Workfest Goals

Develop the Tools for CDR Simulations HDFast HDGEANT Geometry Definitions Remote Access

Education of the rest of the collaboration Needs for CDR Data Model Other Items

Software Web Page Distributed Meetings Long Term Simulation Effort – Goals & Design

Workfest Minutes

Speaker – Richard Jones Subject - Status of GEANT

10 Months until the CDR is due

Software Design Create PublishMonitor Development

HDFAST

MCFAST fairly mature (monitoring development). Reasonably stable at this point.

GEANT 3 Designed and in prototype How to compare with MCFAST

GEANT 4 Forseen, probably have a compatible geometry

definition (GEANT4 might change). Event Generators cwrap, genR8, weight Facilities (JLab, Regina, IU, FSU and UConn.) Expected Simulation Projects – PWA, Detector

Optimizations, Background Studies

HDFAST – I/O Summary

Genr8 ascii (ascii2stdhep) stdhepCommand & Geometry HDFast Root(RDT) or stdhep

asciiUsing Root (Root Data Tree)Root available at http://root.cern.ch

UConn Cluster

Pentium 450/800 – 36 processors Rackmount (2U cases)

Dual CPU – 512 MByte

Switch

Dual CPU’s

Disk

pvfs

Disk

raid

80 Mb/s 35 Mb/s

I n t e r n e t 2

...Nortel 450

dual Pentium IIIcompute nodes

100Mb/s

Raidserver

PVFSserver

massstorage

nodes

650GB

350GB

UConnPhysicsBeowulfCluster

UConn100Mb/s

ESNet

JLab

I.U. HPSS

150 TBnear-line storage

MSS250 TB

long-term archive

U Conn Computing Cluster

Mantrid – Indiana University

Processing/Storage ModelD

32 processors in 16 nodes 32 45 GB disks (1.44 TB

total)

/home/HallD

Node 00 /data0/HallD /data1/HallD

Has slides Has a prototype web based

access system for event generation (cwrap).

http://anthrax.physics.indiana.edu/~teige/tests/cwrap_request.html

U Regina – 50 Alpha Cluster

50 Nodes

PBS 500 MHz Alphas

9 Gbyte disk per node Note a 500 MHz alphas is

roughly comparable to a 1 GHz Pentium III.

Has slides See: www.phys.uregina.ca

/~brash/openspace.ps

www.phys.uregina.ca/~brash/connectivity.ps

Second Day: To Do

Running HDFAST and HDGEANT. Paul gave a demo.

Decide on data model how information is moved from package to package. (Data Model Working Group)

Web Site (Working Group) Features required to integrate clusters

Hall D Computing Web Site

Goals: Everyone can contribute without

excessive site management. XML based description of documents. Automatic searching and organization

tools. Still need overview documents.

Contents: Hall D Website

Everyone maintains their own site Everyone has a summary page and link to

Hall D computing resources and searching tools Links and searches will be managed automatically

Everyone contributes documents to the Hall D computing archive describing their computing activities

Each document has an XML metadata description of what it contains

How To: Hall D Website

Within your website create a single XML metadata document describing all of your documents.

Let me know where it is. Publish DTD so local sites can be

validated (http://comphy.fsu.edu/~dennisl/halld/dtds/website.dtd).

How To: Hall D Website

<?xml version="1.0"?><documents><document location='http://comphy.fsu.edu/~dennisl/halld/computing' version='1‘

format=“html”> <altdocument location=“design.pdf” format=“pdf” type=“relative”/> <title>Hall D Computing Design Page</title> <topic>Design</topic> <author email=“dennisl@phy.fsu.edu”>Larry Dennis</author> <date>May 21, 2001</date> <keyword>grid</keyword> <keyword>computing</keyword> <keyword>acquisition</keyword> <keyword>analysis</keyword>

<keyword>simulations</keyword> <abstract>

This is final word on Hall D computing. </abstract></document> <document … repeat as needed … </document></documents>

Geant I/O Package

Binary Stream Binary StreamEvents In Events Out

Control In stderr

stdout Log

metadata

GEANT 3 – Richard’s Plan

Produce a standard geometrySee

http://zeus.phys.uconn.edu/halld/geometry

Use the geometry for Monte Carlo, event display, logical geometry model for use in analysis.

Monte Carlo Data Model - Input

<run> <event <interaction <vertex … <particle … </event> …</run>

Conceptual Model Logical

Model

Physical Model OpenStart with an I/O APISome others exist

Monte Carlo Data Model - Output

… <detector <BarrelDC <ring <sector <strawhit

<eloss <time …

Conceptual Model

LogicalModel

Physical Model OpenStart with an I/O API

Monte CarloEvent Generator Interactions Simulation

Real dataDAC Digitized DataCalibration Translator

Hits:

DOE/NSF Initiatives & Resources

Groups Working on Software

CMUU ConnU ReginaFSUIUFIUJLab (Watson, Bird, Heyes, Hall D)

RPIODUGlasgow

Raw events

hits

Tracks/clusters

Particles

How much of thiscan be automated?

Larry -- Things To Do

Give everyone information about ITR and SciDAC

Get Web Site Started Design (Elliott, Scott) Prototype Grid Nodes (Ian, Elliott)

Richard -- Things To Do

Input Interface to GEANT from event generators, XML input.

Finish Geometry Prototype Output Interface for GEANT Prototype Document and Publish the above

Scott -- Things To Do

Web access for Mantrid Interfaces from generators to/from

XML

Paul -- Things To Do

Maintain HDFast Teach people how to use HDFAST

Greg -- Things To Do

DTD for event structure DTD for cwrap input

Ed -- Things To Do

Full OSF support for genr8, HDFAST, GEANT3, translators

Web Interface for UR Farm Barrel Calorimeter Studies with

GEANT3

Elliott -- Things To Do

Explore CODA Event Format Assist Greg with Event DTD Explore GEANT4 Hall D Computing Design Prototype for Grid Nodes Remote Collaboration Tools

Design Focus

Get the job done Minimize the effort required to

perform computing Fewer physicists Lower development costs Lower hardware costs Keep it simple

Provide for ubiquitous access and participation – improve participation in computing

Goals for the Computing Environment

1. The experiment must be easy to conduct (coded software peopletwo person rule).

2. Everyone can participate in solving experimental problems – no matter where they are located.

3. Offline analysis can more than keep up with the online acquisition.

4. Simulations can more than keep up with the online acquisition.

5. Production of tracks/clusters from raw data and simulations can be planned, conducted, monitored, validated and used by a group.

6. Production of tracks/clusters from raw data and simulations can be conducted automatically with group monitoring.

7. Subsequent analysis can be done automatically if individuals so choose.

Goal #1: Easy to Operate

100 MB/s raw data. Need an estimate of designed good event rate to set online trigger performance

Automated system monitoring Automated slow controls Automated data acquisition Automated online farm Collaborative environment for access to experts Integrated problem solving database links current

to past problems and solutions Well defined procedures Good training procedures

Goal #2: Ubiquitous expert participation

Online system information available from the web.

Collaborative environment for working with online team.

Experts can control systems from elsewhere when data acquisition team allows or DAQ inactive.

Goal #3: Concurrent Offline Production

Offline Production (raw events tracks/clusters) can be completed in the same length of time as is required for data taking (including detector and accelerator down time). This includes: Calibration overhead. Multiple passes through the data (average of

2). Evaluation of results. Dissemination of results

Goal #4: Concurrent Simulations

Simulations can be completed in the same length of time as is required for data taking (including detector and accelerator down time). This includes: Simulation planning. Systematic studies ( up to 5-10 times as

much data as is required for experimental measurements).

Production processing of simulation results. Dissemination of results.

Goal #5: Collaborative computing

Production processing and simulations can be planned by a group.

Multiple people can conduct, validate, monitor, evaluate and use produced data and simulations without unnecessary duplication.

A single individual or a large group can manage appropriate scale tasks effectively.

Goal #6: Automated computing

Production processing and simulations can conducted automatically without intervention.

Progress is reported automatically. Quality checking can be performed

automatically. Errors in automatic processing are

automatically flagged.

Goal #7: Extensibility

Subsequent analysis steps can be done automatically if individuals so choose.

The computational management system can be extended to include any Hall D computing tasks.