Message boards :
Technical News :
Small Word (Sep 20 2007)
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 . . . 8 · Next
Author | Message |
---|---|
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
Well with people caching more work it means that you are more likely to get paired with a slow computer (slow to return result) rather than a fast one. That's life. Actually, if you "recruit" Aunt Bertha's computer and give it a weeks cache, BOINC will figure out how much it runs, and adjust the cache size appropriately. One WU is more than one weeks' work. |
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
The deadlines for that are just far too long. When we had short deadlines, people got all panicky about "panic mode." EDF is your friend. It allows shorter deadlines without loss of work. |
ML1 Send message Joined: 25 Nov 01 Posts: 21241 Credit: 7,508,002 RAC: 20 |
I hope you will agree, from my posting history, that I'm not a credit whore (and I'm sure that you weren't directing that comment at me - no offence taken). The comment was certainly general, especially so being as the credits race has been discussed by many people so often. There is amazing enthusiasm for this project by some people, which I think is great stuff. Witness all the passion that explodes from the forums on occasion also! I consider both the project and the science to be very worthwhile. There is even science in how all this and the science are being done. With all that, this is still a volunteer project and Berkeley always suffer a little from the participants getting a little 'too enthusiastic' to the point of trying to push the project beyond what Berkeley can ever hope to offer. We are all still very much helping Berkeley due to their provision of the infrastructure. As an aside, a lot of participants also get a lot of fun and even education out of all this! I agree that BOINC is designed to tolerate and recover from stresses like this, but I still think, and say, that it's antisocial to test those stress limits when they can be avoided with a little thoughtfulness. I see a LOT of thoughtfulness go into how the people (one person often) at Berkeley go about juggling the systems that they have. It is quite a story just for getting the few snippets that we do get for how they are straining whatever hardware, network, OS, filesystem, or whatever beyond normal operation, and still keeping it all going. Sometimes (often?) with out-of-hours unpaid working... I'll admit that I've wondered how secure the master science database is against HDD corruption. Then again, there's postings to reassure us that indeed backups are run. I just hope they also have backups OFFSITE to guard against fire, theft, earthquakes, meteor strike, or 'whatever'... The Berkeley team have been juggling all this for over a decade. I should think that they have a very good idea of what they are doing! Much more so than many 'professionals'... Often with Science, there is also lots of Patience! Happy crunchin', Martin See new freedom: Mageia Linux Take a look for yourself: Linux Format The Future is what We all make IT (GPLv3) |
Simon Wiesenthal Send message Joined: 8 Feb 02 Posts: 105 Credit: 80,113 RAC: 0 |
I don't get those who attack those who, like myself, wonder, "Why *should* I waste my computer, waste power, and the money spent on power, pay for replacement parts, etc., on work units that are never used, because Berkeley doesn't have its act together?" It would be one thing if it were the occasional, rare occurence. No, it's continual. What *would* be irrational would be just to say, "Oh, no no, I don't mind wasting my money on work that's never used. You keep up the good work, failure-prone though it is. I have infinite patience." |
Simon Wiesenthal Send message Joined: 8 Feb 02 Posts: 105 Credit: 80,113 RAC: 0 |
Nope, not done venting. In fact, I'm just getting started. When I first joined this project back in 2001, built up some credit, lost it with an email address change, rejoined in 2002, things were fairly cool here. You had project people saying things like, "Oh, no, we don't want you running 24/7 just for SETI@home's sake; we want you to be green." Well, I'm an enthusiast, and not a big believer in my machine's capability to destroy Earth, so I ran 24/7. Recently (within the last two or three years), there has been almost a paradigm shift in attitude. Now it's all about who has the biggest, fastest collection of machines. Gone is the, "There's this old man, who has a Heathkit and a 300-baud modem, who is so proud of the handful of work units he's able to contribute in a year." No, now it's some kind of juvenile pissing contest between junior-high-like -- albeit vicious -- little geeks with nothing better to do. There was a time when the molasses-slow machines and B/Ws were a hefty part of the project. You've managed to drive most of them off. Congratulations. Maybe we can get the U.S. Treasury to mint you a coin, Heroes. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14679 Credit: 200,643,578 RAC: 874 |
I see a LOT of thoughtfulness go into how the people (one person often) at Berkeley go about juggling the systems that they have. It is quite a story just for getting the few snippets that we do get for how they are straining whatever hardware, network, OS, filesystem, or whatever beyond normal operation, and still keeping it all going. Sometimes (often?) with out-of-hours unpaid working... Absolutely - couldn't agree more. If only the same thoughtfulness could be exhibited by the user community (crunchers, or sometimes ex-crunchers). Maybe then the Berkeley people would have to do less fire-fighting, and could get on with the scientific analysis Simon so rightly reminds us of. |
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
I don't get those who attack those who, like myself, wonder, "Why *should* I waste my computer, waste power, and the money spent on power, pay for replacement parts, etc., on work units that are never used, because Berkeley doesn't have its act together?" One big change from Classic to BOINC: When classic ran out of work, it simply recycled what it had. Most were happy because they were always crunching, even if the work had been adequately crunched. BOINC will let you go idle instead of providing "busy work." Unlike Classic (without add-ons) BOINC will queue work for times when servers are down. It's all much better than it was under Classic. Your other statement: Berkeley's position has not changed. BOINC (and SETI) are designed to exploit a waste product -- idle CPU cycles. Certainly, there are many folks here that have spent a lot of money to create waste CPU cycles -- and the race for credits is huge motivation for some. ... but Berkeley isn't asking for machines built just for SETI. In fact, things would probably be better for SETI if 20% of that money had been donated (to bring in better servers) than spent on hardware "at home." |
ML1 Send message Joined: 25 Nov 01 Posts: 21241 Credit: 7,508,002 RAC: 20 |
Nope, not done venting. In fact, I'm just getting started. Take a look at the "P90" thread. Excellent fun! And don't be put off by the (one) Giga-Geek working all hours to show off the fastest one PC in the world. (Then again, was there not a liquid nitrogen cooled Overclock of some system to 5 GHz + last year?) There's still a lot of fun and passion and Science in this game. Welcome onboard, even if the forums are a little 'hot' sometimes. Happy crunchin', Martin See new freedom: Mageia Linux Take a look for yourself: Linux Format The Future is what We all make IT (GPLv3) |
kittyman Send message Joined: 9 Jul 00 Posts: 51478 Credit: 1,018,363,574 RAC: 1,004 |
Nope, not done venting. In fact, I'm just getting started. Simon....your opinions may be well taken, but your rhetoric is a bit, shall we say, aggressive. I am 50 years old, and am not sure if I should be referred to as being part of a "juvenile pissing contest between junior-high-like -- albeit vicious -- little geeks with nothing better to do." I crunch Seti for the science, but it has also become a hobby and somewhat of a sport for me to see how much crunching I can eke out of my little 8 rig farm. And I do not think I have ever been vicious to anybody ever in this forum. Believe it or not, the "molasses slow" machines crunching away in the background contribute the lion's share of the work done here. All contributions are worthwhile and welcome here. It is just that the current stable of ever-faster crunching rigs in the top 100 computer listings are the most visible. And because their owners tend to be very involved in the project, they are also the most vocal as well. And I am very proud of what I am contributing here! I hope you can understand that some folks' enthusiasm for the project is different from yours. I have contributed money directly to the project as well as spending money for new computers for my own enjoyment. Computers have changed, Seti has changed, but the goal remains the same....to find some proof that there is life beyond ours in existence somewhere. I hope you will remain involved, to whatever extent you choose. And the kitties say...."Just keep 'em crunching!" "Time is simply the mechanism that keeps everything from happening all at once." |
ML1 Send message Joined: 25 Nov 01 Posts: 21241 Credit: 7,508,002 RAC: 20 |
Nope, not done venting. In fact, I'm just getting started. Ooooooops! Downgrade that by one third! That should read "Take a look at the "P60" thread." Happy crunchin', Martin See new freedom: Mageia Linux Take a look for yourself: Linux Format The Future is what We all make IT (GPLv3) |
Andy Lee Robinson Send message Joined: 8 Dec 05 Posts: 630 Credit: 59,973,836 RAC: 0 |
That should read "Take a look at the "P60" thread." That thread should be taken out and shot for wasting so much time and energy! |
Palomar Jack Send message Joined: 9 Nov 04 Posts: 44 Credit: 503,405 RAC: 0 |
So, you're saying that this 754 pin AMD Athlon 3200 is beating the pants off a whole raft of Core II Duos and quad? I appreciate the compliments, but I don't buy it. The hardware is overclocked, true, and Win XP has had every service not needed killed, true, but it still won't even be close. Those machines should have turnarounds at least 1/4 of this one and RACs at least four times higher, but they don't, and the science and we sit and wait a month for a Core II, duo, or quad to return work. "When all possibilities are eliminated what remains, however improbable, must be the truth." And that's life, too. |
ML1 Send message Joined: 25 Nov 01 Posts: 21241 Credit: 7,508,002 RAC: 20 |
... and the science and we sit and wait a month ... Are those hosts just running a maximum sized cache? Meanwhile, we've waited many years for the radio signals to arrive at earth, and then a while longer again for the data to then be sent to us for processing, a small further wait might be irritating but no big deal. Neglected hosts are only a small nuisance. And Boinc is deliberately designed to allow for unreliable hosts. There's more to the Big Crunch than just credits :-) (And you get the pleasure of a RAC boost when any delayed credits do roll in later! ;-) ) Happy crunchin', Martin See new freedom: Mageia Linux Take a look for yourself: Linux Format The Future is what We all make IT (GPLv3) |
Andy Lee Robinson Send message Joined: 8 Dec 05 Posts: 630 Credit: 59,973,836 RAC: 0 |
And Boinc is deliberately designed to allow for unreliable hosts. Deliberately designed badly! The current deadline system is a development short cut to get the system quickly up and running. Now that the system is running, it should be revised and done properly. |
ML1 Send message Joined: 25 Nov 01 Posts: 21241 Credit: 7,508,002 RAC: 20 |
... The current deadline system is a development short cut to get the system quickly up and running. What needs doing differently or improving? Cheers, Martin See new freedom: Mageia Linux Take a look for yourself: Linux Format The Future is what We all make IT (GPLv3) |
Andy Lee Robinson Send message Joined: 8 Dec 05 Posts: 630 Credit: 59,973,836 RAC: 0 |
... The current deadline system is a development short cut to get the system quickly up and running. Simply: If boinc detects a user disappearing without 'agreement' for a few days, his uncrunched workunits automatically return to the pool. He can download new ones on resumption. This is no inconvenience. By agreement, I mean anyone not having indicated they'll be processing offline. Likewise, if a user knows he's going away and the machine won't be crunching in his absence, he can push a button to return uncrunched WUs, instead of using the too drastic detach from project. This is a day's programming for an intelligent programmer to sort out on the server side. I believe clients already have the capability to react to recalls in one form or another, in the form of cancellations or deletions, so they can automatically be removed from their queues. This is pretty much a model of checking out books from a library. If you know you're going to emigrate, go blind or get murdered, then a process exists to return the books before the deadline and avoid a financial penalty, which doesn't exist in boinc's case! This is just so beneficial to everyone and logical to me, I just can't imagine why anyone wants to resist it. |
PhonAcq Send message Joined: 14 Apr 01 Posts: 1656 Credit: 30,658,217 RAC: 1 |
The system needs to be intervention free; your description requires some manual case handling, which will certainly fail to meet the goals. "Set and Forget" should be the watchword. Is there a simple flowchart online regarding this process so that we can understand what is intended today? I notice I'm getting a lot of wu's that time out in a couple of months, eventhough most of my computers have a 1w or less turn around (due to my current cache settings). On the face of it, a two month time out doesn't bother me, but seems to bother a lot of people looking for a responsive rac metric. |
kittyman Send message Joined: 9 Jul 00 Posts: 51478 Credit: 1,018,363,574 RAC: 1,004 |
The system needs to be intervention free; your description requires some manual case handling, which will certainly fail to meet the goals. "Set and Forget" should be the watchword. OK.....just to throw out a thought. How about adding another stat for the hosts. Let's call it 'Potential RAC'. Calulate the RAC for the hosts including the pending credits, should they be awarded as claimed. The current RAC would still be all that really matters, but it might be interesting to have a 'Potential RAC' taking into account what is waiting on the wingmen. "Time is simply the mechanism that keeps everything from happening all at once." |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13855 Credit: 208,696,464 RAC: 304 |
The system needs to be intervention free; your description requires some manual case handling, which will certainly fail to meet the goals. "Set and Forget" should be the watchword. I agree. At most the deadlines could be changed to better reflect actual processing times. But making things even more complicated, just so people get credit sooner, isn't a good move. Grant Darwin NT |
ML1 Send message Joined: 25 Nov 01 Posts: 21241 Credit: 7,508,002 RAC: 20 |
OK, a few of my ramblings: ... If boinc detects a user disappearing without 'agreement' for a few days, his uncrunched workunits automatically return to the pool. He can download new ones on resumption. ... Sounds like a 'heartbeat' check between the project servers and the clients. ... Likewise, if a user knows he's going away and the machine won't be crunching in his absence, he can push a button to return uncrunched WUs, instead of using the too drastic detach from project. Well, a 'detach' shouldn't be necessary in any case. The 'will be away, on holiday' idea has been discussed before. I don't know how JM7 (or someone else) can add anything more meaningful than the existing "No New Work" setting. This is a day's programming for an intelligent programmer to sort The code is Open Source and Dr A is very amenable to people offering patch files for new code. I believe clients already have the capability to react to recalls And that could be an alternative method whereby you have a "dump all unstarted WUs" and report that they have been dumped back to the project servers. ... return ... before the deadline and avoid a financial penalty, which doesn't exist in boinc's case! The circumstances are somewhat different to that of a library. The overhead of reissuing WUs is just the cost of server-side storage and retransmission, a very small cost. This is also a volunteer effort so there cannot be the idea of 'penalties' other than to protect the project from overt abuse. The present mechanism of reissuing a WU after a timeout is about the best option that I can think of unless you are going to implement some sort of 'heartbeat monitoring'. My thoughts are that the easiest to implement for those very conscientious participants is the 'dump unstarted WUs' GUI button. However, using "No New Work" earlier would be even more conscientious. And do not some of the stats sites offer a statistic for pending credit? Happy crunchin', Martin See new freedom: Mageia Linux Take a look for yourself: Linux Format The Future is what We all make IT (GPLv3) |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.