Message boards : Number crunching : Gone in a flash!
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
Daxa

Send message
Joined: 29 Dec 06
Posts: 100
Credit: 184,937
RAC: 0
Message 16738 - Posted: 27 Apr 2007, 8:29:20 UTC
Last modified: 27 Apr 2007, 8:29:43 UTC

Did anyone clock how fast those 13500 work units were snapped up? I got ONE, and by the time it had run for 15 minutes there was no more work (of course, my WU may not have started right away because I crunch for other projects.)

Well anyway, hooray for the drizzle after the drought!



_______

"Three quarks for Muster Mark!"
. . . . . . . - James Joyce, Finnegans Wake . . . .

ID: 16738 · Report as offensive     Reply Quote
Profile caspr
Avatar

Send message
Joined: 26 Apr 06
Posts: 89
Credit: 309,235
RAC: 0
Message 16743 - Posted: 27 Apr 2007, 13:29:32 UTC

Ok, that explaines why I haven't been able to get any wu's on the last two runs!Some days you'er the windshield, some days you'er the bug!
A clear conscience is usually the sign of a bad memory


ID: 16743 · Report as offensive     Reply Quote
Bob Guy

Send message
Joined: 28 Sep 05
Posts: 21
Credit: 11,715
RAC: 0
Message 16748 - Posted: 28 Apr 2007, 12:05:54 UTC

I got 19 from this batch, and I thought that wasn't alot. They were all gone within an hour of when my computer began downloading the ones I got. I see in the results history that I've gotten a couple of others this month that I was completely unaware of.
ID: 16748 · Report as offensive     Reply Quote
Ravens

Send message
Joined: 14 Dec 06
Posts: 29
Credit: 128,225
RAC: 0
Message 16751 - Posted: 28 Apr 2007, 16:59:22 UTC - in response to Message 16748.  

I got 19 from this batch, and I thought that wasn't alot. They were all gone within an hour of when my computer began downloading the ones I got. I see in the results history that I've gotten a couple of others this month that I was completely unaware of.


You're pretty brave admitting you got 19. I got two- is there a particular setting to help this? My network usage/work cache is set to 1 day - is 5 better to encourage more WU's?
ID: 16751 · Report as offensive     Reply Quote
Aaron Finney

Send message
Joined: 14 Jul 05
Posts: 60
Credit: 140,661
RAC: 0
Message 16752 - Posted: 28 Apr 2007, 18:16:39 UTC - in response to Message 16738.  

Did anyone clock how fast those 13500 work units were snapped up? I got ONE, and by the time it had run for 15 minutes there was no more work (of course, my WU may not have started right away because I crunch for other projects.)

Well anyway, hooray for the drizzle after the drought!




I wonder why I was given 13,410 of them?

Oh well... I have work >:P
ID: 16752 · Report as offensive     Reply Quote
Profile Nightbird

Send message
Joined: 13 Jul 05
Posts: 55
Credit: 41,230
RAC: 0
Message 16753 - Posted: 28 Apr 2007, 20:18:35 UTC

I disabled work requests on this project.
I don't like the idea that somebody, sometimes throws me peanuts.

Do you want to get banned for 31 years, your account and credits deleted at a Boinc project ? Predictor@home is your best choice.
ID: 16753 · Report as offensive     Reply Quote
Bob Guy

Send message
Joined: 28 Sep 05
Posts: 21
Credit: 11,715
RAC: 0
Message 16758 - Posted: 29 Apr 2007, 4:46:04 UTC - in response to Message 16751.  

I got two- is there a particular setting to help this? My network usage/work cache is set to 1 day - is 5 better to encourage more WU's?

My cache setting is at 0.5 days, if that does anything good it just means that my computer might check more often for work. I was just lucky and my computer happened to check at the time when there were some WUs available. I think it also helps if you have a fast DSL or cable connection so the cache fills before the WUs are gone, they really do get gone fast.

I happened to be at my computer when this happened and I though about changing my cache settings to a much larger value, like 5 days, just to fill the cache, but I think that's just rude to the other people here. No, I was just lucky to get the few WUs I got.
ID: 16758 · Report as offensive     Reply Quote
Andreas

Send message
Joined: 2 Aug 05
Posts: 33
Credit: 2,328,412
RAC: 16
Message 16760 - Posted: 29 Apr 2007, 8:13:34 UTC - in response to Message 16753.  

I disabled work requests on this project.
I don't like the idea that somebody, sometimes throws me peanuts.


What's wrong with peanuts? :-)
ID: 16760 · Report as offensive     Reply Quote
KAMasud

Send message
Joined: 7 Oct 06
Posts: 114
Credit: 23,192
RAC: 0
Message 16764 - Posted: 29 Apr 2007, 12:50:56 UTC


:-) Nothing wrong with peanuts, but i for one have gotten sick and tired of hanging around for those peanuts;-) so missed out on the peanuts:-) LoL.
Regards
Masud.
ID: 16764 · Report as offensive     Reply Quote
Daxa

Send message
Joined: 29 Dec 06
Posts: 100
Credit: 184,937
RAC: 0
Message 16769 - Posted: 30 Apr 2007, 2:39:29 UTC - in response to Message 16751.  
Last modified: 30 Apr 2007, 2:47:38 UTC

...I see in the results history that I've gotten a couple of others this month that I was completely unaware of.
@Bob Guy: Same here. This happens a lot ESPECIALLY if your machine has a fast processor: Gone in a flash, Downloaded in a flash, Crunched in a flash, Reported in a flash.



...I got two- is there a particular setting to help this? My network usage/work cache is set to 1 day - is 5 better to encourage more WU's?
@Ravens: I've experimented with different cache settings (from 0.1 to 7) and my labors have revealed that 3 days is pretty good. Less than 3 will usually cause your machine to pick up very few (or ONE) work units even if the run is huge [ >50000 ]. A cache number more than 3 will result in often getting NOTHING from small runs [ <20000 ] ...but I encourage others to tweak my number and post if they find a better Magic Constant!

P.S. I love peanuts, and I don't mind waiting around for them.


_______

"Three quarks for Muster Mark!"
. . . . . . . - James Joyce, Finnegans Wake . . . .

ID: 16769 · Report as offensive     Reply Quote
corkyf16

Send message
Joined: 30 Apr 07
Posts: 3
Credit: 681,339
RAC: 14
Message 16770 - Posted: 30 Apr 2007, 2:53:13 UTC - in response to Message 16769.  

Pardon my ignorance, but where does one 'adjust the cache settings'?
Thanks,
Joe

...I see in the results history that I've gotten a couple of others this month that I was completely unaware of.
@Bob Guy: Same here. This happens a lot ESPECIALLY if your machine has a fast processor: Gone in a flash, Downloaded in a flash, Crunched in a flash, Reported in a flash.



...I got two- is there a particular setting to help this? My network usage/work cache is set to 1 day - is 5 better to encourage more WU's?
@Ravens: I've experimented with different cache settings (from 0.1 to 7) and my labors have revealed that 3 days is pretty good. Less than 3 will usually cause your machine to pick up very few (or ONE) work units even if the run is huge [ >50000 ]. A cache number more than 3 will result in often getting NOTHING from small runs [ <20000 ] ...but I encourage others to tweak my number and post if they find a better Magic Constant!

P.S. I love peanuts, and I don't mind waiting around for them.


ID: 16770 · Report as offensive     Reply Quote
Bob Guy

Send message
Joined: 28 Sep 05
Posts: 21
Credit: 11,715
RAC: 0
Message 16771 - Posted: 30 Apr 2007, 3:42:26 UTC - in response to Message 16770.  
Last modified: 30 Apr 2007, 3:46:02 UTC

Pardon my ignorance, but where does one 'adjust the cache settings'?
Thanks, Joe

Go to your account, General preferences, Network usage.
Adjust your 'Connect to' time to whatever you think is appropriate. For this project my opinion is that 1 day or less is most appropriate. And I like peanuts! I'll wait awhile to get some.

As noted before large 'connect to' times can cause you to completely miss small distributions of work, but if you happen to ask for work at just the right time a large 'connect to' time will get you more WUs downloaded.

You cannot actually adjust your cache size as in number of WUs downloaded but you can set a 'connect to' time large enough so that more WUs will be downloaded when work is available. There are also ways to fake-out the server (to get more work than your settings would ordinarily allow) but I consider that to be unfair to the other people here who would also like to get some work.

Also note that setting your cache here is a global setting and will be applied in any other Boinc project that you are active in.
ID: 16771 · Report as offensive     Reply Quote
Ariel
Avatar

Send message
Joined: 7 Mar 07
Posts: 59
Credit: 7,906
RAC: 0
Message 16777 - Posted: 30 Apr 2007, 8:33:07 UTC - in response to Message 16769.  
Last modified: 30 Apr 2007, 8:45:43 UTC

...I've experimented with different cache settings (from 0.1 to 7) and my labors have revealed that 3 days is pretty good. Less than 3 will usually cause your machine to pick up very few (or ONE) work units even if the run is huge [ >50000 ]. A cache number more than 3 will result in often getting NOTHING from small runs [ <20000 ] ...but I encourage others to tweak my number and post if they find a better Magic Constant!

Using 4 or 5 will work if you do the following:


1. Give lhc@home at least 75% resource share on any given machine.

2. Don't crunch for too many projects on any one machine, but make sure you crunch for at least one other project on each machine - Climate Prediction, Climate Attribution, SETI, Einstein, Rosetta, and the like are good candidates.

3. Keep an eye on the "other" projects running alongside lhc; they may have greedy tendencies that will block out lhc (even with a 0.01% resource share.. yes, I've seen it happen.)

4. Run your machine to death. You will get more lhc WUs if your machine has been crunching like crazy for other project(s) during lhc droughts.


That's my word.



Ariel: Certified "Too Cute for LHC" Cruncher!


. . . . . . . . . . . . -- Consider the lilies.
ID: 16777 · Report as offensive     Reply Quote
Profile Jim Baize
Avatar

Send message
Joined: 17 Sep 04
Posts: 103
Credit: 38,543
RAC: 0
Message 16783 - Posted: 30 Apr 2007, 11:29:18 UTC - in response to Message 16771.  

Pardon my ignorance, but where does one 'adjust the cache settings'?
Thanks, Joe

Go to your account, General preferences, Network usage.
Adjust your 'Connect to' time to whatever you think is appropriate. For this project my opinion is that 1 day or less is most appropriate. And I like peanuts! I'll wait awhile to get some.



Just to point out a small discrepancy, the "Connect to" time is not a "cache" setting. Although it does affect how much work is stored locally, it is not a "cache" setting. This is a very common misunderstanding.

Jim
ID: 16783 · Report as offensive     Reply Quote
Daxa

Send message
Joined: 29 Dec 06
Posts: 100
Credit: 184,937
RAC: 0
Message 16787 - Posted: 30 Apr 2007, 14:29:10 UTC - in response to Message 16783.  
Last modified: 30 Apr 2007, 14:33:18 UTC

...Although it does affect how much work is stored locally, it is not a "cache" setting. This is a very common misunderstanding.

And I suppose the amount of work stored locally is not a work cache. If that is so, then the Ulysses S. Grant in my wallet isn't cash either. That would explain why the liquor store down the street won't sell me that bottle of Patron Silver. Give me a break, Jim. It all comes down to cache, whether it bites you in the butt or not.




_______

"Three quarks for Muster Mark!"
. . . . . . . - James Joyce, Finnegans Wake . . . .

ID: 16787 · Report as offensive     Reply Quote
Profile Jim Baize
Avatar

Send message
Joined: 17 Sep 04
Posts: 103
Credit: 38,543
RAC: 0
Message 16788 - Posted: 30 Apr 2007, 14:35:56 UTC - in response to Message 16787.  

...Although it does affect how much work is stored locally, it is not a "cache" setting. This is a very common misunderstanding.

And I suppose the amount of work stored locally is not a work cache. If that is so, then the Ulysses S. Grant in my wallet isn't cash either. That would explain why the liquor store down the street won't sell me that bottle of Patron Silver. Give me a break, Jim. It all comes down to cache, whether it bites you in the butt or not.






Yes, the work stored locally is a cache, but the "connect interval" is just that, a connect interval, NOT a cache setting.
ID: 16788 · Report as offensive     Reply Quote
Daxa

Send message
Joined: 29 Dec 06
Posts: 100
Credit: 184,937
RAC: 0
Message 16790 - Posted: 30 Apr 2007, 14:51:18 UTC - in response to Message 16788.  
Last modified: 30 Apr 2007, 15:03:11 UTC

Yes, the work stored locally is a cache, but the "connect interval" is just that, a connect interval, NOT a cache setting.


So... a horse isn't a horse if it's a vial of stud spew that's worth $150,000 on the black market. Is that right? If there's any doubt, ask anyone who frequents the local pony track. That goo is worth some serious cash; don't try to tell the bookies it's not a horse!



_______

"Three quarks for Muster Mark!"
. . . . . . . - James Joyce, Finnegans Wake . . . .

ID: 16790 · Report as offensive     Reply Quote
Profile FalconFly
Avatar

Send message
Joined: 2 Sep 04
Posts: 121
Credit: 592,214
RAC: 0
Message 16792 - Posted: 30 Apr 2007, 15:43:18 UTC - in response to Message 16790.  
Last modified: 30 Apr 2007, 15:44:10 UTC

Indeed...

It is a common misunderstanding that "Connect to Network every X days" was not the Cache setting.

Apart from correction values applied to it (Up Fraction, Online Fraction, Running Fraction etc.) it is the basic - and only - local Cache setting to date.
Other supplemental Cache Options are being introduced with the current 5.9.x Alpha test series.

Why they erroneously named it so misleading (as the Client will make much more frequent contact when online, regardless of this alledged Network limitation setting) is beyond me.
Scientific Network : 45000 MHz - 77824 MB - 1970 GB
ID: 16792 · Report as offensive     Reply Quote
Profile Jim Baize
Avatar

Send message
Joined: 17 Sep 04
Posts: 103
Credit: 38,543
RAC: 0
Message 16793 - Posted: 30 Apr 2007, 17:36:01 UTC - in response to Message 16792.  

Indeed...

It is a common misunderstanding that "Connect to Network every X days" was not the Cache setting.

Apart from correction values applied to it (Up Fraction, Online Fraction, Running Fraction etc.) it is the basic - and only - local Cache setting to date.
Other supplemental Cache Options are being introduced with the current 5.9.x Alpha test series.

Why they erroneously named it so misleading (as the Client will make much more frequent contact when online, regardless of this alledged Network limitation setting) is beyond me.


If it was indeed a cache setting, the "connect interval" would be directly proportional to the cache setting. This is not always the case. In some instances it is directly proportional, while in others, making the "connect interval" too large will actually stop the accumulation of work on hand; therefore it is NOT a cache setting. Yes, it does affect the cache, but it is NOT a cache setting.
ID: 16793 · Report as offensive     Reply Quote
Profile FalconFly
Avatar

Send message
Joined: 2 Sep 04
Posts: 121
Credit: 592,214
RAC: 0
Message 16794 - Posted: 30 Apr 2007, 20:31:50 UTC - in response to Message 16793.  
Last modified: 30 Apr 2007, 20:42:25 UTC

*lol*

Okidok, so tell me where the Setting for the Local Cache is then ?
Because apparently, I've been running BOINC without knowing how to set it for more than 2 years ;)

The answer :
The setting is of course directly proportional to the Cache.

However, it is smart enough to detect Situations where the selected amount of Cached work will result in Problems (such as missing the deadline for short-deadline Projects). This is a safety function with override Priorities inherent to the Cacheing feature, but nonetheless it remains exactly that : Set Cache to x Days worth of work.

It's actually smart enough to correct for System downtimes, times where the System is up but not running BOINC and even CPU efficiency (where BOINC is running but only gets a fraction of the total CPU share).

Set your Cache to 1 day and after filling up the Cache, disconnect Network cables and use the System in the same way you did in the last 2 weeks. Magically, it will run dry in exactly the time you had set as Cache.

The effects of this calibration lead several folks to believe it was something else than Cacheing function, but it's plain not true.

Double your Setting and witness it downloading & Caching exactly double the Work... unless it determines that this would hit one of the Criteria mentioned above and its limiting-feature cuts in.

Exceptions were coded into BOINC for exceptional Projects, such as CPDN where the setting was rendered basically irrelevant and specific extensions had to be put in place to accomodate the extremely long runtimes.
Scientific Network : 45000 MHz - 77824 MB - 1970 GB
ID: 16794 · Report as offensive     Reply Quote
1 · 2 · Next

Message boards : Number crunching : Gone in a flash!


©2024 CERN