Too Late To Validate

Message boards : Number crunching : Too Late To Validate
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · Next

AuthorMessage
Profile yoyo_rkn
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist

Send message
Joined: 22 Aug 11
Posts: 739
Credit: 17,612,101
RAC: 10
Germany
Message 118 - Posted: 9 Sep 2011, 15:32:58 UTC

That doesn't sound reasonable. The long wus should run ~2h. Do you restart your Boinc frequently? Which app version runs?
ID: 118 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ChertseyAl

Send message
Joined: 29 Aug 11
Posts: 7
Credit: 11,046
RAC: 0
United Kingdom
Message 125 - Posted: 12 Sep 2011, 7:45:02 UTC - in response to Message 118.  

That doesn't sound reasonable. The long wus should run ~2h. Do you restart your Boinc frequently? Which app version runs?


Actually, a lot of mine run from 8-11 hours, and I've lost a few that were too late to validate. I don't restart the machines, and they aren't used for anything else. Some WUs really do take much longer to run than you expect.

My other WUs are usually between an hour and an hour and a half or just a few minues.

I must admit that I don't understand the logic of timing out late WUs - You don't reissue them, so there is no danger of repeated work. They might well be useful in which case it is a waste of the crunchers time. Even if they have been factored elsewhere it hardly harms you or the project to grant the credit anyway :)

Al.
ID: 125 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile [AF>Le_Pommier] Aillas

Send message
Joined: 7 Sep 11
Posts: 12
Credit: 581,331
RAC: 0
France
Message 126 - Posted: 12 Sep 2011, 11:48:09 UTC - in response to Message 118.  

That doesn't sound reasonable. The long wus should run ~2h. Do you restart your Boinc frequently? Which app version runs?


I'm running boinc version 6.12.34 on Linux.
I restart boinc only when I restart my linux, that means, rarely.

Currently, WU are finished between 2 sec and 12000 sec (> 3 hours).

About my 2 very long WU (more than 18 hours, when I left friday night) I can't give you more information. They were computed using yafu 128.01 and they are not more available in my list of completed tasks. Maybe you can find them in my history if you still have access to it.

ID: 126 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ChertseyAl

Send message
Joined: 29 Aug 11
Posts: 7
Credit: 11,046
RAC: 0
United Kingdom
Message 127 - Posted: 12 Sep 2011, 13:34:41 UTC - in response to Message 125.  

Oh, looks like I got credited for timed out WUs - Nice one! :)

It's a pity I aborted one before I realised. Oh well.

Al.
ID: 127 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile yoyo_rkn
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist

Send message
Joined: 22 Aug 11
Posts: 739
Credit: 17,612,101
RAC: 10
Germany
Message 128 - Posted: 12 Sep 2011, 14:40:25 UTC

The workunits are not immediate deleted if they time out. Therefore there is an additional chance to get credits ;)
But anyway a runtime of 8-12 hours is much to long currently.
Can you give me the result- or wuID?
yoyo
ID: 128 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ChertseyAl

Send message
Joined: 29 Aug 11
Posts: 7
Credit: 11,046
RAC: 0
United Kingdom
Message 129 - Posted: 12 Sep 2011, 15:39:40 UTC - in response to Message 128.  

ID: 129 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile [AF>Le_Pommier] Aillas

Send message
Joined: 7 Sep 11
Posts: 12
Credit: 581,331
RAC: 0
France
Message 130 - Posted: 12 Sep 2011, 15:43:38 UTC

I have 2 WU yafu_C94_1315461657_286_0 & yafu_C94_1315461657_324_0 that compute for more than 18 hours. I think I won't be credited for those wu :(

Hope this help

I search in the event log and found these:

Thu 08 Sep 2011 08:50:13 AM CEST | yafu | Starting task yafu_C94_1315461657_324_0 using yafu version 12801
Thu 08 Sep 2011 08:50:30 AM CEST | yafu | Started download of yafu_C94_1315461657_286
Thu 08 Sep 2011 08:50:31 AM CEST | yafu | Finished download of yafu_C94_1315461657_286
Thu 08 Sep 2011 08:50:31 AM CEST | yafu | Starting task yafu_C94_1315461657_286_0 using yafu version 12801
Thu 08 Sep 2011 10:31:40 AM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Thu 08 Sep 2011 10:31:40 AM CEST | yafu | Restarting task yafu_C94_1315461657_286_0 using yafu version 12801
Thu 08 Sep 2011 12:32:30 PM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Thu 08 Sep 2011 12:32:30 PM CEST | yafu | Restarting task yafu_C94_1315461657_286_0 using yafu version 12801
Thu 08 Sep 2011 02:34:09 PM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Thu 08 Sep 2011 02:34:09 PM CEST | yafu | Restarting task yafu_C94_1315461657_286_0 using yafu version 12801
Thu 08 Sep 2011 04:34:48 PM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Thu 08 Sep 2011 04:34:48 PM CEST | yafu | Restarting task yafu_C94_1315461657_286_0 using yafu version 12801
Thu 08 Sep 2011 06:35:12 PM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Thu 08 Sep 2011 06:35:12 PM CEST | yafu | Restarting task yafu_C94_1315461657_286_0 using yafu version 12801
Thu 08 Sep 2011 08:37:03 PM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Thu 08 Sep 2011 08:37:03 PM CEST | yafu | Restarting task yafu_C94_1315461657_286_0 using yafu version 12801
Thu 08 Sep 2011 10:38:27 PM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Thu 08 Sep 2011 10:38:27 PM CEST | yafu | Restarting task yafu_C94_1315461657_286_0 using yafu version 12801
Fri 09 Sep 2011 12:39:21 AM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Fri 09 Sep 2011 12:39:21 AM CEST | yafu | Restarting task yafu_C94_1315461657_286_0 using yafu version 12801
Fri 09 Sep 2011 02:40:43 AM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Fri 09 Sep 2011 02:40:43 AM CEST | yafu | Restarting task yafu_C94_1315461657_286_0 using yafu version 12801
Fri 09 Sep 2011 04:41:48 AM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Fri 09 Sep 2011 04:41:48 AM CEST | yafu | Restarting task yafu_C94_1315461657_286_0 using yafu version 12801
Fri 09 Sep 2011 06:43:15 AM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Fri 09 Sep 2011 06:43:15 AM CEST | yafu | Restarting task yafu_C94_1315461657_286_0 using yafu version 12801
Fri 09 Sep 2011 08:44:09 AM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Fri 09 Sep 2011 08:44:09 AM CEST | yafu | Restarting task yafu_C94_1315461657_286_0 using yafu version 12801
Fri 09 Sep 2011 10:45:33 AM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Fri 09 Sep 2011 10:45:33 AM CEST | yafu | Restarting task yafu_C94_1315461657_286_0 using yafu version 12801
Fri 09 Sep 2011 12:47:16 PM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Fri 09 Sep 2011 12:47:16 PM CEST | yafu | Restarting task yafu_C94_1315461657_286_0 using yafu version 12801
Fri 09 Sep 2011 02:47:32 PM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Fri 09 Sep 2011 03:49:00 PM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Fri 09 Sep 2011 05:30:59 PM CEST | yafu | Sending scheduler request: To fetch work.
Fri 09 Sep 2011 05:30:59 PM CEST | yafu | Requesting new tasks for CPU
Fri 09 Sep 2011 05:31:01 PM CEST | yafu | Scheduler request completed: got 1 new tasks
Fri 09 Sep 2011 05:31:03 PM CEST | yafu | Started download of yafu-32k-linux64-12805
Fri 09 Sep 2011 05:31:03 PM CEST | yafu | Started download of yafu_C93_1315579582_318
Fri 09 Sep 2011 05:31:04 PM CEST | yafu | Finished download of yafu_C93_1315579582_318
Fri 09 Sep 2011 05:32:41 PM CEST | yafu | Finished download of yafu-32k-linux64-12805
Fri 09 Sep 2011 06:10:02 PM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Fri 09 Sep 2011 08:07:53 PM CEST | yafu | Restarting task yafu_C94_1315461657_324_0 using yafu version 12801
Fri 09 Sep 2011 08:29:50 PM CEST | yafu | Sending scheduler request: To fetch work.
Fri 09 Sep 2011 08:29:50 PM CEST | yafu | Requesting new tasks for CPU
Fri 09 Sep 2011 08:29:52 PM CEST | yafu | Scheduler request completed: got 1 new tasks
Fri 09 Sep 2011 08:29:52 PM CEST | yafu | Result yafu_C94_1315461657_324_0 is no longer usable
Fri 09 Sep 2011 08:29:52 PM CEST | yafu | Result yafu_C94_1315461657_286_0 is no longer usable
Fri 09 Sep 2011 08:29:53 PM CEST | yafu | Computation for task yafu_C94_1315461657_324_0 finished
Fri 09 Sep 2011 08:29:53 PM CEST | yafu | Computation for task yafu_C94_1315461657_286_0 finished


ID: 130 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile yoyo_rkn
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist

Send message
Joined: 22 Aug 11
Posts: 739
Credit: 17,612,101
RAC: 10
Germany
Message 131 - Posted: 12 Sep 2011, 17:09:18 UTC

@ChertseyAl: I do not see really a problem in your results. Maybe your Pentium 4 is just to old and has to less cache.

[AF>Le_Pommier] Aillas: You should avoid the restarts of yafu workunits, e.g. by activating "keep app in memory".

yoyo
ID: 131 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile bsquared

Send message
Joined: 5 Sep 11
Posts: 6
Credit: 113,086
RAC: 0
United States
Message 132 - Posted: 13 Sep 2011, 13:16:09 UTC - in response to Message 129.  

The longest running one was the last one that was a little over 12 hours.

Al.


This would not be unusual for a PIV factoring a C90+ by SIQS. Unfortunately PIV's are notoriously poor sievers, and yafu is not well optimized for them.
ID: 132 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
vaughan

Send message
Joined: 1 Sep 11
Posts: 9
Credit: 13,786,383
RAC: 751
Australia
Message 133 - Posted: 14 Sep 2011, 1:14:54 UTC
Last modified: 14 Sep 2011, 1:20:59 UTC

Dear Admin:

Your ~1 hour estimate is wrong.

I have several computers with YAFU tasks still running well after 1 hour. For example a C2D E6420 @ 2.133GHz has task Yafu_C97_1315917330_94_0 so far at 6 hours 20 minutes and continuing. Progress is stuck on 0.000% (I'm sure I posted about the Progress % not updating but the post seems to have disappeared).

Also another Windows 7 machine this time a Core 2 Quad 6600 G0 stepping @ 2.40GHz. It has a C97 task at over 4 hours and still running.

What has happened is that the tasks clog up the system so they time out. I checked the E6420 and changed the settings preference so that tasks remain in memory. On the Q6600 this was already set.

On the E6420 I reset the project last night as so many tasks were not going to complete. The BOINC estimate was over 4 hours. After the rest stupid thing set the YAFU estimate a 17 minutes - crikey. What happened to the "I increased the estimated time per task" adjustment you made?

Better to err on the side of caution like for example CAS@home where the initial tasks are estimated at several hundred hours. This allows BOINC to self correct the run time and you don't have a swag full of tasks cached.

I am no mathematician but I don't understand the urgency of this sieving project in getting tasks from a factordb crunched in under 10 hours. If the time-frame is so short why not run the project from GPUs like the PrimeGrid Project does for its PPS sieving sub-project? If it remains as a CPU project then I concur with the other posters and strongly urge the project administrator to increase the time limit to several days.

my 2c
ID: 133 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile yoyo_rkn
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist

Send message
Joined: 22 Aug 11
Posts: 739
Credit: 17,612,101
RAC: 10
Germany
Message 134 - Posted: 14 Sep 2011, 5:12:57 UTC - in response to Message 133.  

Hello,
some clarifications.

In Boinc on the server you don't specify the time in hours a tak will need. You specify it in floating point operations a task will need. Currently the tasks are estimated with so many floating point operations, which my i5 laptop will consume in 2h. The task are not fully predictable how long they will need. Much of them are running only very short. Boinc client also stores how long tasks need on your computer and how is this related to the estimation. If the task take usually double of the time or half the time on your computer Boinc uses this to recalculate the estimated time.

Since we going to run longer numbers now, I increased also the deadline to 24h for new tasks.

Regarding not meeting the deadline, I think this isn't really a problem. If Boinc starts a task and sees, that this task is already behind it's deadline, than Boinc doesn't start this task and aborts it. So there is no computing time spend for which you do not get credits.
For tasks which are started before it's deadline but finishing after it's deadline there is a high chance, that you also get credits for them. The workunits are not immediate deleted by the server. They are keept for another 10h So if you report the task in this timeframe you get also credits and the project profits from your result.

Regarding GPU I think you miss the point that not every project, every problem can be implemented on GPU and even if it could it doesn't mean that it is faster. Currently I do not know a source, which is doing factoring in the number range we are doing.

yoyo
ID: 134 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
reklov

Send message
Joined: 2 Sep 11
Posts: 4
Credit: 258,685
RAC: 0
Germany
Message 149 - Posted: 18 Sep 2011, 14:14:51 UTC - in response to Message 134.  

At the moment, I have on an i5 (Linux 64bit, only the 2 real cores in use) tasks running with more than 8 resp 5 hours - other yafu tasks have been suspended from BOINC after 3 resp 5 hours runtime. They are all from chunk C98_1316296219_*. I aborted now a lot of wu's which haven't started yet - but the estimated runtime seems to be complete nonsense for these tasks. Maybe you should adjust the runtime estimation depending on the Cxx classification?
ID: 149 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
reklov

Send message
Joined: 2 Sep 11
Posts: 4
Credit: 258,685
RAC: 0
Germany
Message 153 - Posted: 18 Sep 2011, 15:16:48 UTC - in response to Message 149.  

Now 3rd wu finished today. Runtime were 19,482.51s, 20,599.86s and 33,248.96s (=9h14m09.36s; three further wu's running or suspended.
I don't want to think about the runtime on my old Pentium M :(
ID: 153 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jose [Team Musketeers]

Send message
Joined: 8 Sep 11
Posts: 7
Credit: 35,578
RAC: 0
Argentina
Message 156 - Posted: 20 Sep 2011, 20:37:54 UTC

I'm sorry to disturb, but on two computers I have this problem:

yafu 20/09/2011 01:07:45 p.m. Message from server: Tasks won't finish in time: BOINC runs 98.0% of the time; computation is enabled 100.0% of that


The 30% or 50% of WU are 100% but never finish.
This happens in W7 and XP X86. DualCore 2.95ghz and / 2g ram

What is the problem?
ID: 156 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile yoyo_rkn
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist

Send message
Joined: 22 Aug 11
Posts: 739
Credit: 17,612,101
RAC: 10
Germany
Message 158 - Posted: 20 Sep 2011, 21:26:09 UTC

Is it better now?
yoyo
ID: 158 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile RealSanta

Send message
Joined: 21 Sep 11
Posts: 1
Credit: 51,627
RAC: 0
Germany
Message 159 - Posted: 22 Sep 2011, 6:04:28 UTC
Last modified: 22 Sep 2011, 6:04:53 UTC

Hi yoyo...

here on my i7 the wu also runs til yet 7:48 h and counting (no remaining time is shown)... at the laptop after around 4h the wus finished.

best RS
ID: 159 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jose [Team Musketeers]

Send message
Joined: 8 Sep 11
Posts: 7
Credit: 35,578
RAC: 0
Argentina
Message 162 - Posted: 22 Sep 2011, 11:43:21 UTC

Hi yoyo, there are still problems: Yafu be closed: error
If the workunit ends before 4h : ok
If it does not end: error.


- System

- Provider
[ Name] Application Error

- EventID 1000
[ Qualifiers] 0

Level 2
Task 100
Keywords 0x80000000000000

- TimeCreated
[ SystemTime] 2011-09-22T05:02:13.000000000Z

EventRecordID 1113

Channel Application
Computer Jose-Server


- EventData

yafu
0.0.0.0
4e70b156
ntdll.dll
6.1.7600.16385
4a5bdadb
c0000374
000c283b
e24
01cc78c87fa791dd
C:\ProgramData\BOINC\slots\4\yafu
C:\Windows\SYSTEM32\ntdll.dll
08992056-e4d8-11e0-866c-485b3950f0e4
ID: 162 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jose [Team Musketeers]

Send message
Joined: 8 Sep 11
Posts: 7
Credit: 35,578
RAC: 0
Argentina
Message 171 - Posted: 24 Sep 2011, 14:13:49 UTC

Hi YoYo!! congratulations!!
I believe that with the 128.05 version have been solved my problems.
(at least for me) ;)
ID: 171 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Matthias Lehmkuhl

Send message
Joined: 7 Oct 11
Posts: 34
Credit: 2,445,624
RAC: 76
Germany
Message 204 - Posted: 13 Oct 2011, 11:46:58 UTC

Hi Yoyo,
Result shows sometimes 100% Progress when yafu is finished and gnfs-lasieve should run. Boinc ignores the deadline for e.g. "high priority mode", when the result is at 100% progress and not finished yet.
http://yafu.dyndns.org/yafu/result.php?resultid=349690
Project: yafu
Application: 128.06 YAFU
Result name: yafu_C101_1318234240_115_0
elapsed time: 13:03:35 (04:32:33)
Progress: 100,000
Remaining time: -
Deadline: 12.10.2011 13:44:34

Is it possible to reach 100% progress only when the result is ready for uploading?
Matthias
ID: 204 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile yoyo_rkn
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist

Send message
Joined: 22 Aug 11
Posts: 739
Credit: 17,612,101
RAC: 10
Germany
Message 205 - Posted: 13 Oct 2011, 16:21:47 UTC - in response to Message 204.  

Hi,

the wu shows the whole time 0% and only at the end when the wu has finished it shows 100%, at least in Boinc manger. Seems that you are using something different, e.g. boincview.

yoyo
ID: 205 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Previous · 1 · 2 · 3 · 4 · Next

Message boards : Number crunching : Too Late To Validate




Datenschutz / Privacy Copyright © 2011-2024 Rechenkraft.net e.V. & yoyo