climateprediction.net home page
Irritated - some questions

Irritated - some questions

Message boards : Number crunching : Irritated - some questions
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
Thomas Wiegand

Send message
Joined: 4 Jul 19
Posts: 24
Credit: 60,647
RAC: 1,178
Message 60731 - Posted: 28 Jul 2019, 4:51:36 UTC

I just finished a first work, wondering

- that my task had been calculated before already with a failure.
Why, as so much flops available, they aren't sent out twice = bigger option of a good clean result.

- my pc was ready in about 7 days, but sill a lot units still in work
Isn't it important to get fast results, as climate change IS FAST.
Why not using the fastest PCs available when a new set is sent ?
(WE all want to help, ok - I would evem use my slower PCs in same time to calculate a same set, so after another 10 days there is a proofed 2nd to check.)

Greets to all here,
I am in Thailand and a 9 year old daughter - and need to be prepared for this future to protect her.
ID: 60731 · Report as offensive     Reply Quote
Thomas Wiegand

Send message
Joined: 4 Jul 19
Posts: 24
Credit: 60,647
RAC: 1,178
Message 60732 - Posted: 28 Jul 2019, 5:31:37 UTC

about me:
I am 53, from Germany living since 13 years in Thailand.

My first knowledge about something wrong was until 1990, as I visited every year glacier in Switzerland as saw the melting got faster, what couldn't be if temperatures are same = I knew so that it got warmer then before.
As if the glacier would melt until an balance of existing temperatures, the lost should get less every summer. But it got more.

In my years here in Thailand I then saw then from 2010 until now how weather patters change:
* the monsoon got earlier, less strong, and
* instead of no rain, even clouds around Chiangmai after 1.11. now the rain still comes back until New-year about every 12 days, and every year stronger. started with clouds, some drops until strong day lasting rain now.

The Climate change is real ... and fast.

I might die near normal with 90 years of age, but my daughter will face hard times and not sure if she might die normal with a good age.
ID: 60732 · Report as offensive     Reply Quote
Profile Dave Jackson
Volunteer moderator

Send message
Joined: 15 May 09
Posts: 2463
Credit: 3,124,201
RAC: 383
Message 60733 - Posted: 28 Jul 2019, 9:26:56 UTC - in response to Message 60731.  

I just finished a first work, wondering

- that my task had been calculated before already with a failure.
Why, as so much flops available, they aren't sent out twice = bigger option of a good clean result.

- my pc was ready in about 7 days, but sill a lot units still in work
Isn't it important to get fast results, as climate change IS FAST.
Why not using the fastest PCs available when a new set is sent ?
(WE all want to help, ok - I would evem use my slower PCs in same time to calculate a same set, so after another 10 days there is a proofed 2nd to check.)

Greets to all here,
I am in Thailand and a 9 year old daughter - and need to be prepared for this future to protect her.


HI, welcome to the community here. CPDN doesn't send out two copies of a task to check because unless the computers are identical, the results won't be exactly the same. This is due to the complexities of climate and the way the programs are written. A complex statistical analysis is used to assess the validity of results. Sometimes a task crashing isn't a failure but because the conditions programmed into the model are pushing the boundaries of what is possible and an impossible climate e.g. a negative pressure is produced. Currently two of the tasks I am running have previously crashed. I think they will probably complete on my rather slower machines.

Have a good read around the forums because they certainly contain answers to man of the questions I had. I still have questions from time to time!
ID: 60733 · Report as offensive     Reply Quote
Thomas Wiegand

Send message
Joined: 4 Jul 19
Posts: 24
Credit: 60,647
RAC: 1,178
Message 60734 - Posted: 28 Jul 2019, 11:34:49 UTC - in response to Message 60733.  

CPDN doesn't send out two copies of a task to check because unless the computers are identical, the results won't be exactly the same. This is due to the complexities of climate and the way the programs are written. ...

Have a good read around the forums because they certainly contain answers to man of the questions I had. I still have questions from time to time!


I have two I7-4770 with diff MB etc. also another 5-6 PC which are all diff in CPU, most also HP DELL acer ... .
I guess to calculate 4 tasks with 2 diff downloads in fast 7 days would help a lot in speed (the main 2 diff downloads on the I7-4770) and calculate the same 2 on two other slower machines slower like 10 days, could help to force speed incl. save results.

We need results, that we can prepare our-self.
ID: 60734 · Report as offensive     Reply Quote
Jim1348

Send message
Joined: 15 Jan 06
Posts: 372
Credit: 15,130,852
RAC: 56
Message 60735 - Posted: 28 Jul 2019, 13:15:08 UTC - in response to Message 60734.  

I have two I7-4770 with diff MB etc. also another 5-6 PC which are all diff in CPU, most also HP DELL acer ... .
I guess to calculate 4 tasks with 2 diff downloads in fast 7 days would help a lot in speed (the main 2 diff downloads on the I7-4770) and calculate the same 2 on two other slower machines slower like 10 days, could help to force speed incl. save results.

We need results, that we can prepare our-self.


If you look at the due dates, they are one year out. Speed is not their thing.
ID: 60735 · Report as offensive     Reply Quote
Thomas Wiegand

Send message
Joined: 4 Jul 19
Posts: 24
Credit: 60,647
RAC: 1,178
Message 60742 - Posted: 30 Jul 2019, 5:56:36 UTC - in response to Message 60735.  
Last modified: 30 Jul 2019, 5:57:46 UTC

If you look at the due dates, they are one year out. Speed is not their thing.


THAT I saw. Wondering !

How to make sure the fast machines are on and ready to work when a new set is getting out ?

Is there any other Boinc to support this direction ? - the other times most best is shut them down, and save electric.

Thanks
ID: 60742 · Report as offensive     Reply Quote
mmonnin

Send message
Joined: 28 May 17
Posts: 40
Credit: 2,747,500
RAC: 1,118
Message 60745 - Posted: 30 Jul 2019, 10:37:55 UTC - in response to Message 60742.  

If you look at the due dates, they are one year out. Speed is not their thing.

the other times most best is shut them down, and save electric.

Thanks


I, and others, run other projects when there is no CPDN work available. CPDN has a lack of work, not a lack of CPU power.
ID: 60745 · Report as offensive     Reply Quote
Thomas Wiegand

Send message
Joined: 4 Jul 19
Posts: 24
Credit: 60,647
RAC: 1,178
Message 60767 - Posted: 1 Aug 2019, 2:56:48 UTC
Last modified: 1 Aug 2019, 3:17:12 UTC

Irritation gets more ...

I just got new Tasks 31 Jul 2019, 22:54:56 UTC, N144 resolution
but my Work PC got them, crashed 3 times yet with 4 Tasks,
but the other also running PC with same CPU got nothing. ???

At first, and after 2 restart even 4 tasks where running, now after again crash/freeze only 2 Task came up. - until now now new freeze.

Error:
error while loading shared libraries: libstdc++.so.6:
did then sudo update upgrade ...

see: https://www.cpdn.org/cpdnboinc/results.php?hostid=1487884 - show 4

When there isn't any logic behind all this:
- why not both running fast PC got ?
- why not any post about, or status page related ... still 53 to send ... or I am to dumb to read that page
... I have to rethink this engagement.

I am not sure if I let PC run all time for that 7 days it might take (like before), or close at night.
As there are still so much tasks in progress and I cant estimate if that are old ones on slow less time running PC, or there are often coming new ones so keeping that number high ?

It is not fixing together that too much pc power and to less work ..., if then that in progress should reduce quite fast.
ID: 60767 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 6951
Credit: 20,843,205
RAC: 0
Message 60768 - Posted: 1 Aug 2019, 3:42:27 UTC - in response to Message 60767.  

error while loading shared libraries: libstdc++.so.6:

is because these models are 32 bits, and that missing lib is needed for that reason.

**********************

The "53 waiting to send" is a BOINC anomaly.
Apparently all projects have it, but it only shows up here because the download queue is actually empty.


**********************

All computers are treated equally, fast or slow.
It's not important, as long as they complete the tasks in a reasonable time.
ID: 60768 · Report as offensive     Reply Quote
Thomas Wiegand

Send message
Joined: 4 Jul 19
Posts: 24
Credit: 60,647
RAC: 1,178
Message 60770 - Posted: 1 Aug 2019, 4:24:58 UTC - in response to Message 60768.  
Last modified: 1 Aug 2019, 4:29:01 UTC

error while loading shared libraries: libstdc++.so.6:
is because these models are 32 bits, and that missing lib is needed for that reason.

? 2 Tasks are running but 4 not ... I am confused

and try to update, or install too much possibilities and dependencies. Stopped


All computers are treated equally, fast or slow.
It's not important, as long as they complete the tasks in a reasonable time.

reasonable time: the new one have time until ? 13.07.2020 is see - wow, seams not to be urgent
OK, sound nice all all same, is there any other selection like last contact of a computer to be selected ? Or just bad luck that of my 2 online just one got some ? (as the 2nd one "-32b" had no manager open and a older responce)


added:
I see at other thread, failed tasks are send new, ok -
and go to computer just asked for work.

Yes I followed the "UK Met Office HadAM4 at N144 resolution 0 310" started with 626 about, I was ready with my one just under 600, now is 310 ...
ID: 60770 · Report as offensive     Reply Quote
Profile Dave Jackson
Volunteer moderator

Send message
Joined: 15 May 09
Posts: 2463
Credit: 3,124,201
RAC: 383
Message 60771 - Posted: 1 Aug 2019, 4:45:17 UTC

error while loading shared libraries: libstdc++.so.6:
is because these models are 32 bits, and that missing lib is needed for that reason.
? 2 Tasks are running but 4 not ... I am confused


If some task types crash with a missing lib message and others don't it is because those that don't don't need the one listed. and any others needed are already installed.



If you run the following command Ubuntu 18.04, 10 and 19.04 (can't remember when it changed from a slightly different command.)

sudo apt-get install lib32ncurses5 lib32z1 lib32stdc++-6-dev

It will install the 32bit libs for all the task types that require 32bit library files. Not all are required by all task types




Or just bad luck that of my 2 online just one got some ?


Yes, while the only tasks available are retreads, your computer has to ask for work immediately after the re-sends go out to get them.
ID: 60771 · Report as offensive     Reply Quote
Thomas Wiegand

Send message
Joined: 4 Jul 19
Posts: 24
Credit: 60,647
RAC: 1,178
Message 60772 - Posted: 1 Aug 2019, 4:55:04 UTC
Last modified: 1 Aug 2019, 4:56:31 UTC

"sudo apt-get install lib32ncurses5 lib32z1 lib32stdc++-6-dev" worked,
but I guess I need to restart PC and boinc to make again 4 Tasks running, instead of the 2 now ... I wait for an freeze first,
then I will see if still 2 or 4 Task again


computers are asking,
mean that is the most 1 hour under status "Communication deferred xx:xx:xx" ?
and I guess that "1 hour" is set fixed.
ID: 60772 · Report as offensive     Reply Quote
Thomas Wiegand

Send message
Joined: 4 Jul 19
Posts: 24
Credit: 60,647
RAC: 1,178
Message 60773 - Posted: 1 Aug 2019, 9:44:31 UTC

update,
an electric shutdown 30 min, and restart computer made it,
still only 2 Tasks on 8 CPU - so far a bit irritated why, how and if ...
ID: 60773 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 6951
Credit: 20,843,205
RAC: 0
Message 60774 - Posted: 1 Aug 2019, 10:09:13 UTC

The work here is NOT from Oxford, it's from climate research centers around the world. Various climate physicists are working on different projects, and only submit work when they're ready to do so.

At the moment, there is NO work available, so I don't know why you keep saying that your computers aren't getting any.

Just wait until there IS more work.
ID: 60774 · Report as offensive     Reply Quote
Profile Dave Jackson
Volunteer moderator

Send message
Joined: 15 May 09
Posts: 2463
Credit: 3,124,201
RAC: 383
Message 60775 - Posted: 1 Aug 2019, 12:12:25 UTC

computers are asking,
mean that is the most 1 hour under status "Communication deferred xx:xx:xx" ?
and I guess that "1 hour" is set fixed.


Yes, the one hour is fixed or one hour and one minute to be precise. If you try a manual update within that time, it won't get work but it does start the, "hour" again.

As Les says, there isn't any new work at the moment so anything you get will be retreads of tasks that have failed on other computers. With the Linux tasks I find the vast majority of these are ones that have failed due to the missing lib files.

The good news is that your Linux boxes all have at least the minimum amount of memory to run the new 64bit tasks once they appear.
ID: 60775 · Report as offensive     Reply Quote
Thomas Wiegand

Send message
Joined: 4 Jul 19
Posts: 24
Credit: 60,647
RAC: 1,178
Message 60782 - Posted: 3 Aug 2019, 7:52:09 UTC

as I found this Error while computing Tasks now in
"/var/libs/boinc-client/projects/..."
* does climateprediction know that also ?
- and give them to someone else (already) ?
* or are they checked and running later on my PC again and then reported as Error if cant (even with updated libs)
- or can I force this 2 Tasks to restart, or to copy on other of my PC, so that they are ready more early ?
ID: 60782 · Report as offensive     Reply Quote
Profile Dave Jackson
Volunteer moderator

Send message
Joined: 15 May 09
Posts: 2463
Credit: 3,124,201
RAC: 383
Message 60783 - Posted: 3 Aug 2019, 9:54:56 UTC - in response to Message 60782.  

as I found this Error while computing Tasks now in
"/var/libs/boinc-client/projects/..."
* does climateprediction know that also ?
- and give them to someone else (already) ?
* or are they checked and running later on my PC again and then reported as Error if cant (even with updated libs)
- or can I force this 2 Tasks to restart, or to copy on other of my PC, so that they are ready more early ?


Tasks sent to users computers from CPDN (and other projects for that matter) are assigned to a unique number for that computer and will if returned by another computer as an error.
The link in your post gives an unable to handle request error so I can't be sure about what the error is. However of your tasks that have given errors apart from any missing library ones, they have failed with the same errors on other computers indicating that the problem is not with your machine.

If they have reported as errors they will have been assigned to other computers unless as in one I looked at the task name ends in _2 which means it has had its three attempts and that is it.
ID: 60783 · Report as offensive     Reply Quote
Thomas Wiegand

Send message
Joined: 4 Jul 19
Posts: 24
Credit: 60,647
RAC: 1,178
Message 60784 - Posted: 4 Aug 2019, 0:45:32 UTC
Last modified: 4 Aug 2019, 0:49:10 UTC

hmm, named https://www.cpdn.org/cpdnboinc/workunit.php?wuid=11885091
name hadam4_a0aq_201210_12_826_011885091
application UK Met Office HadAM4 at N144 resolution
created 21 Jun 2019, 8:34:57 UTC
minimum quorum 1
initial replication 1
max # of error/total/success tasks 3, 3, 1
errors Too many total results


and https://www.cpdn.org/cpdnboinc/workunit.php?wuid=11885190
name hadam4_a0dh_201410_12_826_011885190
application UK Met Office HadAM4 at N144 resolution
created 21 Jun 2019, 8:37:10 UTC
minimum quorum 1
initial replication 1
max # of error/total/success tasks 3, 3, 1


and on the other side I cant follow the 1 hour update, as now 3 PCs running 24/7 since last 10,5 hour electrical shutdown, AND sure also the boinc-manager is running with open window, and no power save etc. but
* ID: 1487908: 3 Aug 2019, 23:34:39 UTC (a 24/7 game server)
* ID: 1487884: 3 Aug 2019, 21:18:28 UTC (the one with 2 active Tasks)
* ID: 1487933: 3 Aug 2019, 17:51:25 UTC (a 24/7 webserver)
what ment 3.5 hours min. no connection, no update, ? - but all pc and network are online and active. Even no similar lost of connect timeout that would indicate that.
ID: 60784 · Report as offensive     Reply Quote
Thomas Wiegand

Send message
Joined: 4 Jul 19
Posts: 24
Credit: 60,647
RAC: 1,178
Message 60785 - Posted: 4 Aug 2019, 23:41:54 UTC

after again, the computer "web-server" had been 7 hours after actual "ready to work Tasks" in the list of my units, I shut it down.
As basically I have normal 2 PC running 24/7 and would easy be available to work, but they are too often not "ready", and my work pc gets the Tasks, which I wanna to shutdown in the night but as I think climate change and its prediction is imminent to look for - I let it run when there are Tasks,
I have to mark this pc no new Tasks, as this is against reduce climate change, when I polite the world with more CO2 then it makes sense.
ID: 60785 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 6951
Credit: 20,843,205
RAC: 0
Message 60786 - Posted: 5 Aug 2019, 3:11:56 UTC - in response to Message 60785.  

I'm not sure what you're talking about, but at the moment there isn't any work, and likely won't be for a few weeks.

The only reason that you need a computer running, is if you have some work that hasn't been completed, and that's this computer: ID: 1487884
ID: 60786 · Report as offensive     Reply Quote
1 · 2 · Next

Message boards : Number crunching : Irritated - some questions

©2019 climateprediction.net