Given by Geoffrey C. Fox at ParCo Conference Bonn on Sept 16,97. Foils prepared Sept 12 97
Outside Index
Summary of Material
We describe how a network of servers architecture can naturally support both parallel and distributed computing while |
Java can be used for both the client interface and for programming parallel and sequential scientific problems |
This is illustration in multidisciplinary applications, forces modeling, integration of CORBA, role of component based software, interpreters |
we discuss role of collaboration technologies |
we describe relevance of Web technologies |
Outside Index
Summary of Material
Geoffrey Fox |
Syracuse University |
NPAC |
111 College Place Syracuse NY 13244 4100 |
3154432163 |
We describe how a network of servers architecture can naturally support both parallel and distributed computing while |
Java can be used for both the client interface and for programming parallel and sequential scientific problems |
This is illustration in multidisciplinary applications, forces modeling, integration of CORBA, role of component based software, interpreters |
we discuss role of collaboration technologies |
we describe relevance of Web technologies |
Ideas from HPCC research Good! |
Not enough people/funding in field to implement robust production systems |
Must re-use as much software (including infrastructure software) as possible |
Similarly must build HPCC software in a modular fashion with small enough modules that smallish groups can build effectively |
Different modules are likely to use different base technologies (Fortran v Java v C++ etc.) and so interoperability essential! |
No silver bullet on the horizon - maybe pessimistic but implies better HPCC environments implies better implementations of existing ideas. |
Need to support both production use of MPP's and "rapid prototyping" in development of new applications - latter is not well supported by current HPCC software systems even though need parallel support for prototyping of new 3D simulations |
PC and workstation clusters are of growing important and this typically distributed memory people's technology is contrasted with distributed shared memory tightly coupled MPP's. |
Computational science moving to multidisciplinary (multi-component) applications |
Corresponding growing use of databases (for data-intensive applications) |
Interoperability between disparate heterogeneous platforms, support of multidisciplinary applications, and metacomputing are three related important areas |
"full metacomputing" (decompose general problem on general networked resources) may not be relevant |
The Web is delivering a new operating environment (WebWindows) and a rich distributed computing software infrastructure with especially excellent support for software integration |
There is a need for a new scalable technical operating system (NT v UNIX v WebWindows) |
Bottom of Pyramid has 1000 times dollar value and compute power of best supercomputer (tip of pyramid) but supercomputer has high performance network to support close synchronization needed by classic parallel algorithms |
Use of |
Web Technologies |
is naturally a |
By definition, Web Software will be the "best" software ever built because it has the largest market (and so greatest leverage of investment dollars) and most creative business model (harness the world's best minds together with open interfaces)
|
One should build upwards from the "democractic Web"
|
This allows you to both deliver your application to the general public (not always required but often desireable) and use the best leveraged software |
Note Web Software tends to offer highest functionality as opposed to highest performance and HPCC often requires different trade-offs |
Web Technology is still uncertain and there may be major changes but "enough" capabilities are in place to build very general (~all) applications
|
Rapidly evolving Standards and a mechanism to get rapid consensus |
Fortran 77 -> Fortran90 --> HPF --> Fortran2000 (23 years) |
VRML Idea (1994) --> VRML1 deployed (95) --> VRML2 deployed (early 97) (2.3 years)
|
Classic Web: HTTP Mime HTML CGI Perl etc. |
Java and JavaScript Compiled to almost compiled (applet) to fully Interpreted Programming Language |
VRML2 as a dynamic 3D Datastructure for products and their simulation object |
Java Database Connectivity (JDBC) and general Web linked databases |
Dynamic Java Servers and Clients |
Rich Web Collaboration environment building electronic societies |
Security -- still needs maturing as very clumsy or non existent at present in many cases |
Compression/ Quality of Service for Web Multimedia
|
Emerging Web Object model including integration of Corba (see JavaBeans and Orblets) |
One can "just" use Object Web technologies as a software infrastructure for building parallel, distributed or sequential computing environments which can have a very different architecture from the Web
|
As above, but specifically build a metacomputing environment |
One can view the Object Web as a distributed information system with modest performance and build a metacomputing system with the Web architecture
|
Harness the power of the Web as a computer -- use up the idle cycles on the WebTV's in every home -- typically a Web Client based system |
So this is partly use of Object Web Software to build metacomputing systems of any architecture |
Perhaps VRML or Java3D are important for scientific visualization |
Web (including Java applets) front-ends provide convenient customizable interoperable user interfaces to HPCC facilities |
Perhaps the public key security and digital signature infrastructure being developed for electronic commerce, could enable more powerful approaches to secure HPCC systems |
Perhaps Java will become a common scientific programming language and so effort now devoted to Fortran and C++ tools needs to be extended or shifted to Java |
The universal adoption of JDBC (Java Database Connectivity) and the growing convenience of web-linked databases could imply a growing importance of systems that link large scale commercial databases with HPCC computing resources |
Javabeans, RMI, COM, CORBA, IIOP form the basis of the emerging "pragmatic object web" which analogously to the previous bullet could encourage a growing use of modern object technology which will allow better managed distributed systems |
Emerging collaboration and other distributed information systems could allow new (low end) distributed work paradigms which can be used in computational steering and for basic scientific research |
High end (today) to high end(tomorrow)
|
High end(today) to low end(tomorrow)
|
Low end(today) to low end(tomorrow)
|
Low end(today) to very low end (tomorrow)
|
The new application(s) we can't predict |
1997 |
2007 |
Any large scale metacomputing/distributed computing environment will be shaped by and shape all 5 classes of applications on previous foil |
Even if only aimed at high end applications, the system will be influenced by and influence the "Object Web" or "commodity software infrastructure" which is here defined as "mass-market"/business IntraNet (low to low) use of Internet/distributed Information System |
Parallel Computing systems can be viewed as a special case of a Metacomputer
|
Applications are metaproblems with a mix of module and data parallelism |
Modules are decomposed into parts (data parallelism) and composed hierarchically into full applications.They can be the
|
Modules are "natural" message-parallel components of problem and tend to have less stringent latency and bandwidth requirements than those needed to link data-parallel components
|
Assume that primary goal of metacomputing system is to add to existing parallel computing environments, a higher level supporting module parallelism
|
We can distinguish Decomposition and Integration |
Decomposition is performed by an HPF or other Parallelizing compiler; or by a user writing a Fortran + Message Passing code "by hand" |
MPI integrates decomposed parts together with high bandwidth latency constraints |
Systems such as AVS integrate larger modules together and much of "software engineering" (modular style of programming) involved with this |
Web is a powerful integration model suitable for large coarse modules with modest latency and sometimes modest bandwidth requirements
|
Collaboration, computational steering, multidisciplinary science are all integration and not decomposition problems! |
See Original Foil |
Object Web Software provides a high functionality but modest performance distributed computing (Metacomputing) environment based on either Web (soon to be CORBA IIOP and HTTP/Java Socket) Servers or Clients |
Here we will explore an architecture using servers for control as higher functionality than clients although currently less broadly deployed
|
Object Web Only addresses Integration of already decomposed parts!
|
Middle Tier |
Basic Web Server |
Custom Web Server |
TP Server |
Business Transaction Management |
You Write Software |
at Client and Server |
Old and New Useful Backend Software |
1:User View: Interoperable Web Interface accessing services through Java Compute Services Framework |
2:Network of Java Servers provide distributed services with databases, compute engines, collaboratories, object brokers, instruments
|
Back end "Number Crunchers" linked either by communication at level 2 (slowish but easy) or at level 3 (high performance but more work) |
Compute processes linked either to servers or together by MPI if parallel |
Java Servers |
We have a set of Services hosted by Object Web Servers which form the middleware and accessed by clients |
Groups of clients (electronic societies) are linked by Java server based collaboration systems such as TANGO or Habanero |
Access |
Resources |
Store |
Multimedia Information |
Collaboration Server |
File Systems |
and/or Database |
Object Broker |
Database |
Simulation |
e.g. NEOS |
Netsolve |
Computer |
Person2 |
Shared |
WhiteBoard |
Shared Client Appl |
Person1 |
General User |
Systems like Tango or Habanero built around Java Servers integrate a group of multiple clients as a "Service" at the middle Java Server level |
Building systems in this way automatically includes "people in the loop" -- Computational Steering, Education, Multidisciplinary collaborative design |
Group of collaborating clients |
and client applications |
Database |
Object Broker |
MPP |
Java for the User Interface: This is roughly the "WebWindows Philosophy" of building applications to Web Server/Client Standards |
Java for Coarse Grain Software Integration: see collaboration and metacomputing |
Java as a high performance scientific language: for "inner" (and outer) loops Here parallelism is important but sequential issues also critical and first issues to examine! |
This is least controversial and is essentially WebWindows for User Interfaces |
Fortran was never good at user interfaces! |
Initially Aimed at education where usability higher priority than performance |
Teaching Java and JavaScript greatly aided by interpreted technology which allow integration of demonstrations into lectures |
VPL aimed at allowing embedding of F90, HPF and MPI (etc.) examples in lectures and convenient support of homeworks for transient inexperienced users. |
Features of VPL:
|
Click on SnapShots in Virtual Programming Lab |
Click on SnapShots in Virtual Programming Lab |
Scivis is a client-server (3-tier) data visualization and analysis system by taking full advantage of Java. |
The purpose of this system is provide researchers with a customizable data analysis system to aid their research. |
We also provide a collaborative framework, where the users can exchange data and their own personalized filters. |
Available via http://kopernik.npac.syr.edu:8888/scivis |
Contains over 29K lines of Java code (2K lines of user-definable filters). |
A screen dump from a Scivis Session |
VRML naturally gives 3D visualization with usual Web advantage of running on PC's and Workstations |
Its universality implies can use in industry to specify products so can design, manufacture and market from the same(related) specification |
Should impact PDES/STEP and such industry product specification standards |
VRML will need extension to handle this but it is a good start and allows user defined types |
VRML and Parallel Computing?
|
NPAC Web Based Geographical Information System in Stand Alone Mode |
A GIS application is a specialized OpenInventor viewer, however it accepts any OpenInventor 2.1 scene model. That's why it's so easy to integrate it with third party applications, which produce IO/VRML output. The images show GIS integration with Weather Simulation application. A GIS viewer can also display animated objects controlled by Simulation Engine. |
It is natural to base on either a network of Web Clients or Web Servers
|
Web Client Models Include SuperWeb (Javelin) from UCSB and are well illustrated by the January 1997 hotwired article "Suck your Mips". |
Greater functionality but less power and pervasiveness is a pure Web Server model as proposed by NPAC
|
Note total compute power in all Web "clients" is about 100 times that in all Central Supercomputers |
http://www.packet.com/packet/ Hot Wired Tuesday January 7 Edition |
Applet calculates pi while you read article! |
This is "middleware" which is implemented in simplest form as a network of Java Servers
|
Access |
Resources |
Store |
Multimedia Information |
Collaboration Server |
File Systems |
and/or Database |
Object Broker |
Database |
Simulation (Network-enabled |
servers such as NEOS, Netsolve) |
Sequential |
or Parallel |
Computer |
As a first step, implement multi-module systems with each module linked via Java Servers
|
Where necessary "escape" down to classic HPCC technologies for data transport keeping control at server level
|
This seems very convenient in JDK 1.1 "event model" which is mechanism used by Javabeans to communicate
|
1)Simple Server Approach 2)Classic HPCC Approach |
3)Hybrid Approach with control at server and |
data transfer at |
HPCC level |
4)Invoke High Performance Message Transfer between Observers and Sources specified in Message Event |
Server Tier |
Data Source |
Data Sink (Observers) |
5)Actual Data Transfer |
High Performance Tier |
2)Prepare |
Message Event in Source Control |
1)Register Observers with Listener |
Here are some examples of using our approach where large scale industry investment in Web technology appears to add significant value to metacomputing systems built with Web architecture
|
Multidisciplinary and Computational Steering Applications
|
Visual and Interpreted Programming Environments
|
Technologies to get High Performance CORBA |
Integration with Forces Modeling (Distributed Event driven Simulation) |
Integration with Networked enabled servers such as NEOS and Netsolve
|
Note Java also integrates compiled and interpreted approaches and so leads to more convenient programming environments
|
JavaScript is a fully interpreted language but not really Java |
Applets are half-way between traditional compiled and interpreted approaches |
Web "systems" can behave like Interpreters with interactive commands at client (gives Web version of MATLAB) |
Web Client |
including |
Java Applets |
Web Server |
Java/Fortran/C++ |
Application Backend |
User Interpreted Commands |
invoking preloaded |
Java/Javascript objects |
Server invokes extrinsic |
processor |
(true interpreter |
or compiler invoked |
dynamically) |
Running Program |
interruptable at |
extrinsic interfaces |
Compiler offers:
|
Interpreter offers:
|
JAVA applet |
HTTP |
server |
HPF |
server |
Node 1 |
Node 2 |
Node 3 |
Node 4 |
Node 5 |
Node 6 |
Web Browser |
Server Host |
1. Contact HTTP server to download the Java applet |
2. Use the applet to start instrumented HPF application |
3. HPF server starts (implemented as an HPF extrinsic) |
4. Java applet establishes communication with the HPF server |
5. HPF server accepts user's requests: |
- suspend/resume execution of the HPF code |
- send data |
- interpret new HPF statements |
- dynamically link and execute shared objects |
HPF statements and Visualization requests in Java, HPF etc. |
HPF extrinsic procedure |
Shared Object |
Lexical Analysis: PCRC frontend. |
HPF compilation: pghpf |
Pool of shared objects |
HPF RTS |
HPF |
server |
Dynamical linking |
Instrumented HPF |
JAVA |
applet |
Applications requires a range of capabilities in any language |
High level ("Problem Solving Environment") manipulating"large" objects
|
Intermediate level Compiled Code targetted at "sequential" (multi-threaded) architecture
|
Lower level runtime exploiting parallelism and memory hierarchies
|
Numerical Objects in (C++/Fortran/C/Java) |
Expose the Coarse Grain Parallelism |
Expose All Levels of Memory Hierarchy |
a) Pure Script (Interpreted) |
c) High Level Language but Optimized Compilation |
d) Machine Optimized RunTime |
b) Semi- Interpreted |
a la Applets |
Memory Levels in High |
Performance CPU |
Nodes of Parallel/ Distributed System |
Simulation |
Basic Display |
Image Filter |
is another |
module |
Output Display after Filter |
Runs as a |
parallel |
module |
using |
Java Server |
host |
Bunch of Filters and Displays |
defined in |
Java Graph editor and |
running on grid of Java Servers |
Original Image |
Visual Basic/C++/J++ and ActiveX or Beanboxes with Javabeans give visual approach to software objects
|
Enterprise Javabeans and COM are extending this to distributed computing |
Using Web technologies for grid and building modules out of (whatever Javabeans/COM evolves to) allows one to deliver to user HPCC programming environments with comparable friendliness to those in PC world |
New Java Frameworks (Enterprise, Media) bring new family of Java servers that can implement an object based middleware and backend layers for the Web based services including computing. |
One critical development in the distributed objects domain is the IIOP (Internet Inter-ORB Protocol), mandatory in CORBA2 and offering a new server-to-server communication standard between commercial ORBs from various vendors. |
At the moment, Web Server market is dominated by customized or/and proprietary systems but there is a new strong tendency to standardize the Web middleware around the CORBA paradigm. |
'Java based ORB' is a CORBA2 compliant ORB written exclusively in Java, and hence fully portable and ready to run also on the Wintel platform, where distributed objects were so far under control of Microsoft COM/ActiveX model. |
The following figure illustrates a possible CORBA-HPCC integration framework ( HPCORBA ) |
Each node of a parallel machine runs a lightweight (e.g. Nexus based ) ORBlet which enables both local and remote object services. |
Coming CORBA3/MOM based messaging and object-by-value frameworks adapted to provide high-performance (e.g. MPI based) object-oriented (IIOP based) inter-node communication. |
Parallel Computation graphs are represented by Interface Definition Language (IDL) in a language independent way and optimized by parallel compiler technologies. |
Specific language bindings would result in HPF, HPC++ or HPJava implementations. |
This framework provides interoperability with the emergent Object Web software industry and offers an elegant solution for scalable parallel I/O ( linking HPCORBA with Oracle CORBA ) |
DoD modeling community is currently evolving towards the HLA(High level Architecture) framework with the RTI (Run Time Infrastructure) based communication bus. |
The goal of HLA/RTI is to enhance interoperability across more diverse simulators than in the DIS realm, ranging from real-time to time-stepped to event-driven paradigms. |
HLA defines a set of rules governing how simulators (federates) interact with each others. Federates describe their objects via Object Model Template (OMT) and agree on a common Federation Object Model (FOM). |
The overall HLA/RTI model is strongly influenced by the CORBA architecture and in fact the current prototype development is indeed CORBA based. |
We suggest that next step is to combine CORBA2 (Initial HLA/RTI is CORBA1) with NPS prototype ideas to give a fully object and Web integrated event driven simulation environment. |
Java3D is natural visualization environment in this scenario |
A large number of new Java APIs for advanced Web Services are emerging from JavaSoft and partners. Many current problems such as with the NPS WebDIS networking will be soon solved in a more robust, stable and elegant fashion by the new high-level APIs. |
Java APIs are organized in Java Frameworks. Current list of frameworks include:
|
Java Security Framework - support for authentication encryption, digital signatures. |
Java Commerce Framework - Java Wallet, Java Cassettes (digital credit cards). |
Java Beans Framework - componentware API with support for GUI negotiation and merging, persistence (JAR files), event filtering, introspection, visual application builders |
Java Media Framework - Java2D (with Adobe), animation (with Macromedia), audio/video (with Intel), Java3D (with SGI), JSDA (Java Shared Data API). |
All APIs listed are either already operational within JDK1.1, or still in works (spec only or alpha or beta release) but with the final release dates in 1997. |
Numerical Computing Framework
|
Distributed Simulation Framework
|
Computing Services Framework
|
High Performance Framework
|
Distributed Computing Framework
|
Enables development of Web Interfaces to run a given job on any computer compliant with this framework just as JDBC gives a universal interface to any relational database |
Compiling, Executing |
Scheduling jobs as in Codine or LSF |
Accessing and storing into File Systems |
Visualization Interface (how applets access server side information) |
The Computing Services Framework allows vendors to compete on User Front End (GUI) or back end services |
The framework is implemented as a set of drivers which map generic Java Interfaces to particular software (e.g. a compiler) on particular machines. |
Although we have 384 foilsets imported to the Web -- this is not the correct way of doing it!
|
Originally we thought a "Web Foil" is an enhanced HTML and we built a prototype using HotJava |
However now believe this is not powerful and correct Web implementation of "foils" is a JavaBean stored as a serialized Java Object
|
Start with a full drawing program supporting "art" and importing of images where basic entities are Java objects |
Start with system integrated with Tango |
Basic Unit is a "foil" -- typically NOT scrolled (but supporting this) |
Foils are arranged into foilsets dynamically as in current WebWisdom and allow notes(addons) etc. |
Import existing PowerPoint/Persuasion from "outline" text with automatic font size scaling
|
Allow user to add highlighting and images "cut" from original screendump of PC foil |
Latest Tango Capabilities June 97 |
Java is basis of Web Collaboration Systems with Applets Coordinated by Java Server |
Habanero from NCSA was one of first |
TANGOsim uses more modern Web Technology and incorporates a Discrete Event Simulator |
TANGO is a software framework supporting computer-based communication and collaboration
|
TANGO supports client side electronic societies of people, instruments and their applications. |
TANGO links these to a network of Java Servers which manage TANGO sessions and interfaces them to the world wide Server Infrastructure that underlies our model of Web applications
|
Electronic societies or groups are managed by core database in TANGO |
Systems like Tango or Habanero built around Java Servers integrate a group of multiple clients as a "Service" at the middle Java Server level |
Group of collaborating clients |
and client applications |
Database |
Object Broker |
MPP |
Technically, TANGO is a distributed system based on event broadcasting
|
Core system is implemented in Java and as a browser plug-in |
Base programming module is a Java applet
|
From Tango - A Java/WWW-Based Internet Collaborative Software System part of NPAC Overview May 1997 |
From Tango - A Java/WWW-Based Internet Collaborative Software System part of NPAC Overview May 1997 |
Global architecture is a mesh of servers
|
Session metaphor is a meeting room with tools
|
Asynchronous collaboration is (will be) enabled via session recording to database backend |
TANGO modules can be downloaded from anywhere
|
Application select |
buttons |
Application |
Domains |
Open/close and |
floor control |
Session |
information |
User information |
TANGO session, sync, and |
floor control buttons built |
into the application control |
panel. |
Generic tools: chats (1D, 2D, 3D*), whiteboards (paint and draw) |
Multimedia tools: audio/video conferencing, collaborative indexed streaming video |
Shared browsers: slide shows, fully synchronized, shared JavaScript, customizable client-side apps |
Virtual University: WebWisdom - interactive contents delivery system, a set of applets for physics curriculum |
Domain specific applets: 2D and 3D (collaborative VRML) Visible Human, collaborative AutoCad (feasibility study only) |
Core system implemented by a team of 3 programmers within ~6 months, available.
|
APIs for Java, C/C++, and JavaScript available |
Two generations of GUI implemented |
"Lessons learned" are being incorporated into TANGO II, scheduled for release end of October
|
Basic architectures
|
Basic architectures
|
Sharing possible only if applications deterministic
|
Copies of applications must be present everywhere
|
Environments must be identical
|
Initial state of a newcomer impossible to define
|
Cannot maintain consistency among copies
|
The pros win! Web model enables event broadcasting! |
TANGOsim |
Basic |
Replicated Applications |
1)Virtual Users 2)Customized Views |
TANGO Java |
Collaboratory |
Server |
HTTP |
Server |
MultiMedia Mail |
C2 Commander |
Chat |
VTC |
Event Driven |
Simulation |
Engine |
C2 Radar Officer |
3D GIS |
Scripting |
Language |
C2 Weather Officer |
Message Routing |
SW/Data Distrib. |
Other |
Collaborators |
MultiMedia Mail |
Chat |
Simulation |
Engine Controller |
All Clients |
Typical Clients |
Feb 97 Demonstration of Tango |
From Tango Project for CEWES Collaborative Tool Meeting |
TANGO links people and shared applications such as chat board, audio video conferencing, visualizations, shared white board, common AUTOCAD design and related tools |
CFD |
TANGO Server |
Database |
Object Broker |
MPP |
Structures |
MPP |
Engineer |
+ core |
services |
Visualization e.g.CAVE |
Shared AutoCAD |
Engineer |
+ core |
services |
This combines TANGO for collaboration with WebFlow to link server side applications |
If necessary WebFlow would support high performance inter-module communication as in structures-CFD Linkage example but it would always implement control and this allows TANGO integration with server side computation
|
WebFlow communication model is a dynamic dataflow |
Of course other server side compute models are possible and in general need (web-linked) data bases, file systems, object brokers etc., |
WebFlow supports dataflow model where user must supply routines to process input of data that drives module and output of data for other modules |
TANGO supports shared state and user supplies routines that read or write either
|
Can be done for applications like AUTOCAD as vendor supplies necessary API |
CFD |
Structures |
We have a set of Services hosted by Web Servers and accessed by clients |
Groups of clients (electronic societies) are linked by collaboration systems such as TANGO |
Access |
Resources |
Store |
Multimedia Information |
TANGO Server |
File Systems |
and/or Database |
Object Broker |
Database |
Simulation |
Computer |
Person2 |
Shared |
WhiteBoard |
Shared Client Appl |
Person1 |
General User |
Clearly Java Collaboration Systems are natural implementations of general environments that mix computers and people |
Computational Steering -- a simulation is like a participant in a Tango session which has
|
Need to link to Tango, Java data analysis/visulaization front ends as well as distributed resource management systems such as ARMS from Cornell |
Note synergy with Java Server based distributed computing such as WebFlow which builds an AVS like environment with graphical interfaces to software Integration |
More ambitious to upgrade discrete event simulation component of TANGOsim to support full SIMNET/DSI (Distributed Simulation Internet) functionality. |
Note that Java is natural language for DSI/Forces Modelling because these typically use object parallelism which fits both language and applet/JavaBean capabilities. |
See discussion in http://www.npac.syr.edu/projects/javaforcse |
Participants at JSU |
Teacher/Lecturer at NPAC |
Java for User Interfaces and MetaComputing is natural from its design! |
Java for your favourite Conjugate Gradient routine (etc.) is less obvious ..... |
Java likely to be a dominant language as will be learnt and used by a broad group of users
|
Java may replace C++ as major system building language
|
Clearly Java can easily replace Fortran as a Scientific Computing Language as can be compiled as efficiently and has much better software engineering (object) and graphics (web) capabilities
|
Java can unify classic science and engineering computations with more qualitative macroscopic "distributed simulation and modelling" arena which is critical in military and to some extent industry |
Key question is performance of Java |
Note Web Software can be run on High Performance IntraNets such as Iway so hardware need NOT be a problem! |
Java is currently semi-interpreted and (as in Linpack online benchmark) is about 50 times slower than good C or Fortran
|
Java --> (javac)--> Downloadable Universal Bytecodes --> (Java Interpreter) |
--> Native Machine Code
|
However Language can be efficiently compiled with "native compilers" |
Java ----> (native compiler) |
---> Native (for Particular Machine) Code |
Lots of Interesting Compiler issues for both compiled and scripted Java |
My SGI INDY gets .54 Megaflops for Java 100 by 100 Linpack |
It has 200 Mhz R4400 and current Netlib benchmark for this chip is 32 mflops for optimized Fortran |
For better resolution see JPEG Version |
see http://www.netlib.org/benchmark/linpackjava/ |
Note Just in Time Compilers are giving a factor of 10 from June 96 Measurements! |
see http://www.netlib.org/benchmark/linpackjava/ |
Syracuse and Las Vegas Workshops saw no serious problem to High Performance Java on sequential or Shared Memory Machines |
Some restrictions are needed in programming model
|
For instance, Avoid Complicated Exception handlers in areas compilers need to optimize! |
Should be able to get comparable performance on compiled Java C and Fortran starting with either Java Language or JavaVM bytecodes |
The Interpreted (Applet) JavaVM mode would always be slower than compiled Java/C/Fortran -- perhaps by a factor of two with best technology |
One can use "native classes" which is just a predownloaded library of optimized runtime routines which can be high performance compiled Java, C, C++, Fortran, HPF etc. modules invoked by interpreted or compiled Java
|
Use Native Classes selectively for
|
1)Classic solution of large scale PDE or Particle dynamics problem
|
2)Modest Grain size Functional Parallelism as seen in overlap of communication and computation in a node process of a parallel implementation.
|
3)Object parallelism seen in Distributed Simulation where "world" modelled (typically by event driven simulation) as set of interacting macroscopic (larger than grid points) objects
|
4)MetaProblems consisting of several large grain functionally distinct components such as
|
Java: 1) Not Supported, 2) is Thread mechanism, 3) is Java Objects or Applets, 4) is JavaBeans or equivalent |
Fortran: 1)is supported in HPF, 2--4) are not supported |
The Web integration of Java gives it excellent "network" classes and support for message passing. |
Thus "Java plus message passing" form of parallel computing is actually somewhat easier than in Fortran or C. |
Coarse grain parallelism very natural in Java |
"Data Parallel" languages features are NOT in Java and have to be added (as a translator) of HPJava to Java+Messaging just as HPF translates to Fortran plus message passing |
Java has built in "threads" and a given Java Program can run multiple threads at a time
|
Can be used to do more general parallel computing but only on shared memory computers
|
Combine threads on a shared memory machine with message passing between distinct distributed memories |
"Distributed" or "Virtual" Shared memory does support the JavaVM as hardware gives illusion of shared memory to JavaVM |
Message Passing |
Message Passing |
Java Wrappers (native classes or Server socket connections) around existing data parallel Fortran or C++ |
Native Java and MPI
|
Data Parallel Extensions of Java
|
Java threads for data parallelism on SMP's |