Earliest Deadline First

Message boards : Number crunching : Earliest Deadline First
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · 4 . . . 5 · Next

AuthorMessage
W-K 666 Project Donor
Volunteer tester

Send message
Joined: 18 May 99
Posts: 18996
Credit: 40,757,560
RAC: 67
United Kingdom
Message 218612 - Posted: 20 Dec 2005, 12:06:33 UTC
Last modified: 20 Dec 2005, 12:10:37 UTC

My computer has just entered EDF for no apparent reason that I can work out, to crunch Einstein.
DTG 20/12/2005 11:52 UTC, Last reboot 23hr:30min ago, for NIC driver update.
Resumed Einstein unit on completion of Seti unit.

Pentium M 1.86GHz

Resourse Share
Einstein 33.33%
Seti 58.33%
SetiB 8.33%

STD
Einstein -1568
Seti +2387
SetiB -817

LTD
Einstein -1173
Seti +4500
SetiB -3327

Estimated time to crunch
Einstein 8hr:03m
Seti 0hr:37m
SetiB (234hr, silly figure) actual 29hr:04m

Units held on Computer
Einstein 9, one at 97.5% 24m to complete, Running. Deadline for last one 02/01/2006 12:41:50
Seti 85, non preempted, Deadline for last one 03/01/2006
SetiB 1, preempted 66.63% in 18hr:30min, Deadline for last one 10/04/2007 (? see TMR post.)

Hours allocated until Einstein Deadline @33.33% resourse share 13 * 24 / 3 = 104, estimated time for Einstein units to crunch 8 * 8.05 = 64.4 + 0.4 = 64.08. therefore there is an excess of nearly fourty hours, so why the EDF.

Even if we take off one day to meet the 24hour rule there is still an exess of hours required to meet the deadline without entering EDF

P.S. Connest to network for Einstein is 0.1 days (actually don't understand why in downloaded 8 units)
ID: 218612 · Report as offensive
Profile MJKelleher
Volunteer tester
Avatar

Send message
Joined: 1 Jul 99
Posts: 2048
Credit: 1,575,401
RAC: 0
United States
Message 218627 - Posted: 20 Dec 2005, 12:42:42 UTC

I've seen this happen on my computer, doing different projects. Whenever I've seen it, at the next reschedule time it goes back to round-robin. I'd just observe it for a couple of hours and see what it does.

MJ

ID: 218627 · Report as offensive
Astro
Volunteer tester
Avatar

Send message
Joined: 16 Apr 02
Posts: 8026
Credit: 600,015
RAC: 0
Message 218630 - Posted: 20 Dec 2005, 12:54:04 UTC

rerun benchmarks
ID: 218630 · Report as offensive
Profile Gary Roberts
Volunteer tester

Send message
Joined: 31 Oct 99
Posts: 95
Credit: 2,301,228
RAC: 0
Australia
Message 218634 - Posted: 20 Dec 2005, 13:13:22 UTC - in response to Message 218612.  

My computer has just entered EDF for no apparent reason that I can work out, to crunch Einstein.
....
P.S. Connest to network for Einstein is 0.1 days (actually don't understand why in downloaded 8 units)


You can't have a different "connect" setting for the different projects, unless you set up different venues. Actually, I'm not even sure the one computer can have different venues for the different projects?? So, EAH will not be 0.1, it will be whatever Seti is - around 2 days or so at a guess. BOINC will take into account the expected time for all work so if your beta unit is estimated at 234 hours (even though you say it will only take 29) BOINC wont know that. It's not surprising you are in EDF.

But what does it matter? BOINC will crunch a few EAH and then go back to round-robin and your resource share will be honoured eventually.

Because your Seti work is so highly optimised, your claims are very low - around 5 or 6. Very often you get pulled up by others but quite a few times you help drag others down. You should delay submitting seti results as long as possible to give other higher claimers a chance to validate first. So, being in EDF with EAH running for the next day or two is actually the best possible outcome for your score :).


Cheers,
Gary.
ID: 218634 · Report as offensive
Profile Landroval

Send message
Joined: 7 Oct 01
Posts: 188
Credit: 2,098,881
RAC: 1
United States
Message 218637 - Posted: 20 Dec 2005, 13:36:37 UTC - in response to Message 218634.  

You can't have a different "connect" setting for the different projects, unless you set up different venues. Actually, I'm not even sure the one computer can have different venues for the different projects?? So, EAH will not be 0.1, it will be whatever Seti is - around 2 days or so at a guess.

I don't know about the different-venues-for-different-projects; I've not tried it. In general, though, the BOINC client will use whichever set of preferences has been updated most recently.

FWIW, I agree that in the long run, the scheduler will take care of it. A project gets a few hours extra time today and gives up a few hours tomorrow, big deal. And if you have a host that's consistently claiming lower credit, running a longer work queue on that host may be a good idea.

If you think education is expensive, try ignorance.
ID: 218637 · Report as offensive
W-K 666 Project Donor
Volunteer tester

Send message
Joined: 18 May 99
Posts: 18996
Credit: 40,757,560
RAC: 67
United Kingdom
Message 218661 - Posted: 20 Dec 2005, 14:40:54 UTC
Last modified: 20 Dec 2005, 14:43:20 UTC

Even if it's using the home settings why did it download 8 Einstein units, 33% of 4 * 24 = 32hrs and at 8hrs/unit it should have only downloaded 4 Einstein units.

I angree it's not critical here but if it is a fault then it could be, and hosts could end up not meeting deadlines, so the suggestion to set my connect time higher is not an option.
ID: 218661 · Report as offensive
Profile Tern
Volunteer tester
Avatar

Send message
Joined: 4 Dec 03
Posts: 1122
Credit: 13,376,822
RAC: 44
United States
Message 218680 - Posted: 20 Dec 2005, 15:27:43 UTC - in response to Message 218661.  

Even if it's using the home settings why did it download 8 Einstein units, 33% of 4 * 24 = 32hrs and at 8hrs/unit it should have only downloaded 4 Einstein units.


Resource share is not used in the download calculation. So if you say 4 days, you get 4 days, not 33% of 4 days. Work already _on_ the host _is_ used in the calculation, so it actually gives you "4 days less some complicated calculation using the other work regardless of project". (Love that technical accuracy there...) But then the simulation to determine if EDF is needed _does_ use the resource shares...

It's a known issue, it's "on the list".
ID: 218680 · Report as offensive
W-K 666 Project Donor
Volunteer tester

Send message
Joined: 18 May 99
Posts: 18996
Credit: 40,757,560
RAC: 67
United Kingdom
Message 223691 - Posted: 31 Dec 2005, 9:14:25 UTC
Last modified: 31 Dec 2005, 9:15:13 UTC

Now I realy am confused:

31/12/2005 06:59:01|SETI@home|Starting result 11fe05aa.11460.17538.242328.1.5_1 using setiathome version 418
31/12/2005 06:59:03|SETI@home|Started upload of 03no03aa.19167.5360.1040906.1.197_0_0
31/12/2005 06:59:14|SETI@home|Finished upload of 03no03aa.19167.5360.1040906.1.197_0_0
31/12/2005 06:59:14|SETI@home|Throughput 1474 bytes/sec
31/12/2005 07:31:21||Using earliest-deadline-first scheduling because computer is overcommitted.
31/12/2005 07:36:17||request_reschedule_cpus: process exited
31/12/2005 07:36:17|SETI@home|Computation for result 11fe05aa.11460.17538.242328.1.5_1 finished
31/12/2005 07:36:17||Resuming round-robin CPU scheduling.

31/12/2005 07:36:17|SETI@home|Starting result 11fe05aa.11460.17585.959636.1.246_2 using setiathome version 418
31/12/2005 07:36:19|SETI@home|Started upload of 11fe05aa.11460.17538.242328.1.5_1_0
31/12/2005 07:36:27|SETI@home|Finished upload of 11fe05aa.11460.17538.242328.1.5_1_0
31/12/2005 07:36:27|SETI@home|Throughput 4035 bytes/sec
31/12/2005 07:45:21||Using earliest-deadline-first scheduling because computer is overcommitted.

Computer is attached to three projects, Seti, Einstein, and SetiB.
Seti is only one with units at moment,
Einstein because of original post is in heavy LTD -34847.582730
SetiB is set to 'no new work' as I expect it to become mainstream soon and my resource share is low so don't think I can do another unit in the time frame.

So why the EDF?

ID: 223691 · Report as offensive
Profile Jord
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 15184
Credit: 4,362,181
RAC: 3
Netherlands
Message 223711 - Posted: 31 Dec 2005, 10:09:30 UTC
Last modified: 31 Dec 2005, 10:12:15 UTC

What is your Connect to set to and how many work units do you actually have?

Earliest Deadline First
Using earliest-deadline-first scheduling because computer is overcommitted.

Uhm, if it is this computer, the site claims you have 133 units in cache??
ID: 223711 · Report as offensive
Profile MikeSW17
Volunteer tester

Send message
Joined: 3 Apr 99
Posts: 1603
Credit: 2,700,523
RAC: 0
United Kingdom
Message 223712 - Posted: 31 Dec 2005, 10:18:09 UTC - in response to Message 223691.  

Now I realy am confused:

<snip>

So why the EDF?


I don't know.
But the good news is I don't care, i don't need to, it doesn't matter.

Over the past months BOINC has amply demonstrated that does work according to my shares and does it on time.
I consider all these messages spurious as they have no effect on my systems doing the work and returning it on time.

The only time I expect anyone needs to consider deadlines is when they have a large cache (connect days) and have to change a machine use from say 24/7 to 8/5. That could give BOINC an impossible headache, and they need run-down the cache well before the change.


ID: 223712 · Report as offensive
W-K 666 Project Donor
Volunteer tester

Send message
Joined: 18 May 99
Posts: 18996
Credit: 40,757,560
RAC: 67
United Kingdom
Message 223718 - Posted: 31 Dec 2005, 11:43:33 UTC - in response to Message 223711.  
Last modified: 31 Dec 2005, 11:47:45 UTC

What is your Connect to set to and how many work units do you actually have?

Earliest Deadline First
Using earliest-deadline-first scheduling because computer is overcommitted.

Uhm, if it is this computer, the site claims you have 133 units in cache??

Yes its that computer and if you look at crunch time max is 40mins, so 133 * 40 / 60 = 88.667 hrs, connect to is 4.33 days = 104 hrs computer is on 24/7 and connected by, always on, broadband.

The unit being crunched, the closest to deadline is not due until 09/01/2006 03:39:58. It will connect twice before that deadline, even if it doesn't request more work.

The last unit in cache is due 14/01/2006 11:23:37. The one that was last in cache before post earlier today is due 14/01/2006 04:52:22. So not a problem there as it should be delivered, depending on Einstein and its LTD, by 04/01/2005 04:00 approx. about 10 days before deadline. So will not be late, will not break the 24hrs before rule.

Resourse share Seti:Einstein:SetiB is 98:56:14, added equals 168, hrs in week.

I've read all the info and still cannot see any criteria that it infringes to force EDF. Unless there is something not stated.


edit] And if you're wondering last change to preferences was before closure of Classic notices in November.
ID: 223718 · Report as offensive
Profile Jord
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 15184
Credit: 4,362,181
RAC: 3
Netherlands
Message 223719 - Posted: 31 Dec 2005, 11:47:11 UTC - in response to Message 223718.  

When you check the guestimate times of all work units in the Work tab (the time to completion number) and add those up, does it then get outside the 4 day connect to?

ID: 223719 · Report as offensive
W-K 666 Project Donor
Volunteer tester

Send message
Joined: 18 May 99
Posts: 18996
Credit: 40,757,560
RAC: 67
United Kingdom
Message 223720 - Posted: 31 Dec 2005, 11:48:15 UTC - in response to Message 223719.  
Last modified: 31 Dec 2005, 11:49:34 UTC

When you check the guestimate times of all work units in the Work tab (the time to completion number) and add those up, does it then get outside the 4 day connect to?

No. time in to completion col is 38:25
ID: 223720 · Report as offensive
Profile Keck_Komputers
Volunteer tester
Avatar

Send message
Joined: 4 Jul 99
Posts: 1575
Credit: 4,152,111
RAC: 1
United States
Message 223767 - Posted: 31 Dec 2005, 13:59:23 UTC

What are your on, active and CPU effeciency factors? All of these are used when figuring how much wall time it will take to get the CPU time listed in the work tab.
BOINC WIKI

BOINCing since 2002/12/8
ID: 223767 · Report as offensive
W-K 666 Project Donor
Volunteer tester

Send message
Joined: 18 May 99
Posts: 18996
Credit: 40,757,560
RAC: 67
United Kingdom
Message 223781 - Posted: 31 Dec 2005, 14:18:31 UTC - in response to Message 223767.  

What are your on, active and CPU effeciency factors? All of these are used when figuring how much wall time it will take to get the CPU time listed in the work tab.


% of time BOINC client is running 98.7839 %
While BOINC running, % of time work is allowed 98.7481 %
Average CPU efficiency 0.983544
Result duration correction factor 0.136217
ID: 223781 · Report as offensive
Jack Gulley

Send message
Joined: 4 Mar 03
Posts: 423
Credit: 526,566
RAC: 0
United States
Message 224000 - Posted: 1 Jan 2006, 1:49:50 UTC

One possible clue as to what might have happened. Look at the two results times just before the one he sent and got the warning message. Those had times that were 100 sec and 200 sec longer than normal. This would have caused his average result time to go up slightly and a recalculation with that number might have shown him over committed. Then that number would have gone back down with that result time and the ones following. Should not have caused a problem with those numbers, but something to look into.
ID: 224000 · Report as offensive
Profile Keck_Komputers
Volunteer tester
Avatar

Send message
Joined: 4 Jul 99
Posts: 1575
Credit: 4,152,111
RAC: 1
United States
Message 224113 - Posted: 1 Jan 2006, 11:29:23 UTC - in response to Message 223781.  

What are your on, active and CPU effeciency factors? All of these are used when figuring how much wall time it will take to get the CPU time listed in the work tab.


% of time BOINC client is running 98.7839 %
While BOINC running, % of time work is allowed 98.7481 %
Average CPU efficiency 0.983544
Result duration correction factor 0.136217


Your DCF seems very low. It should go up quickly if that is the problem though.

Your connect time is 4 days if I am reading the previous posts correctly. That means that any time a deadline is less than 8 days away EDF will be triggered.
BOINC WIKI

BOINCing since 2002/12/8
ID: 224113 · Report as offensive
Profile MikeSW17
Volunteer tester

Send message
Joined: 3 Apr 99
Posts: 1603
Credit: 2,700,523
RAC: 0
United Kingdom
Message 224118 - Posted: 1 Jan 2006, 11:58:16 UTC

Yes, the DCF does seem a bit low. My Pemtium M 2.0GHz has a DCF of .16
Assuming we are talking about your 1.86GHz that is.
If it's your other system it is far too low for a 1GHz system.

What is your actual WU times averaging?

ID: 224118 · Report as offensive
W-K 666 Project Donor
Volunteer tester

Send message
Joined: 18 May 99
Posts: 18996
Credit: 40,757,560
RAC: 67
United Kingdom
Message 224120 - Posted: 1 Jan 2006, 12:05:19 UTC

Average time to crunch a unit is about 35 minutes
ID: 224120 · Report as offensive
W-K 666 Project Donor
Volunteer tester

Send message
Joined: 18 May 99
Posts: 18996
Credit: 40,757,560
RAC: 67
United Kingdom
Message 224121 - Posted: 1 Jan 2006, 12:07:44 UTC

I assume your Pent M is a laptop, mine isn't it's on AOpen mobo with 915 chipset and 533 DDR2 memory.
ID: 224121 · Report as offensive
1 · 2 · 3 · 4 . . . 5 · Next

Message boards : Number crunching : Earliest Deadline First


 
©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.