  I am sure this is effecting the speed at which I can process as I am using a slow k6-400Mhz processor and writing that amount of data can take time.


My computer is a P-II-400 and it runs smooth as anything - I was really surprised but it was hardly noticable.  The results file is quite big because for some particles I have to store 10 different decay-times (as doubles, so 80 bytes), and 80*20000 = 1.6M already, then there is the other particle information data plus there are more than 20000 particles (more like 35000) at most stages of the simulation.

I could make the file a bit smaller by not storing certain fields for muons (only pions have the decay-times), but the problem there would be that the saving would get slower as I'd have to query each particle individually (instead of one VERY LARGE fwrite command that I'm using now).

quote:
Also my recycle bin is filling up very quickly as the file gets deleted/re-written every 1 to 10 minutes (I have saves set to 10 mins - 600 secs).


??????  It ought to just wipe the file.  Maybe it was using that bloody "DeleteFile" Windows function.  I'll use fopen(...,"wb"); to zero the length of the file first.  Then you'll get a copy of it in the recycle bin at the end of each unit still, but it won't take up to much space.

Note: Stephen Brooks did not notice that feature because his computer is set to "delete everything in recycle bin (and a few things outside the recycle bin, like IE history) instantly".


"As every 11-year-old kid knows, if you concentrate enough Van-der-Graff generators and expensive special effects in one place, you create a spiral space-time whirly thing, AND an interesting plotline"
Pascal
2002-08-05 06:56:19
When will version 4.21 come out?

___________________________
1: Athlon TB-C, 1.2 GC/s, 256 MB DDR-RAM, Erazor x², ADSL-Flatrate, NIC Intel, Win 98 SE Mainboard MSI-6380 Rev.  1
2: Pentium III, 600 MC/s, 256 MB RAM, NIC Intel, Win 98 SE
Stephen Brooks
2002-08-05 08:37:23
I've got nearly all the changes I wanted to make already done, but right now I've got a rather frustrating bug to deal with.  It might be available tomorrow, or alternatively it might be a couple of weeks before I sort it out!


"As every 11-year-old kid knows, if you concentrate enough Van-der-Graff generators and expensive special effects in one place, you create a spiral space-time whirly thing, AND an interesting plotline"
Pascal
2002-08-05 08:44:37
Hey, that's ok.  I hope it will be a good bit faster than the actual client wink

___________________________
1: Athlon TB-C, 1.2 GC/s, 256 MB DDR-RAM, Erazor x², ADSL-Flatrate, NIC Intel, Win 98 SE Mainboard MSI-6380 Rev.  1
2: Pentium III, 600 MC/s, 256 MB RAM, NIC Intel, Win 98 SE
: contact : - - -
E-mail: sbstrudel characterstephenbrooks.orgTwitter: stephenjbrooksMastodon: strudel charactersjbstrudel charactermstdn.io RSS feed

Site has had 25162631 accesses.