cmserch XXL (long) 1.0.2 0.31 save during work

Everything about the project RNA World
Nachricht
Autor
Matesko
Fingerzähler
Fingerzähler
Beiträge: 2
Registriert: 06.04.2012 09:33

cmserch XXL (long) 1.0.2 0.31 save during work

#1 Ungelesener Beitrag von Matesko » 06.04.2012 10:35

Hi,
I have problem with some units.
It is count normaly, but save checkpoint only first 20-38 sec.
And nothing else.
example:
cms_GA-p[e5-20MB_Lin64s]_1_Ornithorhyndus-anatinus-(platypus)_CM000414.lin.EBBL_RF00030_RNase_MRP_1324677953_59026

I got 21.3.2012 12:52:20, end is 10.4.2012 13:52:17
Last checkpoint save 00:00:20
CPU time 149:22:17
Time 154:23:24
finish 38,698%

It is same for Windows7 & Linux
After restart PC work started from 20 sec and I am not able to finish on time.
Whot is wrong?
Du hast keine ausreichende Berechtigung, um die Dateianhänge dieses Beitrags anzusehen.

Nemac
FAQ-Leser
FAQ-Leser
Beiträge: 186
Registriert: 05.01.2008 13:33

Re: cmserch XXL (long) 1.0.2 0.31 save during work

#2 Ungelesener Beitrag von Nemac » 06.04.2012 11:00

Hey,

rna doesn't support checkpointing

read here: http://www.rechenkraft.net/wiki/index.p ... ointing.3F

Benutzeravatar
Michael H.W. Weber
Vereinsvorstand
Vereinsvorstand
Beiträge: 23174
Registriert: 07.01.2002 01:00
Wohnort: Marpurk

Re: cmserch XXL (long) 1.0.2 0.31 save during work

#3 Ungelesener Beitrag von Michael H.W. Weber » 06.04.2012 11:24

In other words: If you do not use a 32 bit Linux in which memory randomization has been deactivated in the kernel, you will not have checkpointing available with this project.

Michael.
Fördern, kooperieren und konstruieren statt fordern, konkurrieren und konsumieren.

Bild Bild

Matesko
Fingerzähler
Fingerzähler
Beiträge: 2
Registriert: 06.04.2012 09:33

Re: cmserch XXL (long) 1.0.2 0.31 save during work

#4 Ungelesener Beitrag von Matesko » 06.04.2012 11:37

Ok,
I will change setting of projekt

THX

robertmiles
XBOX360-Installer
XBOX360-Installer
Beiträge: 86
Registriert: 23.02.2010 18:43
Wohnort: northern Alabama, US

Re: cmserch XXL (long) 1.0.2 0.31 save during work

#5 Ungelesener Beitrag von robertmiles » 06.05.2012 04:44

Do you expect many people to finish a workunit with an estimated runtime of about 7 weeks, and almost no checkpoints?


Workunit 5964755

name cms_GA-p[e5-20MB_Lin64f]_1_Caenorhabditis-elegans_BX284601.lin.EMBL_RF00028_Intron_gpI_1328586124_11004

application cmsearch XXL (long) 1.0.2

created 7 Feb 2012 20:52:32 UTC

minimum quorum 2

initial replication 2

max # of error/total/success tasks 15, 40, 6

estimated runtime on reference system 7w 4d 13h 3m 9s (4626189.1849962 s)


Or are you trying to emulate the last several months of Predictor@Home, when they weren't generating any new workunits, just sending repeatedly failed workunits again and again?

It looks like you should analyze the two copies that returned computing error, and check if they even give suitable output for telling why the computing errors occurred.

I'l let my copy run longer, but mainly because that computer is acting up enough that I want the antivirus program run now in progress to finish.

Benutzeravatar
yoyo
Vereinsvorstand
Vereinsvorstand
Beiträge: 8170
Registriert: 17.12.2002 14:09
Wohnort: Berlin

Re: cmserch XXL (long) 1.0.2 0.31 save during work

#6 Ungelesener Beitrag von yoyo » 06.05.2012 08:33

Hi Robert,

just deselect XXL workunits in your project settings if you don't like them. There are most of the time also S workunits available.

yoyo
HILF mit im Rechenkraft-WiKi, dies gibts zu tun.
Wiki - FAQ - Verein - Chat

Bild Bild

Ananas
WU-Schieber
WU-Schieber
Beiträge: 1184
Registriert: 27.04.2008 18:37
Wohnort: Nordlichter Köln

Re: cmserch XXL (long) 1.0.2 0.31 save during work

#7 Ungelesener Beitrag von Ananas » 06.05.2012 08:37

robertmiles hat geschrieben:Do you expect many people to finish a workunit with an estimated runtime of about 7 weeks, and almost no checkpoints? ....
When I have my 18-weeks-result done, I will tell you ;-) (p.s.: it's at 40 days CPU time now, showing 35% - the 100% phase sometimes needs extremely long though so the 35% might not be very exact)


The two crashes look very different, I doubt that they have a common cause.

We have seen 0xc3 several times, it is usually not symptomatic for the result and other boxes might still be able to finish them.

The other one has exit code 0x01 without showing any error message, hard to tell anything about it, maybe RAM shortage or an aborted task through the task manager or so. It would be good to know if there was a windows error message box on the screen but we would need that information from the user.
vi BOINC/checkin_notes
:1,$s/bug/feature/g
:wq!

Erzaehlen sich Biologen eigentlich Klein-RNA-Witze?

robertmiles
XBOX360-Installer
XBOX360-Installer
Beiträge: 86
Registriert: 23.02.2010 18:43
Wohnort: northern Alabama, US

Re: cmserch XXL (long) 1.0.2 0.31 save during work

#8 Ungelesener Beitrag von robertmiles » 07.05.2012 14:08

robertmiles hat geschrieben:Do you expect many people to finish a workunit with an estimated runtime of about 7 weeks, and almost no checkpoints?


Workunit 5964755

I'll let my copy run longer, but mainly because that computer is acting up enough that I want the antivirus program run now in progress to finish.
Sorry, 7 weeks without a checkpoint is just too much for a computer that has problems running a full week without rebooting.

Most of the XXL workunits are much shorter, and can actually finish in less than a week.

ConflictingEmotions

Re: cmserch XXL (long) 1.0.2 0.31 save during work

#9 Ungelesener Beitrag von ConflictingEmotions » 09.05.2012 16:25

Ananas hat geschrieben:
robertmiles hat geschrieben:Do you expect many people to finish a workunit with an estimated runtime of about 7 weeks, and almost no checkpoints? ....
When I have my 18-weeks-result done, I will tell you ;-) (p.s.: it's at 40 days CPU time now, showing 35% - the 100% phase sometimes needs extremely long though so the 35% might not be very exact)
Actually I have this estimated 9 week WU running on an i7 920 (hyperthreading enabled and no overclocking) with half the memory of the reference system with a newer kernel (not sure why boinc thinks the integer speeds are different). For about the last month that WU has been the only thing running on it as I am hoping a little turbo boosting might get it down quicker. About 4 days ago it jumped from about 60% to 100%. As I write this, boinc manager says the runtime of 5,097,600 secs (1416 hours) and the reference says 5,918,420 seconds. If the reference system is correct then there are 9-10 days left :roll2:

Usually I find that the reference system is an overestimate especially with a fast cpu (but not always). A previous WU on that same i7 920 host took 1,534,432.41 sec compared to the 2,872,543.9 sec on reference system.

ConflictingEmotions

Re: cmserch XXL (long) 1.0.2 0.31 save during work

#10 Ungelesener Beitrag von ConflictingEmotions » 10.05.2012 20:15

ConflictingEmotions hat geschrieben: Actually I have this estimated 9 week WU r
Of it dies on me with an 139 (0x8b) error eventhough the host is still going and cmsearch is still chewing up the cpu time.

The Knighty NI

Re: cmserch XXL (long) 1.0.2 0.31 save during work

#11 Ungelesener Beitrag von The Knighty NI » 06.06.2012 11:00

Hi

Still having trouble with the XXL Wu's.

Despite putting a new HD into the box with a full new fresh install of everything including BOINC, the XXL's are refusing to play and run to completion. They are still stalling somewhere between 22-48 hrs and re-starting from scratch.

My guess is that I don't have enough memory for these large WU's so they keep resetting. They are not like the old big WU's you used to have, just kept on shrubbing without a murmur. Just a complete pain if you had a power cut and had to start all over again.

De-selected them in the preferences on the website. However, they keep popping up on my puter from time to time. Its most annoying because I hate having to abort WU's from any project and by the time I spot them they have usually been running for several hours which is a complete waste of time.

What am I doing wrong?

Here is a screen shot of my preferences:

Bild

Ananas
WU-Schieber
WU-Schieber
Beiträge: 1184
Registriert: 27.04.2008 18:37
Wohnort: Nordlichter Köln

Re: cmserch XXL (long) 1.0.2 0.31 save during work

#12 Ungelesener Beitrag von Ananas » 06.06.2012 21:08

Heartbeat trouble kills a lot of results, this Berkeley bug even hits the fastest machines and can have various reasons, when for example more than one RNA results starts at the same time or more than one Malaria result checkpoints synchronous.

I have enabled only XXL and enable downloads only when work is needed, for a short moment. If I receive more than 1 result, I try to catch everything but the first with a "suspend", before BOINC has the chance to start it. When the first new one is started, I re-enable the first suspended one, then the next and so on.

I'm running a test wrapper that is not hit by the heartbeat bug at all anymore but all other projects on that box are affected of course.

@Yoyo : Despite of that missing test (do they end when BOINC ends? Michael, anything new from you about this?), you should release the new one, it is really a lot more reliable - afterall I currently have one result with 71 days on that box and the box had tons of heartbeat crashes in all other projects within that time. So one could definitely say that it has passed the stability test.
vi BOINC/checkin_notes
:1,$s/bug/feature/g
:wq!

Erzaehlen sich Biologen eigentlich Klein-RNA-Witze?

Zurück zu „RNA World Discussions (english)“