areaDetector Developments

Slides:



Advertisements
Similar presentations
NGAS – The Next Generation Archive System Jens Knudstrup NGAS The Next Generation Archive System.
Advertisements

INTRODUCTION TO SIMULATION WITH OMNET++ José Daniel García Sánchez ARCOS Group – University Carlos III of Madrid.
Operating System.
A PLFS Plugin for HDF5 for Improved I/O Performance and Analysis Kshitij Mehta 1, John Bent 2, Aaron Torres 3, Gary Grider 3, Edgar Gabriel 1 1 University.
A new Network Concept for transporting and storing digital video…………
File Systems.
AreaDetector Data Processing Pipeline In EPICS V4 Dave Hickin Diamond Light Source EPICS Collaboration Meeting Diamond Light Source 01/05/2013.
Real-Time Video Analysis on an Embedded Smart Camera for Traffic Surveillance Presenter: Yu-Wei Fan.
CHEP04 - Interlaken - Sep. 27th - Oct. 1st 2004T. M. Steinbeck for the Alice Collaboration1/27 A Control Software for the ALICE High Level Trigger Timm.
1 Lecture 1  Getting ready to program  Hardware Model  Software Model  Programming Languages  The C Language  Software Engineering  Programming.
SRW File Transfer Featuring SRW-5800 w/HKSR-5804.
FLANN Fast Library for Approximate Nearest Neighbors
GigE Knowledge. BODE, Company Profile Page: 2 Table of contents  GigE Benefits  Network Card and Jumbo Frames  Camera - IP address obtainment  Multi.
EtherCAT Driver for Remote I/O James Rowland, Ronaldo Mercado and Nick Rees.
AreaDetector and GStreamer Similarities and Differences Tom Cobb.
1 Input/Output. 2 Principles of I/O Hardware Some typical device, network, and data base rates.
Data Acquisition at the NSLS II Leo Dalesio, (NSLS II control group) Oct 22, 2014 (not 2010)
Operating Systems Operating System
A. Homs, BLISS Day Out – 15 Jan 2007 CCD detectors: spying with the Espia D. Fernandez A. Homs M. Perez C. Guilloud M. Papillon V. Rey V. A. Sole.
A Streaming Video Server for areaDetector Tom Cobb Presented by Nick Rees.
[1] Reference: QCam API reference manual document version Charge Coupled Device (CCD)
Parallel HDF5 Introductory Tutorial May 19, 2008 Kent Yang The HDF Group 5/19/20081SCICOMP 14 Tutorial.
Operating System. Architecture of Computer System Hardware Operating System (OS) Programming Language (e.g. PASCAL) Application Programs (e.g. WORD, EXCEL)
1 High level view of HDF5 Data structures and library HDF Summit Boeing Seattle September 19, 2006.
1.Overview 2. Hardware 3. Software Interface 4. Triggering 5. Installation 6. Configuring.
A Metadata Based Approach For Supporting Subsetting Queries Over Parallel HDF5 Datasets Vignesh Santhanagopalan Graduate Student Department Of CSE.
February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.
GumTree Feature Overview Tony Lam Data Acquisition Team Bragg Institute eScience Workshop 2006.
Y. Kotani · F. Ino · K. Hagihara Springer Science + Business Media B.V Reporter: 李長霖.
An Introduction to Linux Name: Haixin Wang ID :
CCGrid 2014 Improving I/O Throughput of Scientific Applications using Transparent Parallel Compression Tekin Bicer, Jian Yin and Gagan Agrawal Ohio State.
File System Implementation Chapter 12. File system Organization Application programs Application programs Logical file system Logical file system manages.
ATCA based LLRF system design review DESY Control servers for ATCA based LLRF system Piotr Pucyk - DESY, Warsaw University of Technology Jaroslaw.
Detectors for Light Sources Contribution to the eXtreme Data Workshop of Nicola Tartoni Diamond Light Source.
Slide: 1 NeXus and Synchrotrons: Challenges and Requirements V.A. Solé – ESRF Software Group NeXus Data Format Workshop, PSI, May
Grid Computing at Yahoo! Sameer Paranjpye Mahadev Konar Yahoo!
CCGrid 2014 Improving I/O Throughput of Scientific Applications using Transparent Parallel Compression Tekin Bicer, Jian Yin and Gagan Agrawal Ohio State.
Next Generation Operating Systems Zeljko Susnjar, Cisco CTG June 2015.
“Live” Tomographic Reconstructions Alun Ashton Mark Basham.
May 2003National Coastal Data Development Center Brief Introduction Two components Data Exchange Infrastructure (DEI) Spatial Data Model (SDM) Together,
AreaDetector workshop Summary and notes…. Overview iocBuilder File Writer Changes areaDetector driver pattern shared vector Asyn client without an IOC.
Modeling Big Data Execution speed limited by: –Model complexity –Software Efficiency –Spatial and temporal extent and resolution –Data size & access speed.
Connections to Other Packages The Cactus Team Albert Einstein Institute
High Speed Detectors at Diamond Nick Rees. A few words about HDF5 PSI and Dectris held a workshop in May 2012 which identified issues with HDF5: –HDF5.
UNIX & Windows NT Name: Jing Bai ID: Date:8/28/00.
August 2003 At A Glance The IRC is a platform independent, extensible, and adaptive framework that provides robust, interactive, and distributed control.
Chapter 9: Networking with Unix and Linux. Objectives: Describe the origins and history of the UNIX operating system Identify similarities and differences.
With TANGO S. Poirier – Data management group.
Damian Gordon.  This module will serve as an introduction to Operating Systems.  It provides an overview of the major components of a computer system.
CEG 2400 FALL 2012 Windows Servers Network Operating Systems.
NI Big Physics Summit CERN
INTRODUCTION TO HADOOP. OUTLINE  What is Hadoop  The core of Hadoop  Structure of Hadoop Distributed File System  Structure of MapReduce Framework.
Seminar On Rain Technology
NetCDF Data Model Details Russ Rew, UCAR Unidata NetCDF 2009 Workshop
AreaDetector: What’s New? Mark Rivers GeoSoilEnviroCARS, Advanced Photon Source University of Chicago.
ORNL is managed by UT-Battelle for the US Department of Energy ADnED Handling V4 Neutron Event Data Matt Pearson SNS ORNL 18 th -22 nd May 2015.
Parallel Virtual File System (PVFS) a.k.a. OrangeFS
Nat 4/5 Computing Science Software
Modeling Big Data Execution speed limited by: Model complexity
Operating System.
Tango Integration of Modern 2D Detectors
Introduction to HDF5 Session Five Reading & Writing Raw Data Values
Spark Presentation.
HDF5 Metadata and Page Buffering
Chapter 3: Operating-System Structures
Constructing a system with multiple computers or processors
Lecture 15 Reading: Bacon 7.6, 7.7
Outline Operating System Organization Operating System Examples
COMP755 Advanced Operating Systems
Types of Parallel Computers
Presentation transcript:

areaDetector Developments Ulrik Pedersen, Jon Thompson, Tom Cobb and Nick Rees

Area Detector Developments Introduction History Driver developments PCO cameras. Medipix 3 camera. Plugin developments HDF5 NDFilePlugin NDFilePlugin modifications adOpenCV 13 June 2011 Area Detector Developments

Area Detector Developments History Originally, all Diamond area detector data was to be handled by our upper level acquisition software (GDA). However, speed wasn’t great (Java) and there was no consistent framework. Consequently, GDA team selected EPICS area detector as their detector framework. Now the EPICS team has taken this over and most of our new development is creating areaDetector plugins. Big advantage is that it has three data streams: Compressed video via ffmpeg for interactive visualisation Uncompressed frame data via CA for on-the-fly analysis High speed data streaming to disk 13 June 2011 Area Detector Developments

Area Detector Developments Driver developments 13 June 2011 Area Detector Developments

Area Detector Developments PCO Camera’s PCO make a large range of Science Grade CCD and CMOS detectors. Diamond has PCO1600: 1.6k x 1.2k x 14 bit CCD. PCO4000: 4k x 2.6k x 14 bit CCD with 4 Gbyte frame memory PCO dimax: 2k x 2k x 12 bit x 1300 fps CMOS with 36 Gbyte frame memory. All come with a Windows API and Silicon Software CameraLink frame grabber cards. Driver development at SLAC and APS as well as at Diamond. Have Windows drivers for all cameras now, but still adding features. Plan to use Silicon Software Linux drivers to create Linux areaDetector driver, so we can get better write performance. There aren’t any high speed clustered Windows file systems. 13 June 2011 Area Detector Developments

Area Detector Developments Excalibur Excalibur is an internal Diamond/STFC development of the high speed direct detection Medipix 3 system. Sensor array is 8x6 Medipix 3 devices, each 256x256 pixels, for a total frame size of 2048x1536 twelve bit pixels. Medipix 3 has advantages over Pilatus since it has smaller pixels and pixel voting logic, which eliminates non-linearity's near pixel boundaries. Diamond has some other developments based on the same detector. 13 June 2011 Area Detector Developments

Area Detector Developments Excalibur Hardware 10 GigE Network Each row of 8 devices (a stripe) is read out by separate Front End Modules (FEMs) into a Linux readout node over a 10G bit Ethernet link. The readout nodes stream the stripes independently to a Lustre file server in HDF5 format over the external 10 GigE network. At the maximum data rate of 100 frames per second, each stripe generates 840Mbps, a total rate of just over 5Gbps to the fileserver. A master node provides synchronization, real time (low rate) summary image, global configuration, etc.. 13 June 2011 Area Detector Developments

Area Detector Developments Excalibur Software New plugins required for Front End Module (Fem) readout, Master/Slave communications and HDF5 file writer. Many existing area detector processing plugins (statistics, region of interest, etc) are not shown. 13 June 2011 Area Detector Developments

Area Detector Developments Plugin developments 13 June 2011 Area Detector Developments

HDF5 file format and libraries HDF5 file format and libraries supported by the HDF group www.hdfgroup.org/HDF5 Open source and distributed at no cost Unix and Windows platforms supported Designed for NASA to store large datasets Self describing binary file format Hierarchical data structures can be browsed like a directory tree Multi dimensional imaging data Instrumentation metadata can be stored as simple or complex data types Long term support for accessing archived data Parallel file IO access on supported storage systems including Lustre Requires MPI and MPI I/O e.g. Open-MPI Multi node streaming to disk Fast processing using HPC cluster systems Access selected sections of datasets across dimensions in “hyperslabs” Efficient disk IO when post processing across multiple frames like tomography reconstruction Several language bindings to the C libraries C++, fortran and Java bindings provided by the HDF group Python, Matlab, IDL and several others... 13 June 2011 Area Detector Developments

Area Detector Developments HDF5 and Nexus Nexus is: A definition of keywords and data structures implemented as an HDF5 file. An API which creates Nexus files using the HDF5 library, but not giving access to advanced HDF5 features. So a Nexus file is an HDF5 file, but an HDF5 file is not necessarily a Nexus file. However, if you need access to advanced HDF5 features, you can still write a Nexus file if you abide by the Nexus definitions. One driver is to provide increased performance Want to be able to write at 1 Gbyte/sec on a PC with a 10 GigE interface. 13 June 2011 Area Detector Developments

AreaDetector HDF5 plug-in NDFileHDF5 inherits the areaDetector NDFilePlugin class Operated like other AD file plug-ins so client interface is common Works on Linux and Windows File hierarchy is formatted in a Nexus compatible way Allow files to be opened directly in Nexus readers like GDA User can select disk IO “chunk” size by number of rows to optimise for disk IO speed or post processing performance depending on application Lustre file system perform best with 1MB chunk sizes Tomography reconstruction is simple with chunks of 1 row per frame Works with HDF5 supported compression libraries Lossless SZIP provided as a separate library from the HDF group Lossless Z compression Bit packing for cases where there are only few valid data bits per word Continuing development Parallel write plug-in for Excalibur project where 6 parallel nodes will write image stripes into a single HDF5 file AreaDetector HDF5 file reader to use areaDetector plug-ins for post processing Test framework to check disk system performance in different configurations 13 June 2011 Area Detector Developments

Area Detector Developments X-Y scanning Can have two additional “virtual” dimensions for X-Y scanning: 13 June 2011 Area Detector Developments

Area Detector Developments NDFilePlugin Problem: Filename defined in NDFilePlugin over-rode filename specified in the detector driver plugin. If file writing got behind and the control software changed the file name pattern, buffered old data got written with the new name pattern. Solution: NDFilePlugin now can run in a mode where the filename is obtained from the NDArray parameters. Name is set at the time the data is taken, not when it is written. 13 June 2011 Area Detector Developments

Area Detector Developments adOpenCV Plugin Experimental plugin based on Open Source Computer Vision (OpenCV) library OpenCV is a C/C++ library originally developed by Intel and now supported by Willow Garage. http://opencv.willowgarage.com/wiki/Welcome Mainly aimed at real time computer vision. Free for use under the open source BSD license. Plugin can be used for: Sample detection and alignment. Collision detection. Generic filtering of images Contains base classes to convert between the areaDetector and OpenCV data structures Need to ensure that image recognition is relatively simple, so manage lighting and contrast levels. 13 June 2011 Area Detector Developments

Area Detector Developments Summary areaDetector has major benefits for beamlines. However, with this flexibility comes increasing demands for more features and greater driver support. Diamond is now spending more on areaDetector development than any other beamline work. 13 June 2011 Area Detector Developments