Skip to Content.
Sympa Menu

opal - Re: [Opal] OPAL-1.6 cyclotron tracker multicores problem

opal AT lists.psi.ch

Subject: The OPAL Discussion Forum

List archive

Re: [Opal] OPAL-1.6 cyclotron tracker multicores problem


Chronological Thread 
  • From: "Adelmann Andreas (PSI)" <andreas.adelmann AT psi.ch>
  • To: "Adelmann Andreas (PSI)" <andreas.adelmann AT psi.ch>
  • Cc: Chuan Wang <wangchuan02513 AT 163.com>, "opal AT lists.psi.ch" <Opal AT lists.psi.ch>
  • Subject: Re: [Opal] OPAL-1.6 cyclotron tracker multicores problem
  • Date: Fri, 8 Jun 2018 12:14:34 +0000
  • Accept-language: en-US, de-CH
  • Authentication-results: localhost; dmarc=skipped

I mean parallel file system … 
------
Dr. sc. math. Andreas (Andy) Adelmann
Staff Scientist - Deputy Labor Head LSM
Paul Scherrer Institut OHSA/D17  CH-5232 Villigen PSI
Phone Office: xx41 56 310 42 33 Fax: xx41 56 310 31 91
Phone Home: xx41 62 891 91 44
-------------------------------------------------------
Friday: ETH HPK G 28   +41 44 632 75 22
============================================
The more exotic, the more abstract the knowledge, 
the more profound will be its consequences.
Leon Lederman 
============================================





On 8 Jun 2018, at 21:05, Adelmann Andreas (PSI) <andreas.adelmann AT psi.ch> wrote:

Well usually you should have a pallet file system such as Lustre or GPFS.
In you case you only read, however maybe with 32 cores this poses already problem.
I am not the expert here maybe Achim could comment.

A
------
Dr. sc. math. Andreas (Andy) Adelmann
Staff Scientist - Deputy Labor Head LSM
Paul Scherrer Institut OHSA/D17  CH-5232 Villigen PSI
Phone Office: xx41 56 310 42 33 Fax: xx41 56 310 31 91
Phone Home: xx41 62 891 91 44
-------------------------------------------------------
Friday: ETH HPK G 28   +41 44 632 75 22
============================================
The more exotic, the more abstract the knowledge, 
the more profound will be its consequences.
Leon Lederman 
============================================





On 8 Jun 2018, at 20:19, Adelmann Andreas (PSI) <andreas.adelmann AT psi.ch> wrote:

To me this is a file-system or H5 problem. 
What type of file system are you using?

Cheers Andreas 
------
Dr. sc. math. Andreas (Andy) Adelmann
Staff Scientist - Deputy Labor Head LSM
Paul Scherrer Institut OHSA/D17  CH-5232 Villigen PSI
Phone Office: xx41 56 310 42 33 Fax: xx41 56 310 31 91
Phone Home: xx41 62 891 91 44
-------------------------------------------------------
Friday: ETH HPK G 28   +41 44 632 75 22
============================================
The more exotic, the more abstract the knowledge, 
the more profound will be its consequences.
Leon Lederman 
============================================





On 8 Jun 2018, at 09:56, wangchuan02513 <wangchuan02513 AT 163.com> wrote:

Dear, Colleagues,
My name is Chuan Wang from China Insitute of Atomic Energy. Currently I am testing the parallel efficiency of the new cluster of our cyclotron group with OPAL cyclotron tracker and the input file is per attached. When I test the  case with 4 cores and 16 cores it seems OK, however when I test it with 32 cores (16 cores/cpu, 2cpus in one node), I got error messages which is also per attached.
Does anyone have a clue on that?
Thank you for your help.

--
Sincerely yours
Chuan Wang

e-mail  : cwang AT ciae.ac.cn
              wangchuan02513 AT 163.com 

Tel     : 0086-10-69357590(O) 
MP    : 0086-10-18600605915
Addr. : P.O.BOX. 275(3), Beijing, China 
            China Institute of Atmoic Energy
            Beijing Radioactive Ion Facility 

PostCode: 102413


 
<efieldNoSC.in><screen_4core.dat><screen_32core.dat>






Archive powered by MHonArc 2.6.19.

Top of Page