Message boards : Number crunching : Very Short Running on Tasks
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Tom95134

Send message
Joined: 4 May 07
Posts: 250
Credit: 826,541
RAC: 0
Message 23991 - Posted: 25 Jun 2012, 3:10:23 UTC

I have gotten 3 new Tasks int he past 15 minutes and they all run for about 15~25 seconds and then the completion bar goes to 100% and they report. When I look at my Tasks they indicate Completed and waiting for validation.

Is there something wrong or is this what everybody is seeing?

I'm concerned about running too many Tasks and having them not validate then getting thrown out of the New Task queue.
ID: 23991 · Report as offensive     Reply Quote
Profile Conan
Avatar

Send message
Joined: 6 Jul 06
Posts: 107
Credit: 511,942
RAC: 0
Message 23992 - Posted: 25 Jun 2012, 7:47:33 UTC

If the work units finish very quickly there are two main reasons this could happen; The Work Unit has an error, or the particle beam the Work Unit is trying to simulate crashed into the wall and did not make a circuit round the LHC (or similar analogy).

When you are waiting for validation then the Work Unit was not in error but has finished for whatever reason, and is just waiting to see if the a second run by another computer confirms your results.

I don't believe you have anything to worry about. Mixed in with the short ones are much longer ones.

Conan
ID: 23992 · Report as offensive     Reply Quote
Eric Mcintosh
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 12 Jul 11
Posts: 852
Credit: 1,619,050
RAC: 0
Message 23994 - Posted: 25 Jun 2012, 9:29:53 UTC

Yes; I am exploring phase space at rather high amplitudes
and I exoect all particles to be lost rtaher quickly at
these values. There are also cases which are much more stable
and will run longer. I am also comparing two different methods of
computing the beam-beam inetraction. Once I have identified
the more stable region there will be more longer cases to run.
Eric.
ID: 23994 · Report as offensive     Reply Quote
Tex1954

Send message
Joined: 24 Apr 11
Posts: 37
Credit: 1,105,291
RAC: 0
Message 24024 - Posted: 1 Jul 2012, 21:52:25 UTC - in response to Message 23994.  

Out of 1142 tasks since June 25th batch, I still have 560 pending. Looks like 98% of the pending are the short 2-5 second tasks.

Me thinks there is a problem when half the tasks run short and we never get credit... not even 0.01 points for the download/upload bandwidth used...

LOL!

8-)
ID: 24024 · Report as offensive     Reply Quote
Tex1954

Send message
Joined: 24 Apr 11
Posts: 37
Credit: 1,105,291
RAC: 0
Message 24032 - Posted: 2 Jul 2012, 23:08:47 UTC - in response to Message 24024.  

As an update, still got a few of the shorter tasks, but seems a lot more stable now running (typically) from 600 secs (10 min) to the absolute longest of 23,000 secs (6 hrs 22 min) on AMD 1100T/1090T systems running 1600MHz DDR3 @ 3.6GHz speed.

Looks like we got a winning batch here...

:)

ID: 24032 · Report as offensive     Reply Quote
Profile Tom95134

Send message
Joined: 4 May 07
Posts: 250
Credit: 826,541
RAC: 0
Message 24037 - Posted: 3 Jul 2012, 19:59:48 UTC

Tasks I am getting now are running full length.
ID: 24037 · Report as offensive     Reply Quote
Eric Mcintosh
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 12 Jul 11
Posts: 852
Credit: 1,619,050
RAC: 0
Message 24038 - Posted: 3 Jul 2012, 20:41:20 UTC - in response to Message 24037.  

Thanks for that feedback. I am now running ba detailed
comparison of two beam beam algortithms in a region of
reasonable stability.

We are not out of trouble yet but I am preparing 3 executables for
Linux and 3 for Windows and a colleague is producing a MAC exeutable.
The 3 versions are
Run anywhere
Use SSE2
Use SSE3
Eric.
ID: 24038 · Report as offensive     Reply Quote
Profile Tom95134

Send message
Joined: 4 May 07
Posts: 250
Credit: 826,541
RAC: 0
Message 24040 - Posted: 3 Jul 2012, 21:07:59 UTC - in response to Message 24037.  

Tasks I am getting now are running full length.

Well, I spoke too soon. Tasks seem to be running very short. Indicated times are for 14~15 hours but they finish in about 30 minutes (or less). They do seem to validate okay and there is no error showing up.
ID: 24040 · Report as offensive     Reply Quote
Harri Liljeroos
Avatar

Send message
Joined: 28 Sep 04
Posts: 442
Credit: 24,152,406
RAC: 14,344
Message 24041 - Posted: 3 Jul 2012, 21:19:11 UTC - in response to Message 24038.  

Thanks for that feedback. I am now running ba detailed
comparison of two beam beam algortithms in a region of
reasonable stability.

We are not out of trouble yet but I am preparing 3 executables for
Linux and 3 for Windows and a colleague is producing a MAC exeutable.
The 3 versions are
Run anywhere
Use SSE2
Use SSE3
Eric.


How do you recognize the different versions? I have seen only executables named Sixtrack.exe, current version is 443.07. All my computers are SSE3 capable so how do I know it was detected right?
ID: 24041 · Report as offensive     Reply Quote
Toby Broom
Volunteer moderator

Send message
Joined: 27 Sep 08
Posts: 598
Credit: 376,387,246
RAC: 44,972
Message 24042 - Posted: 3 Jul 2012, 21:48:14 UTC - in response to Message 24041.  

it say's 444.01 (sse3)
ID: 24042 · Report as offensive     Reply Quote
tullio

Send message
Joined: 19 Feb 08
Posts: 607
Credit: 3,807,439
RAC: 291
Message 24059 - Posted: 4 Jul 2012, 17:43:42 UTC

My Opteron 1210 at 1.8 GHz, SSE3 capable, is running a task which has taken 18+ hours to reach 44% on my Linux box. A wingman of mine with an AMD Phenom II running Windows has completed the task in two seconds. How is this possible? Is he using his Radeon HD 6900 graphic board?
Tullio
ID: 24059 · Report as offensive     Reply Quote
Toby Broom
Volunteer moderator

Send message
Joined: 27 Sep 08
Posts: 598
Credit: 376,387,246
RAC: 44,972
Message 24060 - Posted: 4 Jul 2012, 18:00:03 UTC - in response to Message 24059.  

Seems like your task isn't the new AMD optimised tasks. It will take a while for you to complete this task or you could abort it and get the new ones.
ID: 24060 · Report as offensive     Reply Quote
Filipe

Send message
Joined: 9 Aug 05
Posts: 35
Credit: 6,667,857
RAC: 50
Message 24061 - Posted: 4 Jul 2012, 18:04:45 UTC
Last modified: 4 Jul 2012, 18:06:48 UTC

when you report your task, it will be an inconclusive validation, and will be sent to a third host.

Probably some kind of error for this 2 second computation.

see this work unit for example.

No GPU computing in this project.

Filipe
ID: 24061 · Report as offensive     Reply Quote
Harri Liljeroos
Avatar

Send message
Joined: 28 Sep 04
Posts: 442
Credit: 24,152,406
RAC: 14,344
Message 24072 - Posted: 4 Jul 2012, 22:45:21 UTC
Last modified: 4 Jul 2012, 22:53:38 UTC

What are the necessary conditions to get your CPU capabilities recognized correctly? My Vista 32 bit with Q9400 + Boinc 6.10.58 does not list the SSE3 support which CPU-Z recognizes. Boinc messages list sse2 and ssse3 though but LHC still sends me the standard version of sixtrack.exe (444.01). Same happens with Win XP 32 bit, T9500 + Boinc 6.10.43.

My Win7 64 bit laptop with i7 + Boinc 6.12.34 and Win xp64 with E5645 + Boinc 6.12.34 get the correct sse3 application although boinc does not list sse3 available on them either.

So you should have at least 6.12.34 to get the right application?

Edit: correct signature and avatar
ID: 24072 · Report as offensive     Reply Quote
[AF>FAH-Addict.net]toTOW

Send message
Joined: 9 Oct 10
Posts: 77
Credit: 3,623,712
RAC: 0
Message 24074 - Posted: 4 Jul 2012, 23:12:18 UTC

Maybe your BOINC client is too old ?

Here is what I get on my i7 920 with 7.0.28 client :

04/07/2012 20:15:45 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 sse4_2 syscall nx lm vmx tm2 popcnt pbe
ID: 24074 · Report as offensive     Reply Quote
Toby Broom
Volunteer moderator

Send message
Joined: 27 Sep 08
Posts: 598
Credit: 376,387,246
RAC: 44,972
Message 24075 - Posted: 4 Jul 2012, 23:55:56 UTC

My reports similar to toTOW.

I don't know how BOINC decides what tasks to send.

It doesn't seem like the version is the problem?
ID: 24075 · Report as offensive     Reply Quote
Harri Liljeroos
Avatar

Send message
Joined: 28 Sep 04
Posts: 442
Credit: 24,152,406
RAC: 14,344
Message 24081 - Posted: 5 Jul 2012, 10:28:01 UTC - in response to Message 24074.  

Maybe your BOINC client is too old ?

Here is what I get on my i7 920 with 7.0.28 client :

04/07/2012 20:15:45 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 sse4_2 syscall nx lm vmx tm2 popcnt pbe


The message I see on my i7 is similar using boinc 6.12.34. And that gets the correct version of the application (sse3). Note that boinc detects ssse3 not sse3.

The Q9400 using boinc 6.10.58 gets this message:
Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 nx lm vmx smx tm2 pbe

and this computer gets only the standard version of sixtrack.exe.

And that is why I am asking what are the right conditions. Which OS and Boinc versions you should have to get it working right?

ID: 24081 · Report as offensive     Reply Quote
Toby Broom
Volunteer moderator

Send message
Joined: 27 Sep 08
Posts: 598
Credit: 376,387,246
RAC: 44,972
Message 24082 - Posted: 5 Jul 2012, 12:27:24 UTC

I always use the latest version of BOINC, I don't see any reason to not do that.

You said 6.12.x works fine so if you perfer an older vresion then upgrade to that?

Once you upgade to 7.x its tricky to drop back.
ID: 24082 · Report as offensive     Reply Quote

Message boards : Number crunching : Very Short Running on Tasks


©2020 CERN