Page 1 2 |  |
| PSEUDO [SwissTeam.Net] 2003-01-14 00:44:45 | Hello Stephen Is it possible, that u can say us when the new Version will be available? Many User dropped, because they are waiting for the new Version. So if u can say us the release date of the new Version, they will know the date to return.  Grüessli Kay [This message was edited by Stephen Brooks on 2003-Feb-05 at 18:13.] |
| Pascal 2003-01-19 10:38:52 | In another thread I read one of Stephen's postings, that the client of version 4.22 will come at a time around Easter. ___________________________ |
| Stephen Brooks 2003-01-19 18:36:23 | How about I set up a mailing list for interested team leaders/members, which I'll post to whenever I release an update or a new version? That way you won't have to keep asking, and can concentrate on other projects until a new version arrives. Err you know, if they made an extra-large black briefcase that also doubled as a 1-passenger car, some fool would complain it was too large to open at their desk |
| Pascal 2003-01-20 23:41:57 | @Stephen, just do it! You may set some information about it on the project's site or on the message board  ___________________________ |
| rsarkezi 2003-01-21 03:41:44 | quote: /me subscribe to this mailing list ----------------------------- Be efficient! Team www.it-academy.cc  |
| PSEUDO [SwissTeam.Net] 2003-01-21 04:43:07 | quote: Great Idea. When u made the list please subscribe me. Thx. Grüessli Kay |
| [DPC]Stephan202 2003-01-21 05:46:54 | I'd like to be on the list too --- Dutch Power Cow. MOOH! |
| TheFinalLoser 2003-01-21 09:46:23 | me too... then i must not visit this website weekly "So the last will be first, and the first will be last." Mt.20:16 |
| px3 2003-01-21 11:03:08 | quote: hmmmm..... what's the sense of this forum ?? my interpretation 'd been so far: users can share ideas and be up to date.. is there really a need of a mailinglist ??? if so, why not close the forum and keep everybody informed via mail ? if it's eaysier to keep users informed via mailinglists, there's no need of a forum, if a forum is what's needed, there's no need of a mailinglist.. IMHO Only my opinion.. Regards, Rolf |
| John Kitchen 2003-01-21 17:50:42 | quote: And a very good opinion too, IMHO! Use the forum as it is intended. If you want to be alerted to a specific user's posts (such as Stephen's), click on the user name and follow the instructions. That way you don't have to check back on the website, you just have to discard unwanted notifications. |
| Pascal 2003-01-21 21:46:26 | :grank: :grank: I forgot the most important thing. I'll also join the mailing list. p.p.k@wanken.de  ___________________________ |
| Stephen Brooks 2003-01-22 09:27:03 | Well I say "mailing list", what I mean is a list of e-mail addresses that I broadcast specific and occasional news-bulletins to, such as releases of major new versions. The main point is so that people who are simply waiting to be told about stuff like that don't have to bother with checking the forum repeatedly. So PX3 I didn't mean a mailing list like USENET or whatever - just a broadcast for major news. I think if you use this forum's mail-notification service to follow my posts you'll get loads of stuff that you didn't really want. Err you know, if they made an extra-large black briefcase that also doubled as a 1-passenger car, some fool would complain it was too large to open at their desk |
| Bluumi [SwissTeam.NET] 2003-02-01 15:13:56 | i want be on the "important-things"-Mailing-liste too. So i can wait on V5 (not the V4.2x Version) without checking the forum... ////// ***** \\\\\\ Stephen... Why did you cry so loud, that V4.22 use dataencryption and is not open source, and "never" release it.... It was a so long stupid "fight" about (not)open source... you lose "hardcore" opensource freak with no reason..  cya Bluumi SwissTeam.NET, the better .NET |
| Stephen Brooks 2003-02-01 16:32:04 | quote: That's because there were a larger number of people voicing concerns about possible cheating and faking of results. I wanted to reassure them that I had a system to stop that. Right now I'm removing repeats at this end, which I think has stopped all cheating for now, but the checksumming stuff I think you're referring to is already in the copy of the code I have here. quote: Ahh, you mean I should have kept quiet about the feature and made it a nasty surprise for them when I released it?  |
| Trent 2003-02-05 09:45:37 |  ....YepCrunching  for  |
| ZeRoC00L 2003-02-08 02:36:30 | quote: Erm... might be possible. How about I go by "window name", i.e. the name that appears when you ctrl-alt-del and get the tasks list? In fact I could have it so you could specify a text file "stop.txt" which had a few lines in it saying things like "Quake", "Media Player", and then Muon would check at intervals with the list of other windows open and stop if any of them had any of those things as a substring (non-case-sensitive) of their title. [This message was edited by Stephen Brooks on 2003-Mar-27 at 21:59.] |
| Trent 2003-03-16 15:51:15 | Can you add me to your mailing list? Crunching for |
| [DPC]Stephan202 2003-03-16 22:21:37 | I guess there is no mailing list yet, or at least no mail  I really like to know how the programming on V5 is going. --- Dutch Power Cow. MOOH! |
| Trent 2003-03-19 07:57:44 | Ya, how is the programming going???When should it become public? Crunching for |
| Stephen Brooks 2003-03-20 12:24:57 | This week I was finally back working on it full-time. I've been rearranging the code somewhat - it's becoming more 'modular', each component becoming more separate from the rest of it. Another offshoot of this is that I think the calculation has become slightly quicker, although it's hard to tell as I haven't done precise benchmarks. Tomorrow I have a redo the way particles detect which components are influencing them (bounding boxes or something), and want to do some profiling on the main loop because I think there _might_ be some unnecessarilty slow things left in there. The next stage when I've got it effectively back to where it was, is remembering exactly what I've said will be in v4.22/v5 and the like. I know we had quite a few ideas, some of which are easier to do than others. Expect a v4.22 release (and perhaps some 4.23 4.24 bugfixes after that) because I want to test the newly-structured code on the _current_ problem to check that it's all OK and stability is right across different platforms. Some time early next week I'll probably be at the stage to ask about features to add, and then the week after I'm hoping to release the 'confirming' version v4.22. By that stage I may also be developing the v5 problem in parallel: the point with all this rearrangement of the code is that the same program can probably run either of the v4.22 or v5 problems, determined only by the lattice input file and perhaps some data files (we've got muon-cooling tanks of liquid hydrogen to model in the v5 ring). Today's weather in %region is Sunny/(null), max. temperature -99999°C |
| Stephen Brooks 2003-03-21 02:22:59 | Well I write that last night and first thing this morning I have a discussion which changes the plan... Apparently the design has changed again (it no longer includes the chicane), but before I get onto that in v5, some people would be interested in finding out what the more general solenoid-variation (new in 4.2) does to _just_ the solenoid channel on its own. The restricted ~13 parameter one I had before got it up from 2.7% to 6.5% transmission (the high was 7.066% but there was statistical noise). So if I release say v4.3 that does just the solenoid channel for a couple of weeks, we'll probably see some very high percentages (11%?). Today's weather in %region is Sunny/(null), max. temperature -99999°C |
| [DPC]Stephan202 2003-03-21 10:39:31 | So we can expect the next simulation to be more succesful? Does V5 simulate the archievement of the same goal in a different way, or does it simulate something totally different? If the first thing is the case, that would mean that the method we simulated until now won't be used, right? --- Dutch Power Cow. MOOH! |
| Stephen Brooks 2003-03-21 12:48:29 | quote: The v4.3 slot be filled by this decay-channel-only optimisation, which is essentially v4.2 _without_ the chicane on the end. It'll be a bit 'boring' I suppose, but at least we should be able to see some high percentages. This won't run for very long - I think it will be possible to use v4.2 successful results as 'seeds' for it, which will give convergence a kick-start. It will also hopefully act as a testbed for a few new features I'm putting in (so there may be some bugs to iron out). quote: It will simulate the same decay channel, but with different components added onto the end, and no chicane. Of course they could decide to change the design _again_ before I write it, but hopefully not. Today's weather in %region is Sunny/(null), max. temperature -99999°C |
| Pascal 2003-03-23 10:24:31 | Nice to hear something about a new version, Stephen. When will you release it? ___________________________ |
| Stephen Brooks 2003-03-24 12:27:37 | I might be able to release v4.3 in as little as a week. Today I did some profiling of the code and found a few slow things and a few stupid things, which I removed. Apparently the version you're running has two 1ms waits inserted for no particular reason, which I've removed. The overall effect in a 30000-timestep simulation is up to 30000*0.002 = 60 seconds, mostly appearing near the end when there are few particles and yet the simulation appears to take a while. The waits actually _do_ serve a purpose with multithreaded modes, but I've reconciled those two things now. Tonight I can work on the results-server, preparing it for the possibility of doing multiple concurrent optimisation projects (and you'll be able to choose which ones you participate in too). I'll need to make a new PHP stats thing to cope with that. -- TIMELINE -- ~1 week: v4.3 solenoids-only optimisation (perhaps going as high as 11%) ~2 weeks: New PHP stats tables... starting off basic, will be improved with time ~3 weeks: v5 and a new accelerator with some different sorts of components too. Note that it's likely that the multi-project nature of this will allow some users to continue running the solenoids-only thing introduced in v4.3 if they want to as well. Today's weather in %region is Sunny/(null), max. temperature -99999°C |
| Stephen Brooks 2003-03-24 16:37:27 | The multi-project results-server end is theoretically in place now, and I think I've also cleared up the issue with 'weird' usernames like "a7C-Bob The Builder" appearing. The ones that are in the stats already will have to be removed by me manually, but hopefully no more will appear. Of course it's half-past midnight, I have to be up by 7am, so I can't do thorough tests on it right now. If results start going strange or missing Tuesday I'll look at it tomorrow evening. I also have to test the multi-project part of it manually because obviously none of your clients are submitting results with project-selector tags. Today's weather in %region is Sunny/(null), max. temperature -99999°C |
| Stephen Brooks 2003-03-27 13:52:20 | Multiple project support appears to work. I'm currently chasing memory leaks and among other things wondering why Muon is trying to free a block of memory that isn't allocated (it appears just to have _made up_ an address because I certainly didn't allocate that address at any time before in the program - I checked!). Other than that the client works fine, and about the only functional thing I still have to put in is a best-results-rechecking queue. Monday or Tuesday looks most likely for a v4.3 release. Today's weather in %region is Sunny/(null), max. temperature -99999°C I also replied to a post of ZeroCOOL's about a month ago, but it's appeared on the previous page. [This message was edited by Stephen Brooks on 2003-Mar-27 at 22:01.] |
| TheFinalLoser 2003-03-28 05:45:04 | jeah version 4.3 at the horizon coool "So the last will be first, and the first will be last." Mt.20:16 |
| Stephen Brooks 2003-03-28 10:33:49 | Progress for Friday... Fixed three separate memory leaks and a couple of other errors. I used a debug library I wrote which tracks _all_ memory usage, so if we're lucky v4.3 will be rock solid. Of course users are amazingly good at setting up their computers in such a way that Muon doesn't work, so I can't make promises. Had a frustrating afternoon because I thought I was within reach of being able to release ahead of schedule today, but what should have been a routine accuracy-verification exercise I needed to do after changing the algorithm turned out to be full of problems. Mostly because the accuracy-checking code was last used somewhere around v3.11 and other things have moved on. I've started uploading backups of my work to the webdrive so it's possible I'll work on it a bit this weekend. The only other feature I've definitely got to put in is re-checking of best-so-far designs (using multiple runs to reduce statistical errors). I think I'll let users set the number of re-checks, although there'll be a minimum of 5 or something. Some people might want to increase this towards the end of the optimisation, to get better accuracy. I was toying with the idea of putting some compression on the results files before sending, but I think I can safely cope with that in a later version (4.31 or something). Another thing to note is that it might take a day or two after the release of v4.3 to get a proper stats page working! Your results will be accepted, I've just got to modify the engine to count them. Stats counts will start from zero again from v4.3 onwards, and the old stats pages for v4.0-4.2x will, I think, still be updated for a while as people submit results from old versions. There's no reason why I can't have both running at once. Today's weather in %region is Sunny/(null), max. temperature -99999°C |
| rsarkezi 2003-03-28 12:27:35 | Hi Stephen, I hope I am at the newest: v4.2x after v4.21b will not be officially released but therefor comes v4.3? On the Muon-startpage only v4.21b is downloadable... ----------------------------- Be efficient! Team www.it-academy.cc |
| [DPC]Stephan202 2003-03-28 13:05:44 | I like the webdrive script! It's cool.  I wonder, what's the muonplayer? When I run it, I get an error... --- Dutch Power Cow. MOOH! |
| Stephen Brooks 2003-03-29 04:13:26 | quote: By "v4.2x" I meant all of v{4.2,4.21,4.21a,4.21b}. The MuonPlayer utility is a program that plays the animations that can be generated from Muon1 by pressing "R" (I think) to toggle video recording. I used this format to send animations to people across the internet. Today's weather in %region is Sunny/(null), max. temperature -99999°C |
| GUS 2003-03-31 06:01:35 | OK new version cool!! So this should be happening in the next week???? on a more important note, what will happen to the stats????? cheers |
| Stephen Brooks 2003-03-31 09:38:44 | Stats from this version will be preserved on another page (along with the final team rankings). Versions 4.0x and 4.1x are unfortunately lost permanently - I might be doing more backups from now on so that doesn't happen again. I was very near being able to release on Friday, but for some reason the accuracy-checking I need to do with the solenoids has been troublesome. I still have hopes for releasing tomorrow, but there's a chance it'll be another week (weds-friday this week is a conference so I'll be busy). Today's weather in %region is Sunny/(null), max. temperature -99999°C |
| Stephen Brooks 2003-04-01 11:04:27 | Well I stayed on an extra hour today but didn't manage to finish it. I'm just debugging the "re-test the highest-yet-observed result" feature. And that's the last thing I'm going to be put in before releasing. So I'll be back working on this on Monday 7th April after the conference. The solenoids thing turned up an interesting problem - the magnetic fields just gradually fade away into space, so there's a question of where you cut them off to save calculating negligable influences. I did some testing and this morning found out that they can't really be cut off until a LONG way from the solenoid (a distance of 13 times its radius from its ends). But that then brings the whole simulation into question because in that space there will be other things affecting the magnetic fields (and I don't yet know how solenoids interact). I've asked a magnet design expert about this problem. Today's weather in %region is Sunny/(null), max. temperature -99999°C |
| Stephen Brooks 2003-04-07 10:37:27 | Nearly got there today - I compiled passable executable programs just before leaving work, but eventually decided instead of rushing into things I'd do it tomorrow after some debugging. The results-rechecking feature was passable, but I think I can do a better system that will reduce errors resulting from "CPU glitches" from being quite so significant. Today's weather in %region is Sunny/(null), max. temperature -99999°C |
| Supp 2003-04-07 16:04:48 | Eagerly waiting rm -rf / |
| Stephen Brooks 2003-04-08 03:16:23 | v43.rar (566K v43.zip (775K These won't be on the webpage until this evening because I can't edit it from here. Also muon1viewresults.exe needs some refitting before it works with the new version's results format, so it's excluded from this first version. My stats-counter is receiving results in the new format fine (or at least it does in tests - I guess we'll find out), but I need to write a PHP version of the stats script for the webpage, so that will be a day or two coming as well. When I fix muon1viewresults.exe it'll either go up as a separate download or be packaged in v4.31 with some other changes I'm going to make. OK so that's the bad news - the good news is... New features: > Results are queued for re-checking if they are the best ones so far > Multiple server manualsend.exe included as standard (and the client sends this way automatically too) > "Interpolate" type in the genetic algorithm improved to include extrapolation as well as interpolation > Fairly massive infrastructure included for multiple and configurable optimisations to be run concurrently - version 4.3 comes shipped with one "lattice file" called SolenoidsOnly.txt, but in later versions we might be wanting to optimise more than one thing at once > Results now scored by "Mpts" = "Millions of particle-timesteps", which is somewhat fairer than the old scoring system > Results now hold checksums for themselves, their scores, and also the validity of the lattice files used in creating their simulation Today's weather in %region is Sunny/(null), max. temperature -99999°C |
| Stephen Brooks 2003-04-08 03:39:42 | Well I was about to send that post out to our "mailing list", but just realised that TheFinalLoser and Supp haven't supplied their e-mail addresses. (Bluumi also didn't but I found it on his website). Today's weather in %region is Sunny/(null), max. temperature -99999°C |
| TheFinalLoser 2003-04-08 04:54:13 | ups, but i got the message anyway...... *going and changing profile* "So the last will be first, and the first will be last." Mt.20:16 |
| Supp 2003-04-08 06:58:52 | Hm...I got e-mail set in profile...but I check this forum often, so rm -rf / |
| [DPC]Stephan202 2003-04-08 07:10:28 | I got the new version running. Works fine so far. Will we get more points in the stats for simulations that are rechecked? Also, is it correct that the username is not added after every result in results.txt? Will this be added just before uploading? --- Dutch Power Cow. MOOH! |
| Bluumi [SwissTeam.NET] 2003-04-08 08:03:51 | i Never post my eMail on a forum But in the swissteam-forum i read that the client is relased ... Now i'm here...  thank you for the eMail, i will read it ... later SwissTeam.NET, the better .NET |
| Stephen Brooks 2003-04-08 08:18:48 | quote: You may have it set, but I think you've got "Don't show my e-mail to other users" ticked too. Actually I was being stupid, as since I'm the board administrator I can view it even if you have chosen to hide it from everyone else. You and TheFinalLoser are now added to the mailinglist properly. quote: Yes, I made sure that as the result was "in quarantine", the Mpts figures add up with every re-run made. quote: Yes, your username gets included in the filename of the file uploaded to the FTP server, the same as before. Today's weather in %region is Sunny/(null), max. temperature -99999°C |
| TheFinalLoser 2003-04-08 10:14:08 | works since six hours without problems..... *tumbs up* the only thing which is missing, are the new stats.... but i think they will start in the next few days "So the last will be first, and the first will be last." Mt.20:16 |
| Pascal 2003-04-08 11:31:23 | As some of the members of team Rechenkraft.net and me found out, not every result in the resultfile (results.txt) does have as many parameters as all the other ones. I checked the files and look there - that are the last parameters of othe simulation. So - it could perhaps be your will - or even not. I just wanted to post it. If anything else mysterious may appear - I may tell you. ..and crunches away.. ___________________________ |
| Stephen Brooks 2003-04-08 11:35:38 | The number of parameters will vary because some designs have more solenoids (fitting into the same length) than others. Today's weather in %region is Sunny/(null), max. temperature -99999°C |
| [AMD Users] Michal Hajicek 2003-04-08 11:51:23 | It works fine for several hours. But there are some additional changes from v4.21b you did not mention, Stephen, I think, at least it seems that number of params varies slightly (53-59 units) - OK, answered, I am writing too slowly. Also checksum breaks the possibility to manually seed param-set (like gogomaus and others did in v4.2). And one more thing: If I understand the rechecking, only results, that are the best so far in the first run are rechecked. If this is the case, it will be good to use some "rechecking range" instead; maybe with minimal_width_limit and an option to choose interval width; and this interval can should grow with the muon yield (e.g. 5% of best percentage). This can eliminate the possibility, that the first run has smaller percentage than an average, which is higher than the best_so_far_design, but will not be calculated because of bad first run. If this is not the case, forget it. I hope my english could be understood. Thanks for any comment and good luck in future development. Michal |
| Stephen Brooks 2003-04-08 15:38:54 | If a design has a "true" percentage higher than the current maximum so far, then there will be a greater than 50% chance that its first run will turn up a percentage higher than that too, so it will be rechecked more than half the time. I considered putting in a 'width' below this, but it struck me it could be arbitrary, it could end up with nearly all results being rechecked, which is not the point, because if I wanted all designs to take 5x as long I'd have put more particles in. The manual seeding won't work for now. Actually there's a "manualmode" routine that I'm part-way through writing, which will allow users to manually create a design and add it to the re-checking queue, but that will come in a much later version. Today's weather in %region is Sunny/(null), max. temperature -99999°C |
| rsarkezi 2003-04-09 03:32:17 | v4.3 works really fine! Upload seems also to be working but I only get a funny message: "New version v4.21b is available from the website! (You are using v4.3) Programm now reads from previous results properly and sending while offline is corrected. *drivelingformuon1viewresults.exe* ----------------------------- Be efficient! Team www.it-academy.cc |
