Difference between revisions of "2012 Software Review Planning Meeting: Mar. 16, 2012"

From GlueXWiki
Jump to: navigation, search
m (Text replacement - "http://argus.phys.uregina.ca/cgi-bin/private" to "https://halldweb.jlab.org/doc-private")
 
(10 intermediate revisions by 3 users not shown)
Line 46: Line 46:
 
                     Best regards,    Rolf  
 
                     Best regards,    Rolf  
 
</pre>
 
</pre>
*:
 
 
* Time-line for Hall-D
 
* Time-line for Hall-D
 +
** April (1st week) 1 slide on tight spots and weaknesses, 1 slide on management plan
 
** April 23-27 - talk outlines.
 
** April 23-27 - talk outlines.
 
** April 30 - Written material for the review (CDR?).
 
** April 30 - Written material for the review (CDR?).
Line 57: Line 57:
 
**:
 
**:
 
* Organization of Hall-D preparations for Review
 
* Organization of Hall-D preparations for Review
** Introductory talk ([[Media:20120309_Cmeyer_Software_Review_preliminary.pdf|preliminary slides]])
+
** Introductory talk ([https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=1949 preliminary sildes]) with notes.
** [https://halldweb1.jlab.org/talks/2012-1Q/manpower_survey.pdf Software Manpower within collaboration] (no change) (Mark)
+
** [https://halldweb.jlab.org/talks/2012-1Q/manpower_survey.pdf Software Manpower within collaboration] (no change) (Mark)
** [[2012 Offline Computing Schedule]](no change) (David)
+
** [[2012 Offline Computing Schedule]] (David)
 
** Software Milestones/Challenges
 
** Software Milestones/Challenges
 
** Calibration/Alignment
 
** Calibration/Alignment
Line 67: Line 67:
 
=Minutes=
 
=Minutes=
  
Attendees:
+
Attendees: Mark Ito, David Lawrence and Curtis Meyer
  
 
== Status of Review Organization ==
 
== Status of Review Organization ==
  
 +
* Several potential reviewers have been identified. Bob is in the processes of contacting them.
 +
** Final schedule depends on chair's availability
 +
* Rolf laid out a rough schedule for preparations during the [https://halldweb.jlab.org/wiki/index.php/2012_Software_Review#Spring_2012_Cross-hall_Collaboration_Effort March 14th follow up meeting]. This is listed in the Agenda near the top of this page.
 +
** All Halls and DAQ group will participate in these meetings.
  
 
== Cross-Hall Collaboration ==
 
== Cross-Hall Collaboration ==
 +
 +
We discussed only briefly the follow-up meeting on Wednesday where [https://halldweb.jlab.org/wiki/index.php/2012_Software_Review#Spring_2012_Cross-hall_Collaboration_Effort cross-hall collaborative efforts] were discussed. Mark, Curtis and I were all at that meeting so not a lot of discussion was needed.
 +
 +
The meeting also included information on the upcoming Software Review which is given in the above.
 +
  
 
== Organization of Hall-D preparations for Review ==
 
== Organization of Hall-D preparations for Review ==
  
 
=== Introductory talk ===
 
=== Introductory talk ===
 +
*We went over the introductory talk which we felt should be fairly graphically orientated.
 +
Our current feeling is that the slide should include the following:
 +
# Overview of Collaboration/Hall D software management.
 +
# Overview (graphically) of what the software needs to do with rough data volumes and possibly estimates of compute resources.
 +
# An "event display" showing a b1pi event and discussing what we need to have to do amplitude analysis.
 +
# A overview of the 3pi analysis done by Jake showing that we have carried out a full analysis from event generators and background generators through GEANT on the grid, actual reconstruction of the events, physics cuts and then an amplitude analysis that could have been done on a GPU.
 +
# A list of what we feel still needs to be done to have us ready for data analysis in 2015.
 +
  
 
=== Software Manpower within collaboration ===
 
=== Software Manpower within collaboration ===
 +
 +
''No Changes''
 +
  
 
=== Offline Computing Schedule ===
 
=== Offline Computing Schedule ===
 +
 +
* David pointed out some updates to the computing activity schedule. Specifically:
 +
** The overall tracking is going to be increased from 4.5 FTE-years to 7 FTE-years
 +
** The percent complete will remain at about 70%
 +
* Curtis noted that the PWA number should include the work done at CMU which adds another 3 FTE-years
 +
  
 
===Software Milestones/Challenges===
 
===Software Milestones/Challenges===
 +
* Curtis has written a one-page summary [https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=1950 document] describing the experience in CLAS in moving large data sets from Jefferson Lab to remote site for analysis.
 +
  
 
=== Calibration/Alignment ===
 
=== Calibration/Alignment ===
 +
Discussions have begun in regular weekly and biweekly GlueX meetings. Plans are expected to start emerging as documents over the next several weeks.
 +
  
 
=== Misc. ===
 
=== Misc. ===
 +
==== Computing resources ====
 +
* Curtis will contact members of the collaboration to try and get an estimate of the current computing resources that are available outside of Jefferson Lab for GlueX. The example from CMU is
 +
<pre>
 +
At CMU: These resources are split with our lattice QCD group with whom
 +
we share operations of the cluster. All CPUs are roughly 2.4GhZ nodes and
 +
all are under three years old at the moment.
 +
 +
32      dual-quad core AMD machines      ->  256 compute cores 1GB core per CPU
 +
12      quad-eight-core AMD machines    ->  384 compute cores 2GB core per CPU
 +
180TB of RAID storage spread across several servers
 +
10Giga-bit backbone from the RAID servers to the switches and 2Giga-bit connections
 +
to each compute node.
 +
 +
 +
Over the next few years, we anticipate purchasing around 16 boxes of the next
 +
generation of CPUs (quad-sixteen or thrirty-two core boxes). We also expect to
 +
double our disk storage capability as we start to retire some of our very old few-TB
 +
servers.
 +
</pre>
  
 +
==== Statistics Group ====
 +
* We discussed the possibility of having a "Statistics Group" within the collaboration. This
 +
is done in many larger collaborations. Mike Williams should be familiar with this and might
 +
be a good person to lead such a group within GlueX. At some point in the future we will
 +
ask him to give a talk at our weekly physics meeting.
  
 
=Action Items=
 
=Action Items=
 +
# Make list of "tight spots" and weaknesses/concerns for end of March Meeting (David)
 +
# Adapt Curtis's slide 2 into 1 page management plan for end of March meeting (David)
 +
# Summary of GRIDFTP experience for transferring data outside JLab (Curtis)
 +
# Contact Richard about 1-pager for simulation on the GRID (David)
 +
# Event display pictures (both ded and hdview2). (David)
 +
# Add slide to Curtis' intro talk summarizing Jake's 3pi exercise (Curtis)
 +
# Increase the PWA activity to 6 FTE-years to include CMU work (David)
 +
# Request list of computing resources now and in 3 years from collaboration (Curtis)

Latest revision as of 16:54, 24 February 2017

Main Hall-D Software Review Page


Connecting

Phone:

  • +1-866-740-1260 : US+Canada
  • +1-303-248-0285 : International
  • then enter participant code: 3421244# (remember the "#").
  • or www.readytalk.com (and code without the #)

Agenda

Dear all,

    Thanks for the work to come to concise summaries on our
previous self-assigned actions. I think this has been a very
useful communication and collaboration between the four Halls
and other interested parties. Very good!

    Please find attached an example charge as given to me by
Chip Watson, and once more the earlier document on planning for
the software review.

    Here's our further tentative planning, assuming a software
review June 7 + 8:
    Between March 29 and April 4:
        Followup preparatory meeting to discuss
    1) few slides intro on scientific computing outlook by 2015
    2) each Hall one slide on anticipated weaknesses/tight spots
        and one slide on planned management structure

    Week of April 23-27:
        Initial Dry Run: Talk Outlines
            (what you plan to present in each slide)

    May 14 or May 15:
        First Dry Run: Near-complete presentations

    Period of May 24, 25, 29, 30 (dates to be determined):
        Final Dry Run: Timed presentations

                    Best regards,    Rolf 
  • Time-line for Hall-D
    • April (1st week) 1 slide on tight spots and weaknesses, 1 slide on management plan
    • April 23-27 - talk outlines.
    • April 30 - Written material for the review (CDR?).
    • May 14/15 - First draft of presentations for dry run.
    • May 24/25 or 29/30 - Final talks with timed presentations.
    • June7/8 Software Review at JLab.
  • Cross-Hall Collaboration
  • Organization of Hall-D preparations for Review

Minutes

Attendees: Mark Ito, David Lawrence and Curtis Meyer

Status of Review Organization

  • Several potential reviewers have been identified. Bob is in the processes of contacting them.
    • Final schedule depends on chair's availability
  • Rolf laid out a rough schedule for preparations during the March 14th follow up meeting. This is listed in the Agenda near the top of this page.
    • All Halls and DAQ group will participate in these meetings.

Cross-Hall Collaboration

We discussed only briefly the follow-up meeting on Wednesday where cross-hall collaborative efforts were discussed. Mark, Curtis and I were all at that meeting so not a lot of discussion was needed.

The meeting also included information on the upcoming Software Review which is given in the above.


Organization of Hall-D preparations for Review

Introductory talk

  • We went over the introductory talk which we felt should be fairly graphically orientated.

Our current feeling is that the slide should include the following:

  1. Overview of Collaboration/Hall D software management.
  2. Overview (graphically) of what the software needs to do with rough data volumes and possibly estimates of compute resources.
  3. An "event display" showing a b1pi event and discussing what we need to have to do amplitude analysis.
  4. A overview of the 3pi analysis done by Jake showing that we have carried out a full analysis from event generators and background generators through GEANT on the grid, actual reconstruction of the events, physics cuts and then an amplitude analysis that could have been done on a GPU.
  5. A list of what we feel still needs to be done to have us ready for data analysis in 2015.


Software Manpower within collaboration

No Changes


Offline Computing Schedule

  • David pointed out some updates to the computing activity schedule. Specifically:
    • The overall tracking is going to be increased from 4.5 FTE-years to 7 FTE-years
    • The percent complete will remain at about 70%
  • Curtis noted that the PWA number should include the work done at CMU which adds another 3 FTE-years


Software Milestones/Challenges

  • Curtis has written a one-page summary document describing the experience in CLAS in moving large data sets from Jefferson Lab to remote site for analysis.


Calibration/Alignment

Discussions have begun in regular weekly and biweekly GlueX meetings. Plans are expected to start emerging as documents over the next several weeks.


Misc.

Computing resources

  • Curtis will contact members of the collaboration to try and get an estimate of the current computing resources that are available outside of Jefferson Lab for GlueX. The example from CMU is
At CMU: These resources are split with our lattice QCD group with whom
we share operations of the cluster. All CPUs are roughly 2.4GhZ nodes and
all are under three years old at the moment.

32       dual-quad core AMD machines      ->  256 compute cores 1GB core per CPU
12       quad-eight-core AMD machines    ->   384 compute cores 2GB core per CPU
180TB of RAID storage spread across several servers
10Giga-bit backbone from the RAID servers to the switches and 2Giga-bit connections
to each compute node.


Over the next few years, we anticipate purchasing around 16 boxes of the next
generation of CPUs (quad-sixteen or thrirty-two core boxes). We also expect to
double our disk storage capability as we start to retire some of our very old few-TB
servers. 

Statistics Group

  • We discussed the possibility of having a "Statistics Group" within the collaboration. This

is done in many larger collaborations. Mike Williams should be familiar with this and might be a good person to lead such a group within GlueX. At some point in the future we will ask him to give a talk at our weekly physics meeting.

Action Items

  1. Make list of "tight spots" and weaknesses/concerns for end of March Meeting (David)
  2. Adapt Curtis's slide 2 into 1 page management plan for end of March meeting (David)
  3. Summary of GRIDFTP experience for transferring data outside JLab (Curtis)
  4. Contact Richard about 1-pager for simulation on the GRID (David)
  5. Event display pictures (both ded and hdview2). (David)
  6. Add slide to Curtis' intro talk summarizing Jake's 3pi exercise (Curtis)
  7. Increase the PWA activity to 6 FTE-years to include CMU work (David)
  8. Request list of computing resources now and in 3 years from collaboration (Curtis)