Message boards :
Number crunching :
Fairer distribuiton of work(Flame Fest 2007)
Message board moderation
Previous · 1 . . . 4 · 5 · 6 · 7 · 8 · 9 · 10 · Next
Author | Message |
---|---|
Send message Joined: 18 Sep 04 Posts: 71 Credit: 28,399 RAC: 0 |
OFF TOPIC I think it's a bit late to be calling OT on a 5-month-old thread. YMMV. :) Please read the first five paragraphs of the first post again. Or read them for the first time, perhaps. Mayhap you'd like to review my participation in the thread itself, eh? There's a good lad. ;-) Whether the project "needs" it from a point of view of the science is OFF TOPIC here OFF TOPIC, you say? My, my. ;-) Frankly, what the project needs is the only real point of interest. I speak as one of the original beta testers. It's my opinion and it's worth what you paid for it. :-) I suspect that the amount of work available would happily be crunched by folks who are happy enough with how the project is currently configured on the server side. If Garfield ever gets off the ground, the discussion becomes moot. yeah fine, you are not worried about it. Tsk, tsk. Smile. It's fun. And, frankly spoken, I take some exception to you jumping on my comment that simply defended BOINC and its scheduling and left 3 weeks of preceeding comments alone. I expect fairer distribution of ranting, matey! ;-) You correctly say it has been hashed over repeatedly. That tells me that the 'argue them into liking it' approach has always failed. Telling people to 'be happy' with something that they perceive to be unfair is never going to work. Was this just a bad hair day or did you really think that me commenting about BOINC scheduling in the client warranted a rant? BOINC scheduling is fine. It does not in any way serve the project for users to be stabbing at the bloody Update button every bloody time their bloody box doesn't have a bloody work unit the bloody instant they bloody well want it. Bloody! ;-) Now, you're talking about something else. Would I like more LHC@home work? Go back and reread my earlier posts in the thread and find out. Or read them for the first time, perhaps. :-P Trying to change what others want from the project is not relevant, nor in my opinion useful. That's interesting because you're wanting to change what I (an "other") want from the project. I want the project to run as smoothly as possible for the science and admins. You want WUs. Pistols at dawn. Don't be late. ;-) And for god's sake, man, smile. |
Send message Joined: 14 Jul 05 Posts: 60 Credit: 140,661 RAC: 0 |
Can we get this damn thread deleted or something? It's about the worst waste of digital space and time spent debating I've seen in a long time. |
Send message Joined: 14 Jul 05 Posts: 275 Credit: 49,291 RAC: 0 |
That's quoting out of context, or READING out of context even. He doesn't want you to change what you want from the project. If you want to care only for the project needs, do it (I may join you), but don't say the other user needs aren't important. BOINC works fine as it is. Just leave it, watch the messages and be happy when the work comes. THAT is what River is complaining about, you can't tell people to "be happy with how it works now". |
Send message Joined: 2 Sep 04 Posts: 39 Credit: 441,128 RAC: 0 |
THAT is what River is complaining about, you can't tell people to "be happy with how it works now". Well, you can. Whether or not anyone listens and agrees is another matter. ;-) Seriously, if the project is happy with their turn-times, then there's really no reason for them to change anything about the way the project is configured. It's not about being "fair" or keeping people happy, neither of which can ever be completely achieved, since they are subjective, and there will always be someone complaining about fairness who is unhappy with the way the project is run. It's about getting the work done. No one has ever explained to me why this is such a big problem. Attach to LHC and 1 or more other projects, and the long term debt will even things out eventually. If/when LHC gets a lot of work, you'll get your share. If you miss out on a short burst of work, it's not going to make that much difference to your stats anyway, in the long run. |
Send message Joined: 30 Sep 04 Posts: 112 Credit: 104,059 RAC: 0 |
This thread has morphed in to something very interesting. I usually don't post here that often, but after reading this thread, I decided to. As I look through this (now) rather long thread I see one recurring theme, no make that two recurring themes. 1.) I see the "Haves" (people that have been here a while and either know the way the system works or have resigned themselves to the way it works) defending the way the current work distribution system is being implemented. I too fall in to this category I guess... 2.) I also see the "Have nots" (mostly new people here 1-2 years or less) complaining about the way the current work distribution system is being implemented and making some demands that something needs to be done to make this project "more fair" to everyone now that we have new admins. It's ironic that the frequency of these types of "requests" only increased AFTER the projects new admins began to post here....hmmmmmmmmm. Well, after almost 3 years of participation in this project, unfortunately it appears that this project leans toward the "work is available when we send it to you" model, no matter who is in charge and I really don't see this changing any time soon because of the sparse nature of the science being generated. Look to the left of this post at my name. See the join date(Sep 30, 2004) below it? It has taken me over 2.5 years just to get to the total that I did, and there was a time when there was NO WORK for the better part of a year, so for ALL those who are new (or recently new) to the project I say to you: "Learn the past history of this project before making demands that are just unreasonable in light of the way this project runs". I personally have 5 machines set to crunch LHC as well as S@H and for the last 3-4 work distributions I have only gotten work on maybe 1 or 2 machines each time. I am not complaining because I understand that this project requires vast amounts of patience (ESPECIALLY with the entire project being moved from one location to another) which some here seem to not have in great quantities. To be successful at LHC requies the understanding that crunching LHC work is a marathon not a sprint and those who only look at the short term are likely to be disappointed here as is evidenced by this and every other recent thread on this topic (at least 5 or 6 that I could quickly see) in the forums. This project IS what it IS and all the complaining about "I want more" & "I want others to get less so I can get some too" is going to basically fall on deaf ears because Neasan & Alex have far more important things to worry about (like a smooth project transition, fixing the broken stats reporting, developing a new sixtrack application, etc...etc...) then the fact that some of you have not received what you feel should be your fair share. There are many other projects out there in the BOINC universe that have a steady and long term supply of work available so why make focusing only on LHC a priority for some?? When LHC has no work (which is often) why not crunch something else along with LHC? I won't make specific recommendiations for backup projects but it shouldn't be too hard to find one that would meet someones needs quite nicely. *** edited to correct spelling, grammar & clarity of thought mistakes.... |
Send message Joined: 15 Jul 05 Posts: 6 Credit: 2,710 RAC: 0 |
There are many other projects out there in the BOINC universe that have a steady and long term supply of work available so why make focusing only on LHC a priority for some?? When LHC has no work (which is often) why not crunch something else along with LHC? I won't make specific recommendiations for backup projects but it shouldn't be too hard to find one that would meet someones needs quite nicely. I run a number of other projects, and periodically check this project to find out if work is available. I have seen a pattern repeating. I visit the website. No work available. Then for a couple days I see unable to connect to the database - too many users. I interpret this to mean there's work but I'm too late - the servers are busy distributing work to the lucky few. Then no work available again. Maybe it's time for me to give up on this project. Nothing travels faster than the speed of light with the possible exception of bad news, which obeys its own special laws. Douglas Adams (1952 - 2001) |
Send message Joined: 30 Sep 04 Posts: 112 Credit: 104,059 RAC: 0 |
tekwyzrd wrote: I run a number of other projects, and periodically check this project to find out if work is available. I have seen a pattern repeating. I visit the website. No work available. Then for a couple days I see unable to connect to the database - too many users. I interpret this to mean there's work but I'm too late - the servers are busy distributing work to the lucky few. Then no work available again. Maybe it's time for me to give up on this project.@tekwyzrd: It's too bad that instead of working within the scope of this project (such as it is) you would instead decide to consider leaving it simply because you don't seem to have very good luck with receiving work. Have you made sure that you do not have a cache set larger then 2-3 days in your preferences? if you have a 10 day cache (or anything larger then 5 days for that matter), chances are you will not get any work. I currently have my cache preference set to 2 days and I seem to be able to get work consistently, but only on 1 or 2 of the 5 PC's I have attached to this project whenever work is sent out. Also, if you suspend the project or have "Won't get new tasks" enabled for the project you will not get any work. If you want to leave that's your right to do so and no one will likely beg you to stay, but your reasoning for wanting to go is not making sense because there's no way for you to not know about how truly sporadic the work is that comes from this project. If you're in it for the credits only, then this is probably not the best project to participate in as someone looking here for the instant gratification that comes from crunching a large number of WU's on a regular basis will most likely be very disappointed. ed. spelling & grammar |
Send message Joined: 22 Mar 06 Posts: 9 Credit: 34,325 RAC: 0 |
The "problem" is self-limiting. The project is way over-resourced for the amount of work it does at the moment. It must be or everyone would get work regularly. The people doing the work are either content with that or they aren't. The people with the lowest threshold leave. If that goes on for long enough, there will be enough work (or few enough workers) to keep everyone busy. Until that point the project isn't getting its limited work done significantly more slowly than they were with lots of workers, most of whom were idle most of the time. There are more than enough workers who are content just to do whatever work comes their way, to more than meet the current requirements of the project. The only possible motivation for risking anything to keep the project over-populated would be if you knew the workload was going to increase enormously sometime soon. But it's my guess that the "silent majority" of contented workers will still be enough even when Garfield shows up. I know how I would change the work distribution arrangements if I was the project administration: not at all, because it isn't broken. Neil |
Send message Joined: 18 Sep 04 Posts: 71 Credit: 28,399 RAC: 0 |
If you want to care only for the project needs, do it (I may join you), but don't say the other user needs aren't important. In the grand scheme of things, I'd say exactly that. It's not possible to keep everybody happy, so the admins should do the obvious: Keep the project happy. THAT is what River is complaining about, you can't tell people to "be happy with how it works now". You can, and those who disagree with the sentiment aren't going to like it. :) |
Send message Joined: 7 Mar 07 Posts: 59 Credit: 7,906 RAC: 0 |
Can we get this damn thread deleted or something? Oh yes, I must concur old chap. And the proverbial most annoying hot dog from the lesser-known Classic of Western Literature, "Perfect Hair Forever" said it even better: "Do the la la la la la la la..." Hoo-hooo, this is my favorite thread ever! Ariel: Certified "Too Cute for LHC" Cruncher! . . . . . . . . . . . . -- Consider the lilies. |
Send message Joined: 15 Jul 05 Posts: 6 Credit: 2,710 RAC: 0 |
@tekwyzrd: It's too bad that instead of working within the scope of this project (such as it is) you would instead decide to consider leaving it simply because you don't seem to have very good luck with receiving work. My motivation for interest in this project is, to put it most simply, to support science for science's sake. This is one of the projects that is truly science oriented rather than a thinly disguised as beneficial to mankind when it is actually designed to use volunteered resources for the benefit some pharmaceutical company. I have my connection interval set to 2 days, the same setting as I've used for over 2 years. I've never used anything over 3 days. I currently have three computers that do work for BOINC projects, and though two of them have P3 processors they are duals. I guess the best way to explain is that it looks like there are obviously more computers doing work for this project than are needed. It's nothing against this project, just a simple fact. Or, some may say, opinion. I'll keep an eye on the site for work but I have it suspended in BOINC. I have noticed that when debt builds it can affect the work that is requested from other projects, sometimes severely enough that I can't get even one full day worth of work. When that happens I zero out all the debts in client_state.xml and requests are back to normal. I also see no sense at hammering away at the scheduler when there's no work. No sense in putting unnecessary stress on the servers. If and when I see work I'll un-suspend this project, and the next time I see the "unable to connect to database - too many connections" on the front page I'll un-suspend and let my computers make their requests along with the multitude of others. Nothing travels faster than the speed of light with the possible exception of bad news, which obeys its own special laws. Douglas Adams (1952 - 2001) |
Send message Joined: 4 Sep 05 Posts: 13 Credit: 536,862 RAC: 0 |
Fairly new with this project, but I see WUs in my results. The technique that works for me is: Connect to 13 projects, set cache size to .001 days and look at the stats page every so often. I've had to raise the resource share for CPDN so that it will finish on sched & on my old 1.8Ghz P4 limit the projects worked on, but aside from that there is no micro-managing necessary. Work comes in...WUs are crunched...Credits are added to the various accounts as earned. I don't see 3+ page lists on the tasks tab ... maybe 1 or 2 at a time per active project, perhaps 7 or 8 projects with a WU or 2 in progress at any given time. Looking at results pages I occasionally find impressive lists of same day returns, but I find the credits add up quite nicely while I'm ignoring the whole thing :P |
Send message Joined: 14 Jul 05 Posts: 275 Credit: 49,291 RAC: 0 |
For your information, BOINC has no way to set the cache size. The setting is "connection interval". Setting it higher may end in SMALLER CACHE in some cases. Specifically, to avoid missing deadlines: if you set it to 10 days, it assumes you won't report your units to the server until 10 days from now. |
Send message Joined: 29 Dec 06 Posts: 100 Credit: 184,937 RAC: 0 |
Yes, it is important to let k00ks know they are k00ks. Otherwise they continue posting their k00k crap. Their nonsense is annoying, antagonising[sic] and insulting. Feel the love. If only Barry White could be here... He'd set things right. Postin' for Love *Sigh* Fire. Fire, fire, fire. Yeah. |
Send message Joined: 18 Sep 04 Posts: 71 Credit: 28,399 RAC: 0 |
You post that crap as if you're the only one in this forum who has figured that out. You, sir, have a gift for not endearing yourself to others. You would also do well to skip the profanity. |
Send message Joined: 18 Sep 04 Posts: 71 Credit: 28,399 RAC: 0 |
Odd, the k00k whiners who masquerade their ridiculous little yearnings Troll. Go away. 0 credits, 0 RAC, 0 credibility. I will not lose a moment of sleep fretting over the fact I have failed to endear myself to selfish imbeciles. If you would refrain from personal attack and insults, you might have SOME credibility. As it is, you have none. Troll. As for profanity, if you don't like words like crap then don't write crap Crap wasn't the word to which I was referring. It just happened to be the one I quoted. Have a splendiferous day, troll. |
Send message Joined: 18 Sep 04 Posts: 71 Credit: 28,399 RAC: 0 |
Seems that you've been endearing yourself over at CPDN, too: I really can't believe the kooks and whiners are posting their nonsense out of concern for this project. Their kook solution is so completely riddled with holes even fish laugh at it. Nobody can actually be that stoopid and remember to breath. So... their only intention for posting that crap is to antagonise and annoy other people. They do it deliberately, They do it repeatedly in the face of all reason which I find to be aa very hostile kind of behavior. So ban them for 30 days. Dagorath strikes again, eh? Full thread here: http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_thread.php?id=5307 |
©2024 CERN