Message boards :
Number crunching :
The 'Zero CPU' problem ... !!!
Message board moderation
Previous · 1 · 2 · 3 · Next
Author | Message |
---|---|
Send message Joined: 18 Sep 04 Posts: 143 Credit: 27,645 RAC: 0 |
> "I'm not kicking out a bunch of errors, they're just fast crunchers. It hurts > me to see them starving for work :(" Taking more than his fair share? I don't > know, but taking how ever many his machine can crunch through sounds fair to > me. True at that Puff, but looking at the farm of 15 machines, the comment may have been correct. Jord BOINC FAQ Service |
Send message Joined: 30 Sep 04 Posts: 112 Credit: 104,059 RAC: 0 |
Actually Ageless, you may be incorrect in that statement.. Client machines can be duplicated in the S@H BOINC project and regularly need to be 'merged' to consolidate all the scattered results. At one time last week, S@H BOINC reported that I had 11 computers under my account which is incorrect because I only have 3. I merged the other results and now things are reporting correctly. I suspect that the same thing can/may happen to LHC.. EDIT: There IS a facility to merge hosts in LHC |
Send message Joined: 2 Sep 04 Posts: 352 Credit: 1,393,150 RAC: 0 |
I have 18 Computers listed in my Account right now when in fact I only have 7 attached to the Project. Some of them could be merged and some not because of how the newer clients read & report the CPU's now ... |
Send message Joined: 2 Sep 04 Posts: 165 Credit: 146,925 RAC: 0 |
> "Please Please please pleeeeeaaaaasssseee increase the units per day." dosen't > really sound like complaining to me... More like a polite request since one of > the two reasons for the quota is (likely) gone. > Actually, with the next version of the server code, computers that return a bunch of garbage WUs will be penalized. Each time your computer returns an error, that computers WU/day quota is reduced by one. The minimum is one. Each time that the computer returns a properly crunched WU, the quota goes up by one. The maximum is the project WU count. This should allow the project administrators to safely increase the maximum WU/day for most of us as a real problem will download one day max worth of work and then be at one WU/day until the problem is fixed. This is substantially better than the current system where the max is fixed as it penalizes the worst offenders and reduces the amount of trashed work sent to the real problems. Of course to keep getting work, your machine has to return more than 50% error free results. BOINC WIKI |
Send message Joined: 2 Sep 04 Posts: 352 Credit: 1,393,150 RAC: 0 |
Thats a good idea John but it's not going to stop anybody from getting WU's that have half a brain ... ;) |
Send message Joined: 24 Oct 04 Posts: 79 Credit: 257,762 RAC: 0 |
> Actually, with the next version of the server code, computers that return a > bunch of garbage WUs will be penalized. Each time your computer returns an > error, that computers WU/day quota is reduced by one. The minimum is one. > Each time that the computer returns a properly crunched WU, the quota goes up > by one. The maximum is the project WU count. This should allow the project > administrators to safely increase the maximum WU/day for most of us as a real > problem will download one day max worth of work and then be at one WU/day > until the problem is fixed. This is substantially better than the current > system where the max is fixed as it penalizes the worst offenders and reduces > the amount of trashed work sent to the real problems. Of course to keep > getting work, your machine has to return more than 50% error free results. > And does the server notify you that your workunit(s) are trash? How would you know the difference from errant workunits from the project side? LOL sounds like democracy American style! |
Send message Joined: 2 Sep 04 Posts: 165 Credit: 146,925 RAC: 0 |
> Thats a good idea John but it's not going to stop anybody from getting WU's > that have half a brain ... ;) > Since it is the server keeping track, it would take a hack of the project server to get WUs if it decided you shouldn't. BOINC WIKI |
Send message Joined: 18 Sep 04 Posts: 143 Credit: 27,645 RAC: 0 |
> Actually Ageless, you may be incorrect in that statement.. > > Client machines can be duplicated in the S@H BOINC project and regularly need > to be 'merged' to consolidate all the scattered results. > I know that, and looking at his farm, he may only have 3 or 4 real computers, but still, those are as he says fast crunchers. If he wants to use them on only LHC, not on anything else, his question is a good one. I can't say anything about his patience or about this project being in user-defined Beta. If the project Devs say this project is in a free status but want people to look elsewhere if they run into too many problems, then it isn't an LHC wannabe cruncher's fault. ;) Jord BOINC FAQ Service |
Send message Joined: 29 Sep 04 Posts: 25 Credit: 77,910 RAC: 4 |
While credits are good for the team, the real reason I leave my machines on 24/7 to crunch for these projects is the science. And, of course, the pretty screensavers... ;) -Gene |
Send message Joined: 27 Sep 04 Posts: 34 Credit: 199,100 RAC: 0 |
Regarding WU, If I remember correctly, here at LHC the WU generation works like this: CERN Scientists give WU to LHC team; LHC team gives them to Us; LHC team receives results and gives them to CERN Scientist. Then they Analyse/Process them and only after this new WU are generated. So while the results are being analysed/processed to later generate more WU, the project has no work. Professor Desty Nova Researching Karma the Hard Way |
Send message Joined: 2 Sep 04 Posts: 352 Credit: 1,393,150 RAC: 0 |
So while the results are being analysed/processed to later generate more WU, the project has no work. ========== Yes, before there were times when they would not have any work for 2-4 days ... |
Send message Joined: 2 Sep 04 Posts: 309 Credit: 715,258 RAC: 0 |
Yes, before there were times when they would not have any work for 2-4 days ... ============ And once there was work, all our hungry crunchers would suck tens of thousands of wu's out of CERN in no time what-so-ever......ahhh, those were the days! |
Send message Joined: 17 Sep 04 Posts: 190 Credit: 649,637 RAC: 0 |
> And once there was work, all our hungry crunchers would suck tens of thousands > of wu's out of CERN in no time what-so-ever......ahhh, those were the days! > > *sniff*! I wish it would be possible! But please consider, all those tens of thousands must be returned for getting the credits, -- em sorry to return the scientific work... |
Send message Joined: 1 Sep 04 Posts: 10 Credit: 57,389 RAC: 0 |
> Believe it or not, there are people that get introduced to a project and do > not know about others. I simply chose to offer a way out of your > "predicament". If a person knows that boinc even exists, they must know that multiple projects exist. My statement stands as is. > > Instead, no, you complain, try to get the quotas raised and deprive other > people of wu's because you are sucking more then your fair share - that sounds > real nice. Enjoy. Where was I complaining? Asking to have the limit increased is complaining? Fair share? 15 machines running full out and not starving for work *is* my fair share. As long as what I'm returning is acurate science, I have every right to _ask_ to keep my 24 hour queue filled as long as there is work in the project queue (at the server end). I did not demand. I did not complain. I asked (well, okay, I begged ;-) ). <a href="http://www.boincsynergy.com"><img src="http://www.boincsynergy.com/images/stats/112lhc.png"></a> |
Send message Joined: 1 Sep 04 Posts: 10 Credit: 57,389 RAC: 0 |
> Actually Ageless, you may be incorrect in that statement.. > > Client machines can be duplicated in the S@H BOINC project and regularly need > to be 'merged' to consolidate all the scattered results. Nope. there are 15 unique PC's. 10 brand new P4 3.0E's, 3 P4 2.8's and a pair of P4 2.4's. I keep a fairly close eye on duplicates I've been involved with Boinc since berkeley's initial beta test, and I'm on the alpha team. I'm well aware of duplicates and many other issues. <a href="http://www.boincsynergy.com"><img src="http://www.boincsynergy.com/images/stats/112lhc.png"></a> |
Send message Joined: 1 Sep 04 Posts: 10 Credit: 57,389 RAC: 0 |
> I can't say anything about his patience or about this project being in > user-defined Beta. If the project Devs say this project is in a free status > but want people to look elsewhere if they run into too many problems, then it > isn't an LHC wannabe cruncher's fault. ;) I've been with Boinc since.... god I can't remember it's been so long. I've got lots of patience. I can't see where I might have given the impression that I'm impatient. I merely asked (ok, begged) the project to up the quota. Suddenly I find myself in the usual "Oh you spoke! Everyone jump on him quick!" scenario. I suddenly feel like I've been JQ'd for asking a valid question :( <a href="http://www.boincsynergy.com"><img src="http://www.boincsynergy.com/images/stats/112lhc.png"></a> |
Send message Joined: 30 Sep 04 Posts: 112 Credit: 104,059 RAC: 0 |
> I've been involved with Boinc since berkeley's initial beta test, and I'm on > the alpha team. I'm well aware of duplicates and many other issues. Sounds like you're all on top of things then. I wasn't presuming you didn't know, just pointing this out IN CASE you didn't. Every BOINC project's just a 'lil bit different then the next one and you'd be surprised what people think they know (or don't) about a given project. |
Send message Joined: 29 Sep 04 Posts: 187 Credit: 705,487 RAC: 0 |
Actually... --- "Oh you spoke! --- ... that's what I did, and... --- Everyone jump on him quick!" --- ... that is what you did ;) Wave upon wave of demented avengers march cheerfully out of obscurity into the dream. |
Send message Joined: 2 Sep 04 Posts: 16 Credit: 15,568 RAC: 0 |
> > Thats a good idea John but it's not going to stop anybody from getting > WU's > > that have half a brain ... ;) > > > Since it is the server keeping track, it would take a hack of the project > server to get WUs if it decided you shouldn't. > Couldn't one just detach & reattach the offending computer to get a full load of workunits each day??? |
Send message Joined: 2 Sep 04 Posts: 352 Credit: 1,393,150 RAC: 0 |
Couldn't one just detach & reattach the offending computer to get a full load of workunits each day??? ========== I don't know if the Server would let you get away with that but there are other work arounds that I know will work without Hacking the Server like John said ... ;) |
©2024 CERN