stephenbrooks.orgForumMuon1GeneralWhy are users dropping out?
Username: Password:
Search site:
Subscribe to thread via RSS
John Kitchen
2002-11-29 10:14:22
Has anyone noticed how quiet it's getting?  The TAM postings on ARS are right down, and contributors to this project are dropping out.

A look at the latest stats shows that out of the top 100 contributors, 33 have failed to submit any results for more than 5 days.  frown OK, I know some of us (myself included) tend to wait until we have a decent collection of results before submitting, but consider this:-

Twenty of the top 100 have not submitted any results for over two weeks. roll eyes I'd guess that most of these have reduced or eliminated their involvement.

If you are thinking of dropping out, then what is the most likely reason?  Is it one of these?  Or something else?  If it is something not on this list, don't vote here, post your concern.  Let's get our concerns to Stephen before we have too many more defectors!
Stephen Brooks
2002-11-29 13:42:53
"Problems submitting results and getting credit" currently has the most votes.  There are a fair number of people who can't run a DC program on some computers unless it's completely "hands-free". So anyway I just made a small program "manualsend_daemon.exe" that instead of sending the results once, runs in the background and checks the size of the file once per hour.  If the file is over 100KB it sends it.

This is now included with the two Windows distributions (the old manualsend.exe now deleted because it was confusing) and also it's in the multisend.zip file where the new sender was.


"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"
rsarkezi
2002-11-29 14:23:29
Erm,

if I would run muon1 only on my AthlonXP2100+ with the best250 at home (which runs only a few hours at the evening), I bet I would also need 2 weeks to get the 100 kb of results

But I have also some machines at my work doing their contribution.  Yesterday I just set up a P3 600 which is only uses 1,5 days a weeks.  Not any longer )))

-----------------------------
Be efficient!
Stephen Brooks
2002-11-29 16:15:37
Yeah, I did wonder about reducing the threshhold a bit (e.g. to 50K), but really it's just the 24/7 machines you'll be most concerned with having automatic submission for.  If you're at the bax [meaning 'box'] to switch it on and off, it can't hurt to send muon results occasionally big grin


"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"

[This message was edited by Stephen Brooks on 2002-Dec-01 at 2:55.]
pvs
2002-11-29 16:40:02
My main Problem is not sending the results.  The problem is, getting credit for them takes to long or results are lost, o.k. last time Stephen said please wait 48 h, but that doesn't really help... (see thread results lost...)
Bewolf
2002-11-29 18:15:34
I stopped participating about a month ago because, as far as I can tell, the project is at a standstill until a new client is released.  It looks like this stage of the simulation as been optimized as far as it can.

I'll probably devote come CPU time to this project once the next stages of the simulation are implemented.
Stephen Brooks
2002-11-29 18:20:44
That is true.  I remember telling people a few times that they could take their boxes off this thing until version 5 or whatever.  Actually version 4.22 will have its uses since it will multi-run the high results - gets rid of rogues, increases accuracy - in fact I was contemplating users set a "rhubarb period" for very-high results in which it keeps re-simulating, then at the end I can look at not just the highest result, but the highest, most-confirmed result.  But you're right, no real science is being done until v5 as far as I can tell - only the meta-science of testing a new DC platform.


"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"

[This message was edited by Stephen Brooks on 2002-Dec-01 at 2:54.]
astra412
2002-11-29 21:33:14
Well, I can only speak for myself, but I'm simply lazy (and busy, which is a bad combination) so I only manual-send my results when I happen to remember and look, which is about once a week.  Though I wouldn't complain about v5 smile!

Astra412
ZeRoC00L
2002-11-30 02:24:16
Any idea when 4.22 or 5.x will be released ?
Herb[Romulus2]
2002-11-30 07:01:28
I like almost to check twice a day my progress and so I'm set for it to send in every 12 hours

On the other hand , if there's a period of no sense for the project at all, it's not so easy for me to switch around with other projects.  This means , if we just have a period of recalculations, that's ok with me. 

I won't go back to dumb key cracking.  This project serves a distinct purpose in science and I'm all for it

-------------------------------
I'd say more, but I can't reach the keyboard from the floor.
Ben Whitten
2002-11-30 07:50:09
Well I'm dedicating as much time this project as I can from my crappy P3 600MHZ just cant get much done nomater how long it's on for (currently having building works on my house so electrics are dodgy so I unplug when builders are in just in case) but there is a huge ammount of computers at my school about 60 new ish nice ones but I've asked and the wont alow it on which was a bumer, but I should be getting a nice new one, or build a good one (maybe send specs to a company and have them build it) enyway when I get that comp a new version of muon will definatly be out, expect a few more results from my direction when I get it big grin
[OCAU]GenomeX
2002-11-30 15:44:01
This project doesnt seem to have the same team rivalry as others.  Overclockers Australia has just seriously increased output, but [H] have nothing.  This is what makes folding so popular I feel.

Muon Progect member for OCAU
Midon
2002-12-01 23:20:40
I can only support Bewolf's statement.  I do not expect any advancement in the percentage for the next time or V5. The simulated design is nearly perfect.  If you get nearly every particle which is accepted at the entry to the exit at the bending chicane what can be done better?

[This message was edited by Midon on 2002-Dec-02 at 20:38.]
Stephen Brooks
2002-12-02 06:44:00
quote:
Originally posted by Midon:
I can only support Bewolf's statement.  I do not expect any advancement in the percentage for the next time or V5.

I don't expect an advance in v4.22 because it'll just be doing the same thing with a different code back-end and maybe some re-checking to get the accuracy sorted.  Version 5 will actually be simulating an almost completely different machine, so totally different designs to explore.


"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"
Ben Whitten
2002-12-02 07:02:10
So the next one will be quite a nice little adventure in a new design, will we be using results from this test to start the new one off?  just as a bit of a head start big grin
[DPC] Jive
2002-12-02 08:35:00
I'll restart crunching when there is

* New Version
* New Top-1000 File

My farm would increase with
* Hands-Free Networkability (proxy)
* System Invisibility (clean service)

And my personal involvement would increase
* When i finish all annoying things that come with buying a house
* I get all holidays & birthdays & aniversaries done which clobber me in the last 2 months every year.
* They finally start planning abit further ahead @ work, instead of just dumping me with "this has to be done by tomorrow" when it's really a 96hour project.


But since i'm still visiting the forum, i could also say i'm still actively participating.
px3
2002-12-02 11:12:44
I'll stay cool

it's really interesting to watch new dc ideas grow smile, regardless if there's an scientific impact on the muon project itself..

Rolf
PX3
px3
2002-12-02 11:23:34
quote:
Originally posted by John Kitchen:
...
A look at the latest stats shows that out of the top 100 contributors, 33 have failed to submit any results for more than 5 days.  frown OK, I know some of us (myself included) tend to wait until we have a decent collection of results before submitting, but consider this:-
...



i'd watched this too.  especially this guy makes me nervous: eek

73. Stephen Brooks 25d - 828 - 5`589 595`654`010 (.303%) 2.926% (399th)

hope he hadn't left the project big grin wink
Stephen Brooks
2002-12-02 11:42:58
Oops!  smile

I switched my _own_ Muon program off because at one stage I was developing the code and wanted to do specific simulations without the background one hogging my CPU.  Come to think of it it would be good testing if I switched it back on again, but as many have figured out, we're already near the optimal design so I thought I'd give my CPU a holiday.

What I'd suggest as an option for people with multiple PCs is to leave perhaps one 'toy' machine running Muon, just to keep the system ticking over, and then shift the rest over to some more worthy thing like THINK, or prime-factorisation or whatever turns you on.  That way I'll have a few people to beta-test when I get the new versions out.  I have Christmas holidays coming (quite long ones) so I might be able to get 4.22 at least done before Christmas.  There might be a "version 5 brainstorming" session on this forum too, so I can figure out what's a priority (for instance I could make the results send as a binary format, reducing disk usage, but ASCII is more portable).


"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"
Midon
2002-12-02 12:40:59
quote:
Originally posted by Midon:
I can only support Bewolf's statement.  I do not expect any advancement in the percentage for the next time or V5.



Oops!  red face

The sentence with V5 was wrong (Never change a finished sentence frown).  I mean that I expect no advancement with the V 4 but I'm waiting impatient for V5 smile.
AySz88
2002-12-02 13:05:42
By the way, Stephen, THINK is done for UD.  Ligandfit is the second phase of the cancer project.
[ARS]odessit
2002-12-02 14:22:15
Probably a combination of issues.  I myself is limited to a home farm due to the need of stable and feature rich client that can run for couple of years without the need for upgrade, I had to deploy my borgs with Prime95. But home machines are still crunching muon project, I just don't send the results that often razz

ARS Team Atomic Milkshake
Unofficial Muon1 FAQ
Supp
2002-12-02 15:55:55
Autoupdate feature (check with every result sending) would help...

rm -rf /
Stephen Brooks
2002-12-02 16:00:10
quote:
Originally posted by AySz88:
By the way, Stephen, THINK is done for UD.  Ligandfit is the second phase of the cancer project.


I wasn't talking about UD, I was talking about http://www.find-a-drug.com/ . The last I heard, UD was in "infinite redundancy", which means it is (like this project) not really doing any useful scientific work, they're just repeating units.  It may have changed in the last few months though - I haven't checked back much.


"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"
[Ars]kenlow
2002-12-02 19:48:36
quote:
Originally posted by Stephen Brooks:
Oops!  smile

What I'd suggest as an option for people with multiple PCs is to leave perhaps one 'toy' machine running Muon, just to keep the system ticking over, and then shift the rest over to some more worthy thing like THINK, or prime-factorisation or whatever turns you on.  That way I'll have a few people to beta-test when I get the new versions out.



Trouble with switching machines is that i have to find a project that I like and once I switch over, it will be unlikely that I will switch back due to the hassel smile
AySz88
2002-12-02 20:36:13
quote:
Originally posted by Stephen Brooks:

The last I heard, UD was in "infinite redundancy", which means it is (like this project) not really doing any useful scientific work, they're just repeating units.  It may have changed in the last few months though - I haven't checked back much.



It has indeed just recently begun the actual scientific work as of a week or two ago.  big grin
Herb[Romulus2]
2002-12-02 23:32:00
quote:
Trouble with switching machines is that i have to find a project that I like and once I switch over, it will be unlikely that I will switch back due to the hassel

Exactly, Amen smile

-------------------------------
I'd say more, but I can't reach the keyboard from the floor.
[DPC] Jive
2002-12-03 07:11:44
Automatic Client-Update would be a definate bonus.
Especially if u could implement an intermediate proxy along the way.

Makes it very easy to build & maintain a farm.
Stephen Brooks
2002-12-03 07:28:34
What exactly do you mean by a 'proxy'?  A program that harvests results from an offline LAN of machines and sends them all under your name?

Auto-update is sort-of possible, although somewhat tricky.  I was thinking of such things as automatic best1000-download in version 5 (although you could choose how regularly this happens - default every 48 hours - to manage bandwidth).  Using a ZIP decompression library would be useful.


"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"
David Bass
2002-12-03 10:58:00
Personally, I think auto-update is only good if it can be switched off.

Whilst not wanting to belittle the programming efforts being performed here, I will say that don't see quite enough quality control to allow a new version to run on my home (and only!) machine without initial supervision.

Not to mention any issues concerning the security model that would be implemented to prevent others from hijacking machines by spoofing a client update...

I understand the need that the owners of large farms feel they have for autoupdate, but personally it does not suit me, and would not even if I had a farm.
Stephen Brooks
2002-12-03 13:58:18
Yes, most likely auto-update (and most of the features requiring internet interaction other than sending, which obviously requires it) will be an optional thing in config.txt.  I also understand why you might not want auto-update turned on - I don't necessarily know beforehand if there's some strange intermittent bug that'll make a version unstable, and of course if one version dies, it will stop polling the server for updates smile. The other option is to make a sort of "daemon" program that runs independently and monitors the situation.  Then it could possibly resurrect a crashed Muon when a fix is available. 

quote:
Not to mention any issues concerning the security model that would be implemented to prevent others from hijacking machines by spoofing a client update...


Well it'd be difficult since the program would request a file from my web-server if signal.dat says there is a new version.  Spoofing that and replacing it by a different file would be as hard as spoofing any other EXE (e.g. ones from WindowsUpdate.microsoft.com!!).  In other words, the auto-update would be a "pull" where the program checks for a new version whenever it sends, rather than a "push" in which my server actively sends a new version onto your client.


"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"
Bluumi [SwissTeam.NET]
2002-12-03 14:06:50
i switch off my clients.  roll eyes

- no really new things ... (was a big bonus in this project ..) [waiting for V5]
- no really steps forward [waiting for V5]
- need to much power to hold the place big grin
- need power for DF.  (Team goes up in DF, and stay at Muon)
- no ServiceClient (M_background & Firedeamon is NOT a solution)
- no UnitProxy, i refuse FTPConnection completly (DF hasn't too)
- to big and long Workunits ... (can only run over weekends without interupts)
- there was a very UNSTABLE time with uploads and credits
- the problem of 3.5% results that won't be real, but not are cheated ..

Bond ... day a other day ... or "i'll be back" - V?
big grin

SwissTeam.NET, the better .NET
David Bass
2002-12-04 00:08:51
quote:
Originally posted by Stephen Brooks:

Well it'd be difficult since the program would request a file from my web-server if signal.dat says there is a new version.  Spoofing that and replacing it by a different file would be as hard as spoofing any other EXE (e.g. ones from WindowsUpdate.microsoft.com!!).  In other words, the auto-update would be a "pull" where the program checks for a new version whenever it sends, rather than a "push" in which my server actively sends a new version onto your client.



Understood - nevertheless, ftp sites have been hijacked before and there is no guarantee that you will maintain control over your current domain for an indefinite period.  (I've seen examples of incompetence by the registration authorities result in more than one sudden change of ownership for popular but niche domains.)

I tend toward the paranoid over this issue, I admit.  But then, if I were running clients clandestinely on other people's machines (cf perceived requirements for a silent service client), even if I had the authority to do so, I would be very wary of an autoupdate that was initiated without my explicit consent.
[DPC] Zephyer
2002-12-04 04:06:43
maybe because RC5-72 just started @ distributed.net ??

Member off [DPC]Alphen // Project Muon
Don't Underestimate The Power Of The NeT
God is my Pilot, but the Devil is my Bombadeer....
Tripodal
2003-01-07 22:39:30
An addition id like to see is the ability to save all results created by all users... my understanding (i may be wrong) is that the client wont re-do results so if you have all results already completed in your results.dat you could get more work done with less redundancy.

I am assuming tho that there is probably a good bit of it already happening.
Supp
2003-01-08 12:06:19
Anyway have you got any timeline for V5?  (just asking...)

rm -rf /
[ARS]odessit
2003-01-08 14:34:14
same question here re: V5

ARS Team Atomic Milkshake
Unofficial Muon1 FAQ
Jetta
2003-01-10 03:02:51
well I know some people need a non ftp method of sending results in.  Not all people have ftp access through work proxys, so a web based or some some other non ftp port submittal method would help bring people in/back as well.

Is there any chance of a non-ftp sollution for V5?
Stephen Brooks
2003-01-10 06:53:28
Some service needs to be present on stephenbrooks.org in order to receive incoming connections (I can't put a server here because my ISP at home doesn't seem to be too reliable on allowing incoming requests).  I don't think my webhost allows CGI scripts that run continuously, but PHP does have the facility to handle file-uploads, something which I've looked into a bit.  I do know of a command to use IE to download a webpage's HTML to disk (similar to wget in UNIX), so perhaps I could put the upload filename in the URL query string (after the '?') and get PHP to upload the file then generate a report page, containing what used to be in signal.dat.

V5 would be the time to do this.  Before that there's still going to be v4.22 and maybe v4.23 because there are features that have to be tested a bit before I do a major reshuffle.
Stephen Brooks
2003-01-10 07:04:55
(Logged in as a guest?  Strange).  The next time I'll be employed properly to do this work is at Easter.  After the hassle in November last year with setting up the new FTP system I didn't want to do much for a while.  The problem was, although that was ultimately very productive, getting it working took a few all-night sessions and it eventually affected my degree work and in part caused me to fail a driving test.  That's why I wanted a break this Christmas (well, actually I'm porting an arcade game I wrote in 1997 to the PC).  So we'll see some upgrades at Easter - perhaps before then, as I'm feeling rather more positive about the whole Muon thing now.
The long period during last summer when I kept thinking I'd produce v4.22 and nothing happened was because I underestimated the effort required to write an entire interpreted programming language.  But (hopefully) it's working now.  It seems to be very close, but with some bugs.


"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"
Rocker_ed
2003-01-12 13:40:45
[DPC]We alone is dropped out because we are now running RC5-72.
This was just a project between RC5-64 and RC5-72.
After RC5-64 was ended we all gone crunching ECCp-109 but that was for a few weeks than that ended. 
Then we had nothing to do and found this project

But now we are running RC5-72,
SORRY

Sorry bad English

Team Member of [DPC]We Alone
Stephen Brooks
2003-01-12 17:22:26
No problem - it's not so important right now anyway.


"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"
Jetta
2003-01-14 03:02:13
Thanks for the response!  Smile Hopefully a non-ftp upload wont be too hard for you!  Smile Thanks for all the hard work!
K`Tetch
2003-01-20 12:18:52
Theres a big problem with RRC%-72, and that is "it's pointless".

D.net stated that the point of RC%-64 was to see how feasable a large-scale distributed computing project, with a known finite finishing point was, as well as to see how robust such a cypher was against a brute force attack.  64-bot lasted 4 years.  It was shown that such a project is feasable.  Now they're doing RC5-72 and churning out exactly the same reasons for it, with the added reason "we want to see how well it scales up". So, either -64 was a waste of time, OR they're lying.  the new point is also one htats been done before, -72 might be 256x bigger than 64, but -64 was 256x bigger than -56, their previous RC5 project.  Same scaling, same problem, just longer.

I used ot be a BIG supoprter of their OGR project as well.  I even put money into the -24 and -25 "prize funds". I also offered to manaually crunch completion values almost a year ago (manual as in writing down each stubb and marking each returned one with a tick if need be) but they turned me down - turns out their OGR client was also done in a sloppy and slip-shod manner, as it'll only do 1- and 2-digit gaps, so the biggest gap the program will do is 99, completely ignoring whole REAMS of results.

SETI@home was s imilar waste of time.  a fixed (or vitually fixed) aim telescope, running through a few hundred channels is nothing.  Whilst Aricebo might be the biggest individual scope, there are "bigger" around.  the VLA in Scirocco, New Mexico is effectivly bigger (since multiple scopes have an effective size of the distance between them), the new "earth wide array" composed of scopes from all over the world, including Jodrell Bank, and Goonhilly, gives an effective scope the size of the earth, but that isn't used.  So few channesls are used, which is also interesting, because the retired META (Megachannel Extra-Terrestrial Array) can easily cope with all the Seti@home data at once, and that was retired 10+years ago, in favour of the new BETA (billionchannel, not megachannel)

Back in early december, when I learnt of the flaws in OGr, I went looking for a new distributed computing project.  i found this one.  It was relevent, origional, and doign work that was not only usefull, but needed, and it gelled with my intrests.
pben
2003-01-21 20:23:58
Why I left is the combined reuslts percentage stoped moving much.  I could see little point in exploring a very small area of the variable space.  So I moved on to folding proteins.  I had hoped that the climateprediction.net would have started by now (Windows only maybe I will keep folding proteins).

The only suprise is how long it took to push me off the first page of stats.  What are you guys a buch of slackers Razz
: contact : - - -
E-mail: sbstrudel characterstephenbrooks.orgTwitter: stephenjbrooksMastodon: strudel charactersjbstrudel charactermstdn.io RSS feed

Site has had 25163114 accesses.