Index  | Recent Threads  | Unanswered Threads  | Who's Active  | Guidelines  | Search
 

Quick Go »
No member browsing this thread
Thread Status: Active
Thread Type: Sticky Thread
Total posts in this thread: 1840
Posts: 1840   Pages: 184   [ Previous Page | 16 17 18 19 20 21 22 23 24 25 | Next Page ]
[ Jump to Last Post ]
Post new Thread
Author
Previous Thread This topic has been viewed 2121811 times and has 1839 replies Next Thread
Dataman
Ace Cruncher
Joined: Nov 16, 2004
Post Count: 4865
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Research Log: Updates from the Harvard Team

Cleanenergy now has a gold badge too ... applause
----------------------------------------


[Feb 25, 2011 7:16:40 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Research Log: Updates from the Harvard Team

Guys, I've got the question.

I'd like to contribute to "Clean Energy Project" (CEP) and internet traffic is not a problem at all. There's another issue...

In order to increase productivity I'm using winAFC tool to arrange CPU affinity and priority for BOINC processes to "above normal" or "high" level. But CEP creates two processes what messed up everything: two processes sharing one CPU core while another CPU core is idle. This slow down CEP and another process.

I wonder if I can do smth about that or there are any plans to release new app to fix that. Sorry for stupid question...

and BTW, how I can add signature to my posts? I can find it in "my profile" blushing
----------------------------------------
[Edit 2 times, last edit by Former Member at Feb 26, 2011 4:41:35 PM]
[Feb 26, 2011 4:32:41 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Research Log: Updates from the Harvard Team

That is strange, it should have put one project on each CPU. Are you getting more than one download at a time of the CEP? I know there is a setting you have to change to be able to download more than one task at a time for both your CPU's to use.

I was kind of under the impression that CEP would use what you had for the most part until another task came along and took some of the CPU power. Work your way up slowly, if you give it too high a priority you can lock your system up.

Aaron
[Feb 26, 2011 11:47:17 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Research Log: Updates from the Harvard Team

winAFC doing the stuff exactly right - one process per one CPU core. But the problem is that CEP starting two processes and that confusing winAFC crying any other project doing it right...

OK, I'll try to get CEP task and I'll bring screenshot to show what I'm talking about.

and BTW - no such issue on linux. that's weird...
[Feb 27, 2011 2:15:10 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Research Log: Updates from the Harvard Team

Hello CTAPbIi,
Yes, CEP2 has always had 2 processes.The main process is wcgrid_cep2_qchem_6.35_windows_intel86 and the secondary process is wcgrid_cep2_6.35_windows_intel86

I have never checked this carefully but I have the impression that the secondary process is mostly setting up the parameters for qchem. I cannot remember ever having seen the secondary process running when I glanced at Task Manager.

Lawrence
[Feb 27, 2011 8:21:05 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Research Log: Updates from the Harvard Team

that's exactly right - there are 2 processes and I can see them in task manager. While I'm using winAFC with mask "*wcg" it messing everything around... I wonder if there's some ideas what should I do...
[Feb 28, 2011 1:57:16 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Research Log: Updates from the Harvard Team

Alan was just named a Big Think Delphi Fellow.
http://bigthink.com/experts
[Feb 28, 2011 1:54:53 PM]   Link   Report threatening or abusive post: please login first  Go to top 
armstrdj
Former World Community Grid Tech
Joined: Oct 21, 2004
Post Count: 695
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Research Log: Updates from the Harvard Team

CEP2 is currently the only application running on World Community Grid which is a compound application. This means that there is a wrapper process which initializes the BOINC environment and then launches a separate process for the science application. The wrapper process runs for the entire task and should use very little CPU time. A new science application process is started by the wrapper for each new science job, 0 - 14, in the task. These are the processes which use the CPU time.

Thanks,
armstrdj
[Feb 28, 2011 3:14:53 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Research Log: Updates from the Harvard Team

that's exactly right - there are 2 processes and I can see them in task manager. While I'm using winAFC with mask "*wcg" it messing everything around... I wonder if there's some ideas what should I do...

Not heard of this one before, but with this tool and my favoured Process Lasso, you can end up with more than 1 science running off 1 core when more processor cores are specified in BOINC than these affinity associating tools are set to use. In the experiment I told PL to assign anything with the HPF2 application mask to cores 2+3. When 4 were running concurrently on BOINC, 2 per core were running and 2 cores were sitting idle. With CEP2 you can set BOINC to never have more than 2 in cache and thus no more than 2 running concurrent and then specify for those the specific cores, but at least 2 cores marked. Then the OS and BOINC itself will take care of the other sciences to utilize the free threads, to include cores 2-3 in the example, else these might end up idling.

--//--
[Feb 28, 2011 3:46:19 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Research Log: Updates from the Harvard Team

CEP2 is currently the only application running on World Community Grid which is a compound application. This means that there is a wrapper process which initializes the BOINC environment and then launches a separate process for the science application. The wrapper process runs for the entire task and should use very little CPU time. A new science application process is started by the wrapper for each new science job, 0 - 14, in the task. These are the processes which use the CPU time.

Thanks,
armstrdj

no complains about that :-) the only thing is that the 2nd process messing everything up...

Not heard of this one before, but with this tool and my favoured Process Lasso, you can end up with more than 1 science running off 1 core when more processor cores are specified in BOINC than these affinity associating tools are set to use. In the experiment I told PL to assign anything with the HPF2 application mask to cores 2+3. When 4 were running concurrently on BOINC, 2 per core were running and 2 cores were sitting idle. With CEP2 you can set BOINC to never have more than 2 in cache and thus no more than 2 running concurrent and then specify for those the specific cores, but at least 2 cores marked. Then the OS and BOINC itself will take care of the other sciences to utilize the free threads, to include cores 2-3 in the example, else these might end up idling.

--//--

nice piece of software, I never heard about it :-) The only thing is I can not understand how I can assign wcg process to a single CPU core, i.e. mask is "0;1" for both of them, but not "0" for the 1st process and "1" for the 2nd one. LS does not allow me to create two different notes. Priority changing works fine though.

But your idea to assign a core to "HPF2 " is brilliant :-) I'll try it soon. But how I can limit BOINC to have two CEP2 tasks in cache?
[Feb 28, 2011 9:18:23 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Posts: 1840   Pages: 184   [ Previous Page | 16 17 18 19 20 21 22 23 24 25 | Next Page ]
[ Jump to Last Post ]
Post new Thread