Message boards : Number crunching : What's with the tiny work units?
Message board moderation

To post messages, you must log in.

AuthorMessage
TroubledBunny

Send message
Joined: 26 Oct 06
Posts: 3
Credit: 9,188
RAC: 0
Message 18984 - Posted: 8 Feb 2008, 10:54:51 UTC

What's the point in distributing work units that take a second or less to process?
The overhead of distributing them and collecting them is more than required processing time.
I get 0 credits for this and I'd rather be crunching something else. If that's your strategy for making sure folks get something to crunch then it's ludicrous.

Pete
ID: 18984 · Report as offensive     Reply Quote
Ken Phillips
Avatar

Send message
Joined: 28 Sep 04
Posts: 1
Credit: 81,533
RAC: 0
Message 18985 - Posted: 8 Feb 2008, 11:50:07 UTC - in response to Message 18984.  
Last modified: 8 Feb 2008, 11:50:33 UTC

What's the point in distributing work units that take a second or less to process?
The overhead of distributing them and collecting them is more than required processing time.
I get 0 credits for this and I'd rather be crunching something else. If that's your strategy for making sure folks get something to crunch then it's ludicrous.

Pete


Pete,

I presume you would prefer it if the simulation process ignored the real world possibility of loss of beam confinement, the beams wandering off somewhere and not colliding when they should, or colliding with something they shouldn't?

This is what Boinc running LHC workunits is simulating! If the simulation fails very shortly after start-up, then it shuts down and reports the data, if the real life LHC was to suffer a similar failure the consequences could well be a lot worse than just zero credits.

The short run workunits are just as valuable (may be more valuable) to the project as the ones that run to completion.

There is nothing ludicrous about building a body of scientific proof, even if it's only from simulations; and that, in common with a lot of other distributed computing is what it's about, not a race for highest earned points.

Best wishes
Ken Phillips
ID: 18985 · Report as offensive     Reply Quote
Perle
Avatar

Send message
Joined: 25 Oct 04
Posts: 83
Credit: 86,727,174
RAC: 46,615
Message 18986 - Posted: 8 Feb 2008, 12:30:43 UTC

Nicely stated Ken.
The combined contribution from so many participants is whats important.

* I am having flashbacks of 'cool hand luke' *
Just like digging a ditch...Some of us have bigger and faster shovels but quite often it is the waterboy that is the most important part of the team.

Not to compare your efforts to that of the under-appreciated waterboy, but that those 'zeros' you have produced may be the most important piece of data that the scientists obtain.
I count a multitude of 0's withing my completed wu's.

peace
ID: 18986 · Report as offensive     Reply Quote
TroubledBunny

Send message
Joined: 26 Oct 06
Posts: 3
Credit: 9,188
RAC: 0
Message 18987 - Posted: 8 Feb 2008, 12:58:41 UTC

Ken,
Thanks for your reply.
I apologise for implying that you are trying to appease the contributors to your project by sending useless simulations.

If these particular results are so highly valuable then why do we not receive a commensurate amount of credit? Even a token would be something.

Pete
ID: 18987 · Report as offensive     Reply Quote
TroubledBunny

Send message
Joined: 26 Oct 06
Posts: 3
Credit: 9,188
RAC: 0
Message 18988 - Posted: 8 Feb 2008, 13:02:32 UTC - in response to Message 18985.  

Perle/peace,
If the waterboy is so valued then wouldn't it be sensible to encourage him in some way?
Pete
ID: 18988 · Report as offensive     Reply Quote
KAMasud

Send message
Joined: 7 Oct 06
Posts: 114
Credit: 23,192
RAC: 0
Message 18989 - Posted: 8 Feb 2008, 13:42:26 UTC


:) LoL'z. You do have a good point there ;) but we still do it 8-) Fools that we are :( Maybe just for the heck of it.
Regards
Masud.
ID: 18989 · Report as offensive     Reply Quote
Colin Porter

Send message
Joined: 14 Jul 05
Posts: 35
Credit: 71,636
RAC: 0
Message 18990 - Posted: 8 Feb 2008, 14:48:32 UTC
Last modified: 8 Feb 2008, 14:49:53 UTC

From what I believe - There is a plan.

Some time in the future (when the LHC@Home BOINC version is updated) all the WU's
that currently show zero as pending credit but actually claim a very small amount
of credit (i.e. 0.9 or less)will all be cleaned up and either credit awarded or just deleted!

Yes, just deleted, sounds bad, but when you consider that I have approaching 100 of these WU's and most claim in the order of 0.00215526329022802, well I think even the real credit chasers can live with losing these sort of amounts in the cause of scientific progress.

EDIT - Dosen't stop them being annoying though. End EDIT
ID: 18990 · Report as offensive     Reply Quote
Profile RedMenace

Send message
Joined: 27 Nov 07
Posts: 5
Credit: 1,179,133
RAC: 0
Message 18991 - Posted: 8 Feb 2008, 16:08:55 UTC - in response to Message 18990.  

Thanks for the explanation. It's the first I've seen on these annoying wu's. Keep in mind that if I was only chasing credits, I wouldn't be touching LHC at all ;) I think it's safe to say pretty much everyone here places science before ego.
ID: 18991 · Report as offensive     Reply Quote
Perle
Avatar

Send message
Joined: 25 Oct 04
Posts: 83
Credit: 86,727,174
RAC: 46,615
Message 18992 - Posted: 8 Feb 2008, 16:16:59 UTC

Yes..encourage the waterboy..
Make him wear the bucket on his head and let him dig the ditch with the ladel.
Then.....flog him until moral improves.

It does appear that many of these zero credit results only required what appears to be very little cpu times.
At least those did not required 3hour or more to complete and earn minimal credit.


ID: 18992 · Report as offensive     Reply Quote
Brian Silvers

Send message
Joined: 3 Jan 07
Posts: 124
Credit: 7,065
RAC: 0
Message 18993 - Posted: 8 Feb 2008, 17:56:22 UTC

I find this thread to be........quite odd.....considering I'm used to seeing tasks that take only 30 minutes on my system and the 4 that I got over the past day are taking 4-5 hours a piece...

I was going to come here and ask "What's with the BIG work units?"

Oh, the irony.....
ID: 18993 · Report as offensive     Reply Quote
Perle
Avatar

Send message
Joined: 25 Oct 04
Posts: 83
Credit: 86,727,174
RAC: 46,615
Message 18994 - Posted: 8 Feb 2008, 19:32:07 UTC
Last modified: 8 Feb 2008, 19:35:17 UTC

Brian when the project started...wu's would run about 2 - 5 hours.

Realise the hardware of the day of course, Amd socket 462 / 754 and 939 or a 940 Opteron.
Intel P4 socket 423 and 478.. 603/604 xeons....netburst and hyperthreading.
Lets not forget the venerable and rock solid P3 slotted and socket 370.
Now we are dual/quad core and multi proc (8 and 16 cores per box) etc.
Bus speeds are up, ram speeds etc, as well as the BOINC code and the Sixtrack code being optimised in various ways.

So my guess is that these new wu's are much more technical than those from a year or 2 ago, but proportional in relation to current hardware.

My dual xeon 3.0 ghz /800fsb wants 15 hours vs. a dual core 6700 at 4.5hours per wu.


ID: 18994 · Report as offensive     Reply Quote
EclipseHA

Send message
Joined: 18 Sep 04
Posts: 47
Credit: 1,886,234
RAC: 0
Message 18996 - Posted: 8 Feb 2008, 22:54:57 UTC

The latest batch of WU's could be 1,000,000 turn vs 100,000 turn.

We used to get those in the past.

Needless to say, the 1m turn took about 10x as long as the 100k turn.
ID: 18996 · Report as offensive     Reply Quote
glaesum

Send message
Joined: 16 Oct 06
Posts: 15
Credit: 144,247
RAC: 0
Message 18998 - Posted: 9 Feb 2008, 0:20:25 UTC - in response to Message 18996.  

The latest batch of WU's could be 1,000,000 turn vs 100,000 turn.

We used to get those in the past.

Needless to say, the 1m turn took about 10x as long as the 100k turn.


just think, in the real-life LHC collider, those 1m laps of the 27km tunnel will take the protons just a minute and a half!
ID: 18998 · Report as offensive     Reply Quote
Brian Silvers

Send message
Joined: 3 Jan 07
Posts: 124
Credit: 7,065
RAC: 0
Message 18999 - Posted: 9 Feb 2008, 0:57:03 UTC - in response to Message 18994.  


So my guess is that these new wu's are much more technical than those from a year or 2 ago, but proportional in relation to current hardware.

My dual xeon 3.0 ghz /800fsb wants 15 hours vs. a dual core 6700 at 4.5hours per wu.



As others have mentioned, it is a higher number of turns around the track. It is precisely 10x the amount of the previous tasks. If it took me 33 minutes before, it now takes 330 minutes...within a few minutes... There really isn't any more detail in the application (we'd have to have a new app version to get that), so there ya have it...
ID: 18999 · Report as offensive     Reply Quote
Jayargh

Send message
Joined: 24 Oct 04
Posts: 79
Credit: 257,762
RAC: 0
Message 19000 - Posted: 9 Feb 2008, 4:44:24 UTC

We have had million turn studies over the years that run the ranges of sizes from 30 credits =3hrs to these type and an even longer version 150 credits=20 hrs on the same type technology.

The million turn studies rarely crash into the collider hence no 0 claims here.
ID: 19000 · Report as offensive     Reply Quote
Richard Haselgrove

Send message
Joined: 27 Oct 07
Posts: 186
Credit: 3,297,640
RAC: 0
Message 19002 - Posted: 9 Feb 2008, 11:23:25 UTC - in response to Message 19000.  

We have had million turn studies over the years that run the ranges of sizes from 30 credits =3hrs to these type and an even longer version 150 credits=20 hrs on the same type technology.

The million turn studies rarely crash into the collider hence no 0 claims here.

Rare, perhaps, but not entirely unknown. WU 2139867.
ID: 19002 · Report as offensive     Reply Quote
PJ

Send message
Joined: 18 Sep 04
Posts: 8
Credit: 1,181,841
RAC: 0
Message 19003 - Posted: 9 Feb 2008, 11:55:58 UTC - in response to Message 19002.  

We have had million turn studies over the years that run the ranges of sizes from 30 credits =3hrs to these type and an even longer version 150 credits=20 hrs on the same type technology.

The million turn studies rarely crash into the collider hence no 0 claims here.

Rare, perhaps, but not entirely unknown. WU 2139867.


Maybe the meaning was that if it runs past the initial minutes it will likely finish?
ID: 19003 · Report as offensive     Reply Quote

Message boards : Number crunching : What's with the tiny work units?


©2024 CERN