| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: 25 projects only 1-10 running |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 19
|
|
| Author |
|
|
bdp003
Cruncher Joined: Oct 2, 2007 Post Count: 10 Status: Offline Project Badges: 
|
Ok, another general question.
How to get more CPUs running your project. That is how to invest a sum of money, buy older cheap computers averege aged or new hefty ones with highest cpu rate. Are there any testimonies out there how other BOINC WCG crunchers set up their park of computers. /B |
|||
|
||||
|
bdp003
Cruncher Joined: Oct 2, 2007 Post Count: 10 Status: Offline Project Badges:
|
And how abt 64 bit CPUs
Are they worth the money in crunching increased speed? /B |
||
|
|
retsof
Former Community Advisor USA Joined: Jul 31, 2005 Post Count: 6824 Status: Offline Project Badges: 
|
That is how to invest a sum of money, buy older cheap computers averege aged or new hefty ones with highest cpu rate. Older cheap computers is definitely the worst. They are best at heating your house during the winter, though. I'd even pass on average aged, because the newer chips are much more economical on power consumption. There's another option...new economical computers. That's new average or new better than average. I didn't say the cheapest. Most of the cheapest have stripped down memory and hard drives and motherboard graphics, which steals part of main memory. For 32 bit computers, memory is cheap. Install 4 Gigabytes and be done with it. AMD computers do better at floating point calculations. My fast X2 3.2GHz computer can keep up with an older Intel Quad. Clock speeds are not directly comparable. I have a couple of single core AMD 2.4 GHz computers that also run other applications. A newer 2.0GHz chip computer here runs much cooler happily overclocked to 2.4GHz than the other two do at stock speed. I wish that AMD would get moving on increasing the clock speed of their 4 core Phenom machines. When it matches 3.2GHz core speed of my X2 I will probably get one. Naturally, the newer Intel machines, especially the bleeding edge ones made for gaming will do more, but at a vastly increased cost. Three or four economical computers for the same money would probably contribute more workunits. Three computers here are on a KVM switch, so only need one monitor, keyboard and mouse. The cost/speed ratio is not a straight line. You pay through the nose for the fastest possible, which is only a small increment over the next fastest. WCG does not have any 64 bit applications at the moment, so doesn't gain anything from those operating systems to date. 64bit computers mainly have a larger memory capacity due to a larger addressing capability.
SUPPORT ADVISOR
----------------------------------------Work+GPU i7 8700 12threads School i7 4770 8threads Default+GPU Ryzen 7 3700X 16threads Ryzen 7 3800X 16 threads Ryzen 9 3900X 24threads Home i7 3540M 4threads50% [Edit 5 times, last edit by retsof at Nov 19, 2008 4:26:48 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The bad part with 60 minute switches is, not quite because BOINC normally switches only at checkpoint saves, that you get many project swaps, loosing minutes a day just on disk I/O. I've got it set to 240 minutes. So if I had 24 jobs running in 4 hour batches where each job needed 24 hours, wouldn't I still have the problem where the jobs wouldn't finish in less than 24 days? Is it that Boinc knows that the job needs finishing so it doesn't switch projects at the end of the 4 hours? |
||
|
|
Thomas515
Cruncher Joined: Aug 7, 2006 Post Count: 22 Status: Offline Project Badges: 
|
I´m crunching for some of this projects to. boincsimap has only work at the beginning of the month, most of the work is done, now there are only some "updates". LHC-workunits are rare and short, about 20 minutes.
Greetings from germany |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
The bad part with 60 minute switches is, not quite because BOINC normally switches only at checkpoint saves, that you get many project swaps, loosing minutes a day just on disk I/O. I've got it set to 240 minutes. So if I had 24 jobs running in 4 hour batches where each job needed 24 hours, wouldn't I still have the problem where the jobs wouldn't finish in less than 24 days? Is it that Boinc knows that the job needs finishing so it doesn't switch projects at the end of the 4 hours? You cant have 24 jobs each running 24 hours unless you have an Octo ;>). The BOINC buffer limit, based on what it knows on current performance (CPU efficiency), daily "on" hours, number of cores, and projected job duration is 10 days, period. That would be one logical reason why WCG uses 12 days as a safety for standard jobs so whatever is fetched should be finished in time, with some system downtime considered. Everything on the Work Buffer / Cache is found under the "Cache aka Additional Work Buffer" topic reached via the Start Here FAQ index. NB: CPDN is an exception. They run for weeks/months and will monopolize the device for longer times if the client computes that the scheduler frequency till deadline is insufficient to complete them. [Added: One exception: If initially attaching to a new project, the client will call for 1 second of work. This means it will get 1 task of any size of that project. Not sure, but it could end you up with too much work in buffer, after which work fetch immediately ceases till the conflict is solved]
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Nov 19, 2008 10:47:26 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Couldn't you have 24 different projects (WCG, SETI, ClimatePrediction, etc) each with a long job? That is what I was trying to express.
|
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
That's what I said. You can't with the noted initial exceptions. The Base limit TOTAL work buffered is 10 days times the number of cores in a device.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Sorry, I understand. Boinc provides a high level control over the max hours buffered across all projects. Thanks Sek.
|
||
|
|
|