| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: Suggestions / Feedback
Thread: Eliminate one reason why members stop contributing |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 75
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello fellow crunchers! Hello Grid technicians!
I wish, that really everyone who knows the internet would also know Worldcommunitygrid, because your site is really great. But nevertheless, it seems hard to gain new members. This makes it even more important not to loose members we got. I am sure, that the main reason for leaving WCG is the expirience of sending results back to WCG "Too late" This is the most frustrating expirience you can get with WCG. So I would like to propose that each new project in WCG should have a batch of workunits for people using their Computer only five minutes a day. This batch should contain special workunits with a time limit of not less than a year. In other words: Users should have one year of time to send them back to WCG. And there should be a check box/option in the prefernces with the name "I need more time to do my workunits". In this way, crunchers can deliberately choose to take and process workunits from the "one year batch", i.e. the batch with a time limit of one year. What do you think about that? Thank you in advance for comments and help to make this idea real. All the best for everyone out there. Good bye. MS |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
An extended deadline for the truly 'part-time' or rather 'sporadic', would be appreciated. Somewhere it said quiet some time ago there was automatically more time awarded for the slower returners, but if that rule is still in effect?
Trouble is, the quorum situation. There's presently no matching of computing power, so one may return in 6-12 hours and would be waiting up to the wingman deadline before an extra 'no reply' copy would be send out. This is something to be addressed first i think, matching of return speed and then designate these at batch by batch level to different computing power / return speed groups. This way, scientists could get prioritized work returned fast, and the stuff not in so much of a rush come back later. Is the boinc work distribution software equipped to do this? Don't know, but it sounds like lots of development. you have to start somewhere though to address the rot, which i saw in a workshop presentation by david anderson is a boinc global issue. More work done due power/core increases, by ever fewer. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
...as some of the scientists are adapting the work units as earlier ones are returned I cannot see any of them waiting a year for a result in order for them to place a new updated batch in the pipeline. It would be alright for a project such as seti as there are light 'years' between the stars so waiting an extra year for a result in something that has take light 50+ years or so to reach us would probably be ok.................imagine a big 'push' on something like an ebola vacine and waiting over a year for a result......no thanks!
 |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello scribe,
I do not want to argue with you, but: Almost all projects last more that a year anyway. So I think that a special batch for slowtimers would not prologue the necessary project time in the end. And: Are you really sure that in all projects such adaptations in reation to work units are being made? Yours sincerely Martin Schnellinger |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
So I would like to propose that each new project in WCG should have a batch of In the dim and distant past one of the original medical project at United Devices (2001) had an advertising slogan along the line "Screensaver, Lifesaver" where the obvious slant was that your PC did work while the screensaver was running, thereby utilising "Spare" cycles. From this it is obvious that the machines were left on for more than 5 minutes, having a machine that is only on for 5 minutes per day is useless under the circumstance you outline. It probably never reaches a checkpoint so it would never complete in 10 years, let alone 1 year!workunits for people using their Computer only five minutes a day. [Edit 1 times, last edit by Former Member at Oct 5, 2014 11:03:10 AM] |
||
|
|
KLiK
Master Cruncher Croatia Joined: Nov 13, 2006 Post Count: 3108 Status: Offline Project Badges: 
|
although I commend the idea...it's not viable!
----------------------------------------why not? 1. Scribe told you...use the spare cycles! 2. my machines goes 24/7...and they were built that way...no overclocks...just stable parts! although they have weekly shutdown option built in! 3. all my customers to whome I repair computer from time to time...i tell them to turn ON the computer in the morning before the coffe or tea...so it's ready to work on when they arrive...and leave it working till they go to sleep...so the time wasted on boot up is next to nothing! also, computers are stressed mostly when the are turned on...so too much turning on, turning off...not good for the machine! 4. I have a machine that is dual-celeron @533MHz...an old BP6 board that I like...how much time would it take - correction, how much cycles (in FLOPs) would it take for a 5min, 20d per week, take to finish calculations on that machine? maybe somebody else uses Intel Atoms? Can't compare that from my sister small laptop to Xeon that I use on the computer that I type from... too much variables... |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello everyone,
Please let me point out, that the period of one year was and is only a proposition. Mabe half a year would do, too? And: It would really be interesting, how many users no longer contributing are in WCG's database. And: How many of them have resingned quickly (not returning any result) finding out that their computer is too weak and used too little time, resulting in..... result too late. Scribe, please do not have bad humour. My appriciation and congratulations for your huge, huge contribution!!!! Yours Martin Schnellinger |
||
|
|
jonnieb-uk
Ace Cruncher England Joined: Nov 30, 2011 Post Count: 6105 Status: Offline Project Badges: 
|
I am sure, that the main reason for leaving WCG is the expirience of sending results back to WCG "Too late" This is just one of any number of reasons why people might leave, or have left, WCG. There is no publicly available evidence to prove that this, or any other suggestion, is the "main" reason. To have a result rejected as "too late" implies that the user has returned at least 1 completed WU. What about the thousands of users who download BOINC, attach WCG as a project and never return even 1 completed WU? I would suggest a more credible reason might be found in the length of WUs rather than their deadline. We live in an age where "instant gratification" ranks highly. A WU which has an estimated completion time of 3 hours or less would surely be more attractive to new users than one which is estimated at several hours longer. In August 2012 FAAH WU runtimes (and most other projects) were deliberately lengthened by 20% so it should be possible to reverse this for FAAH. If necessary these shortened WUs might be allocated to a seperate feeder which is only accessed by new users. A further step might be to make FAAH (with it's shortened WUs) the default project selection for new users. ---------------------------------------- [Edit 1 times, last edit by jonnieb-uk at Oct 5, 2014 2:25:22 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The 'deliberate' increase -and- decrease of work unit content to compute has been done many times, before and after that 20 percent case, to relief the validators and sub systems at the back end when things were tight. The proven capability on a sustained basis was at some point 1.75 to 2 million in results daily, such as last december. We did a last week average of 1 million, i.e. there's room to shrink sizes. Bet you the current position is one of 'ugm is coming, so we won't, until that's running', and all the other stuff wcg just does not want to share with the members, to quell these never ending 'negative toned' discussions.
----------------------------------------The senior architect is invisible, which i'd think is the better positioned person to do a frequent technician post series. At least knreed would in at times come to expand a little, but no, central mojave desert it is on the 'what we're doing' front. [Edit 2 times, last edit by Former Member at Oct 5, 2014 9:46:58 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello everyone,
thank you for your comments. My proposition is not urgent at all, but something basic and to do on the long run. Maybe I will come back to it in half a year? I will go on contributing as long as possible, whether my proposition will become reality or not. To me health related science is of top priority. One more word: The person who invented distributed computing is a genius. Three times hurray for him. Good night Martin Schnellinger |
||
|
|
|