| View previous topic :: View next topic |
| Author |
Message |
The Knighty NI
Prince
Joined: 06 Sep 2007
Posts: 780
Location: Lost in space on a rather small Blue ping pong ball. :)
|
 Posted: Thu May 19, 2011 3:37 pm Post subject: |
 |
|
No current Bee RNA material WU's. I'm still in the process of finding research projects that produce RNA material who would like RNA World to process this for them.
Invalid WU's happen from time to time. Did you check to see if they were ones aborted by the project? If WU's are aborted by the project its usually because they have received enough results for a conclusion on that piece of RNA material.
You most likely would not have done any processing to them.
Inconclusive WU's will usually end up with you receiving credit. There is usually something in the WU's that are causing problems for lots of people. You will probably find that there will be about 5 completed WU's from various people before credit is allocated.
Since I started the project I have never had a WU with an actual error so error rates are very low if any. Probably processed in the region of 3-5000 units to date.
The project have stopped issuing very long WU's, some of which took over 1000 hrs to complete. Uber Credits  IMHO its sad that they have stopped thme  My best one gave me just over 15k credits and took about 300hrs on a 2000Mhz cpu Lol
They do state that for their test applications you may get errors under preferences so if you are concerned about errors don't chose the test application option
RNA World Preferrences
Run test applications?
This helps us develop applications, but may cause jobs to fail on your computer
_________________
What is that in the Shrubbery?
 |
|
| Back to top |
|
 |
Killerrabbit
Major Oblivion
Joined: 23 May 2002
Posts: 4656
Location: in a rabbit hole near you!!
|
| Posted: Fri May 20, 2011 7:51 pm Post subject: |
|
|
Do we really need bees anyway? They tend to sting.
Ni
_________________
 |
|
| Back to top |
|
|
Gemjunkie
Prince
Joined: 03 Jul 2010
Posts: 3519
Location: Earth, lately
|
| Posted: Fri May 20, 2011 8:30 pm Post subject: |
|
|
Totally unnecessary.
If you don't need food.
_________________
(older, before split CPID)
 |
|
| Back to top |
|
|
Michael H.W. Weber
Squire
Joined: 20 Aug 2007
Posts: 18
|
| Posted: Fri May 27, 2011 8:56 am Post subject: |
|
|
Hi,
I just got the message that you guys possibly plan to go for RNA World as PotM, so I thought it might be helpful showing up here to answer questions if there are any. 
Michael.
_________________
http://www.rnaworld.de |
|
| Back to top |
|
|
The Knighty NI
Prince
Joined: 06 Sep 2007
Posts: 780
Location: Lost in space on a rather small Blue ping pong ball. :)
|
| Posted: Fri May 27, 2011 1:50 pm Post subject: |
|
|
Hi Michael
Welcome to KWSN and thanks for coming round.
I'm sure there will be plenty of questions when team members log on
_________________
What is that in the Shrubbery?
|
|
| Back to top |
|
|
Gemjunkie
Prince
Joined: 03 Jul 2010
Posts: 3519
Location: Earth, lately
|
| Posted: Fri May 27, 2011 6:26 pm Post subject: |
|
|
Hi Michael
I aborted a couple of xxl WU's. They were nearly finished with over 20hrs. elapsed, but after shutting down the computer for a couple of hours they resumed at 0hrs. and 0% completion.  Is there a checkpointing issue? I see only this one still listed in my tasks:
http://www.rnaworld.de/rnaworld/workunit.php?wuid=5169236
I've deselected the xxl sub-project in preferences.
_________________
(older, before split CPID)
|
|
| Back to top |
|
|
Sir Forth of Erebourne
Knight
Joined: 12 Sep 2003
Posts: 60
|
| Posted: Fri May 27, 2011 7:46 pm Post subject: |
|
|
Will you be able to supply enough WU for us?
Forth |
|
| Back to top |
|
|
Michael H.W. Weber
Squire
Joined: 20 Aug 2007
Posts: 18
|
| Posted: Fri May 27, 2011 9:34 pm Post subject: |
|
|
| Gemjunkie wrote: | | I aborted a couple of xxl WU's. They were nearly finished with over 20hrs. elapsed, but after shutting down the computer for a couple of hours they resumed at 0hrs. and 0% completion. Is there a checkpointing issue? |
Well, that is indeed normal behaviour as we do not have checkpointing available at this time except if you run 32 bit Linux with memory randomization disabled in the kernel.
Maybe a few more general details on that: we have the CMS S and XXL applications available. The S app will bundle very (!) short WUs in a pack that runs about a few hours each. Such a bundle can hold several hundred WUs if we are for example analyzing viruses (as will be the case very soon). If you run these WUs and reboot your machine, the whole bundle will not go to waste except for the sub-WU that was in progress upon time of reboot. So, it resumes this WU after restart and proceeds through the rest of WUs in this pack and this means you have sort of a checkpointing for all machines. The XXL WUs are not bundled, so, rebooting means that that WU will restart from the very beginning if you do not have 32 bit Linux as stated above. But there is a simple solution to the missing checkpoints: Just do not reboot but go to sleep mode if e.g. a laptop has to be moved or you need to turn the machine off for whatever reason. On top of that the really long running WUs get extra credit to compensate a bit for the long runtimes and lack of checkpointing.
Why don't we have general checkpoints at this time? Because we use highly complex science apps that were designed to run on a massive cluster infrastructure. The code is so complicated that it would have to be rewritten which currently simply is impossible due to lack of manpower. Moreover, because we plan to incorporate many more apps in the near future (including GPU-support at some time) it makes no sense putting these tremendous efforts into checkpointing rather than into the real science. What we are currently working on instead, is a checkpointing that operates from within BOINC. This means that whatever app runs in BOINC, it will checkpoint if belonging to the RNA World project. No further fiddling around required. I think this is the right way to focus our manpower and actually I am a bit surprised that other projects have not come up with such an implementation much earlier. They get the funding, they have the people, they have the time and they do get paid.
You should note one important thing: our project is based on the work of people at Rechenkraft.net that are interested in science without getting payed for what we do for this project. Consequently, developments are sometimes taking more time as would be the case if it was our full-time job. I hope you understand that and maybe even appreciate our efforts.
The bundling approach detailed above for example results in an extreme reduction of client-server communication and therefore server load making our system much mor stable.
You should note one more thing: The S apps will be available only as long as we have small WUs to do - so, disabling XXL means that you **might** run out of supply if we are running low on these packs. But we can try to communicate that in time on the webpage and currently it does not look as if that app would be out of work anytime soon.
| Sir Forth of Erebourne wrote: | | Will you be able to supply enough WU for us? |
Yes. We currently have an almost constant throughput of 6.6 TFLOPS and I do not think we would have a problem with increasing that severalfold.
When exactly would that PotM start?
Ni! 
Michael.
_________________
http://www.rnaworld.de |
|
| Back to top |
|
|
Blurf
Prince
Joined: 03 Jan 2011
Posts: 1090
Location: Upstate NY
|
| Posted: Fri May 27, 2011 10:19 pm Post subject: |
|
|
| Michael H.W. Weber wrote: | When exactly would that PotM start?
Ni!
Michael. |
June 1st
_________________
 |
|
| Back to top |
|
|
Gemjunkie
Prince
Joined: 03 Jul 2010
Posts: 3519
Location: Earth, lately
|
| Posted: Fri May 27, 2011 11:29 pm Post subject: |
|
|
| Michael H.W. Weber wrote: | | Well, that is indeed normal behaviour as we do not have checkpointing available at this time except if you run 32 bit Linux with memory randomization disabled in the kernel. |
It would have been exceeding helpful to know that in advance. 
_________________
(older, before split CPID)
|
|
| Back to top |
|
|
Michael H.W. Weber
Squire
Joined: 20 Aug 2007
Posts: 18
|
| Posted: Sat May 28, 2011 7:49 am Post subject: |
|
|
| Gemjunkie wrote: | | Michael H.W. Weber wrote: | | Well, that is indeed normal behaviour as we do not have checkpointing available at this time except if you run 32 bit Linux with memory randomization disabled in the kernel. |
It would have been exceeding helpful to know that in advance. |
Well, it is all described in the FAQ.
Michael.
_________________
http://www.rnaworld.de |
|
| Back to top |
|
|
Gemjunkie
Prince
Joined: 03 Jul 2010
Posts: 3519
Location: Earth, lately
|
| Posted: Sat May 28, 2011 3:11 pm Post subject: |
|
|
It would have been exceeding helpful to read the FAQ's in advance. 
| Michael H.W. Weber wrote: | | Just do not reboot but go to sleep mode if e.g. a laptop has to be moved or you need to turn the machine off for whatever reason. |
Sleep mode isn't fully powered down though, right? If I unplug and move the computer it'll still lose everything. Don't you mean we should Hibernate the machine? Or does my brain need a reboot? 
_________________
(older, before split CPID)
|
|
| Back to top |
|
|
The Knighty NI
Prince
Joined: 06 Sep 2007
Posts: 780
Location: Lost in space on a rather small Blue ping pong ball. :)
|
| Posted: Sat May 28, 2011 3:33 pm Post subject: |
|
|
Select all RNA World WU's and suspend them first.
Boot down BOINC then switch off your puter the WU's will be saved in their current state.
When you reboot your machine select all the RNA World WU's and restart them.
That way you will not lose any of the processing you have done.
I sadly had a power cut when doing a very large WU's once and lost almost 2 weeks of processing and had to start the WU from scratch Errrrk. Upset.... I was spitting feathers for a week and still feel sore about it.. Lol
_________________
What is that in the Shrubbery?
|
|
| Back to top |
|
|
LanDroid
Prince
Joined: 11 Jun 2002
Posts: 4476
Location: Cincinnati, OH U.S.
|
| Posted: Sat May 28, 2011 8:46 pm Post subject: |
|
|
"I was spitting feathers for a week and still feel sore about it..."
Indeed. Every computer cycle is sacred. I can't imagine losing quadrillions of them.  |
|
| Back to top |
|
|
Blurf
Prince
Joined: 03 Jan 2011
Posts: 1090
Location: Upstate NY
|
| Posted: Sun May 29, 2011 9:41 am Post subject: |
|
|
I finish my load of Prima and figured I'd jump on RNA for a cpl days..not getting any WU's
_________________
|
|
| Back to top |
|
|
Altivec
Knight
Joined: 13 Aug 2003
Posts: 88
Location: Hawaii
|
| Posted: Sun May 29, 2011 12:11 pm Post subject: |
|
|
Same here, no new work units getting out.
_________________
 |
|
| Back to top |
|
|
Blurf
Prince
Joined: 03 Jan 2011
Posts: 1090
Location: Upstate NY
|
| Posted: Sun May 29, 2011 7:20 pm Post subject: |
|
|
Just pulled down 20 WUs
_________________
|
|
| Back to top |
|
|
Altivec
Knight
Joined: 13 Aug 2003
Posts: 88
Location: Hawaii
|
| Posted: Sun May 29, 2011 8:38 pm Post subject: |
|
|
Problems solved, pulled down a bunch an hour ago.
_________________
|
|
| Back to top |
|
|
Michael H.W. Weber
Squire
Joined: 20 Aug 2007
Posts: 18
|
| Posted: Sun May 29, 2011 9:30 pm Post subject: |
|
|
| Gemjunkie wrote: | | Sleep mode isn't fully powered down though, right? If I unplug and move the computer it'll still lose everything. Don't you mean we should Hibernate the machine? Or does my brain need a reboot? |
Yes, you are right - I meant to hibernate the machine.
Michael.
_________________
http://www.rnaworld.de |
|
| Back to top |
|
|
Gemjunkie
Prince
Joined: 03 Jul 2010
Posts: 3519
Location: Earth, lately
|
| Posted: Mon May 30, 2011 2:00 am Post subject: |
|
|
RNA's ahead in the run-off election, you're in danger of being stuck with us for a whole month.
_________________
(older, before split CPID)
|
|
| Back to top |
|
|
|
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
