HDF Update
0 likes270 views
This document summarizes Mike Folk's presentation at the Science Data Processing Workshop from February 26-28, 2002. The presentation provided updates on HDF4 and HDF5, including recent releases and future plans. HDF4 and HDF5 are open source data formats and software libraries for scientific data that support efficient storage of arrays, images, and tables. The presentation outlined ongoing work to improve performance, add new features, and facilitate the transition from HDF4 to HDF5.
HDF Update
- 1. HDF Update Mike Folk National Center for Supercomputing Applications Science Data Processing Workshop February 26-28, 2002 -1- HDF
- 2. Topics • What is HDF? • HDF4 update and plans • HDF5 update and plans -2- HDF
- 3. NCSA HDF Mission To develop, promote, deploy, and support open and free technologies that facilitate scientific data storage, exchange, access, analysis and discovery. -3- HDF
- 4. What is HDF? • • • • • • Format and software for scientific data Stores images, arrays, tables, etc. Storage and I/O efficiency Free and commercial software support Promote standards Broad user base in engineering & science -4- HDF
- 5. Who is supporting HDF? • NASA/ESDIS – Earth science applications, instrument data – All aspects of data management • DOE/ASCI (Accelerated Strategic Computing Init.) – Simulations on massively parallel machines – Emphasis on parallel I/O performance, functionality • DOE Scientific Data Analysis & Computation Program – High performance I/O R & D • NCSA – Grid, Vis, other R&D, user support • Others – Applications, support, some R&D -5- HDF
- 6. HDF4 -6- HDF
- 7. Goals for HDF4 • Support Terra, Aqua & other EOS data users • Maintain and upgrade library and tools • Address differences between HDF4 & HDF5 -7- HDF
- 8. HDF4 milestones since Sept 2000 Q4 ‘00 Q1 ‘01 Q2 ‘01 Q3 ‘01 Q4 ‘01 Q1 ‘02 HDF4 library and utilities: HDF4 Java API & tools: •H F4 D •V . r4 1 sio er •H F4 D . r5 1 2.6 n •V -8- sio er 2.7 n HDF
- 9. HDF4 releases • HDF4.1 r4 (Nov 2000) – – – – Chunking and chunking with compression JPEG compression fixed hdp utility enhancements HDF4-to-GIF and GIF-to-HDF4 converters • HDF4.1 r5 (Nov 2001) – – – – – Compression query functions Vdata: set size of Vdata link blocks Tools updates Added Solaris 2.8 Retired Solaris 2.6, OSF1 V4.0 and HPUX 10.20 -9- HDF
- 10. HDF5 - 10 - HDF
- 11. HDF5 milestones since Sept 2000 Q4 ‘00 Q1 β ‘01 Q2 ‘01 Q3 ‘01 Q4 ‘01 Q1 ‘02 ) afe s ch at dp 3 rea 2 0 1 2( Base .4. . 4 . 1.4. .4. 1.4. Th _ _1 _1 _ _1 _ library: β e” ag β “im e” & High level l e” ab “t “lit library: _ • Java & other tools: HDF4-HDF5 conversion: _ .4. 1 0 _ .4. 1 1 ion rs e nv co lity _ ti u - 11 - Ja _ v 2.2 a α α2 y y rar ibrar lib _ _l HDF
- 12. HDF5 Library Work - 12 - HDF
- 13. New platforms and languages • New platforms – – – – – – • Fortran 90 HP-UX IBM SP IA64 architecture Solaris 2.8 with 64-bit Linux in parallel mode Metrowerks Code Warrior Windows compiler – Solaris 2.6 & 2.7, O2K, DEC OSF, Linux, Cray T3E, SV1 – Parallel on T3E & O2K – Windows • C++ - 13 - – Solaris 2.6 and 2.7, Free BSD, Linux, Windows HDF
- 14. New to HDF5 • • • • • • Virtual File Layer – alternate I/O drivers Array datatype – array as atomic type File sizes greater than 2GB on Linux Performance improvements – parallel & serial Improvements in configuration Tutorials and documentation - 14 - HDF
- 15. Next major release -- HDF5 1.6 • New format and library features – – – – – – Reclamation of free space within a file Enhanced hyperslab/region selection Dimension scale support Bzip2 compression Generic Properties Performance improvements • Parallel I/O performance benchmark suite • Release date: Fall 2002 - 15 - HDF
- 16. High level APIs • HDF5 routines that do more operations per call than the basic HDF5 interface • Goals – Make HDF5 easier to use – Encourage standard ways to store objects in HDF5 - 16 - HDF
- 17. High level APIs • • • • • • Lite – done Image – done Table – partly done Dimension scale – in the works Unstructured grids – in the works http://hdf.ncsa.uiuc.edu/HDF5/hdf5_hl/doc/ - 17 - HDF
- 18. HDF5 High Level APIs – HDF5 Image • For datasets to be interpreted as images/palettes – 2-D raster data like HDF4 raster images • Image operations – Create, write, read, query • Based on “HDF5 Image & Palette Specification” - 18 - HDF
- 19. HDF5 High Level APIs – HDF5 Table • For datasets to be interpreted as “tables” – A collection of records – All records have the same structure – Like Vdatas in HDF4, but more operations • Table operations – Create, write, read, query – Insert, delete records or fields – Future: sort and search - 19 - HDF
- 20. HDF5 tools activities - 20 - HDF
- 21. H5View – new features • • • • • • • Improved editing XML support Supports “image” data sets Line plots of row/column data Save data to a text file Set user preferences http://hdf.ncsa.uiuc.edu/java-hdf5-html/ - 21 - HDF
- 22. Future Viewer work • Objectives – Support transition from HDF4 to HDF5 – Eliminate duplicate work on HDF4 and HDF5 tools – Provide plug-in architecture for tool development • Three stages – Object model to cover both HDF4 & HDF5 • Nearly done – Simple browser for both HDF4 and HDF5 • Summer 2002 – Editor • Fall 2002 – Modular toolkit for browser plug-ins • Fall 2002, if all goes well - 22 - HDF
- 23. XML and web experiments • Explore XML, Web applications for HDF5 – Investigate use of XML – investigate Web XML technologies, use with HDF5 • XML DTD for HDF5 • Conversions involving XML – HDF4 HDF5 XML – Compression and timing study – netcdf to HDF5 translation, via XML style sheet • XML Schema for HDF5 investigated - 23 - HDF
- 24. Other Java-inspired explorations • Tomcat web server and JSP servlets – HDF5 file XML html display in a web browser – Shows promise for providing remote access to HDF5 • HDF5 and CORBA – – – – Investigating using CORBA with Java and HDF5. Data access done in a CORBA servant written in C/C++ Browsing and presenting the information done in Java Uses CORBA is alternative to the Java Native Interface, which calls C directly from Java. - 24 - HDF
- 25. Other Tools Activities • HDF5-to-GIF and GIF-to-HDF5 converters • H5dump subsetting - 25 - HDF
- 26. Facilitating the transition from HDF4 to HDF5 - 26 - HDF
- 27. The transition from HDF4 to HDF5 • We support both HDF4 and HDF5. • We will continue to maintain HDF4, as long as we are funded to do so. • We recommend: • Use HDF5 for new projects • Consider migrating from HDF4 to HDF5 to take advantage of the improved features and performance of HDF5 - 27 - HDF
- 29. HDF4-to-HDF5 Work • Later talk by Bob McGrath to cover – The key technical challenges – NCSA toolkit work – Experiments to support transition - 29 - HDF
- 30. Other Activities of Interest - 30 - HDF
- 31. Parallel HDF5 • • • • HDF5 supports parallel I/O Emphasis on performance Driven by DOE labs Tutorial http://hdf.ncsa.uiuc.edu/HDF5/doc/Tutor/ • See Elena’s talk - 31 - HDF
- 32. Parallel I/O benchmark suite • A set of parallel performance benchmarks that can be distributed with HDF5 • Parallel HDF5 in MPI environment • Measures performance of raw I/O, MPI-I/O, & HDF5 • Draft documentation: http://hdf/RFC/PIO_Perf/PHDF5_performance.html - 32 - HDF
- 33. Other performance studies See Elena’s talk - 33 - HDF
- 34. Thread-Safe HDF5 (Beta) • Uses Pthreads (POSIX threads) library. • Implements Phase 1 of HDF5 thread safety plan – One lock per program • Later phases – 2: Separate locks per major operation (read, write, convert) – 3: Separate locks for all operations. • For details see – http://hdf.ncsa.uiuc.edu/HDF5/papers/mthdf/ - 34 - HDF
- 35. HDF5-DODS* Server • HDF5 data available from DODS clients • Demonstrated with Ferret client • Documentation – HDF5-DODS Data Model and Mapping – The HDF5-DODS Server Prototype – Demo of HDF5-DODS Server Prototype with the DODS-Ferret Client • http://hdf.ncsa.uiuc.edu/apps/dods/ * Distributed Oceanographic Data System - 35 - HDF
- 36. Transform architecture prototype • Provides a mechanism for operating on data as it is moved from one location to another • Inspired by HDF5 I/O operations when moving data between memory and file: – Convert numbers – size, endianness, etc. – Selection – subsetting & subsampling – Linear order – row major vs. column major • Later – Change value/Meaning of data – units, coordinate system, etc. - 36 - HDF
- 37. Outreach • Documentation – Expanded and improved – Searchable, printable – HDF5 User’s Guide • Advanced and parallel HDF5 Tutorial – http://hdf.ncsa.uiuc.edu/HDF5/doc/Tutor/ – Presented at SC2001 http://hdf.ncsa.uiuc.edu/HDF5/papers/SC2001/SC01_tutorial/ - 37 - HDF
- 38. szip Compression Reporting for Pen-Shu Yeh and Wei Xia - 38 - HDF
- 39. Szip Compression in HDF 4 • CCSDS lossless compression – Fast, effective compression method for EOS data – Outperforms other common techniques in size & speed • Szip software – Implements CCSDS lossless compression – Originally developed at University of New Mexico – Implemented in HDF-4 by UNM • HDF-4 with CCSDS-szip option (HDF-4-Szip) – Delivered to GSFC and tested – Tested with MODIS Level-1B data from GSFC DAAC - 39 - HDF
- 40. Sample results Convert 343 Megabyte MODIS file Conversion performed on PC-Linux 7.1, Pentium II, 300Mhz Compression Ratio Compression and decompresson time (seconds) 2.80 3.00 700 600 500 400 300 200 100 0 558 Huff 2.28 2.50 575 Huff 2.00 1.60 1.50 RLE szip RLE szip 1.00 86 72 42 64 0.50 0.00 CompressTime DecompressTime Rle - 40 - Huff szip HDF
- 41. Szip-HDF4 software • HDF4 library routine SDsetcompress – compress existing dataset or create new compressed dataset • Utilities – bin2hdf – input array from binary file, compress using HDF-4-Szip, and output compressed data in HDF file – hdf2chdf – input data from HDF file, compress using HDF4-Szip, and output compressed data in new HDF file – chdf2bin – input HDF file (with or without compression), and output user-selected arrays as separated binary files - 41 - HDF
- 42. Future Work • Compress IEEE 32 bit float data with HDF-4-Szip • Use HDF chunking routines – Especially beneficial for subsetting performance • Test HDF-4-szip on different computer platforms – Currently HDF-4-Szip has been tested under Linux only – Others to test: Sun, SGI and HP, Windows 98 and NT • Advocacy in the scientific community – Add User’s Guides and documentation to HDF docs. – Present and demo at workshops and conferences. • Resolve distribution issues with NCSA and UNM - 42 - HDF
- 43. Thank you! Information Sources HDF • HDF website – http://hdf.ncsa.uiuc.edu/ 5 • HDF5 Information Center – http://hdf.ncsa.uiuc.edu/HDF5/ • HDF Helpdesk – hdfhelp@ncsa.uiuc.edu • HDF users mailing list – hdfnews@ncsa.uiuc.edu - 43 - HDF