Parallel programming model

Parallel programming model

• 1.
  PARALLEL PROGRAMMING MODEL •Themodel of a parallel algorithm is developed by considering a strategy for dividing the data and processing method and applying to reduce interactions.
• 2.
  SHARED VARIABLE MODEL 1.A program is a collection of processes. parallelism depend on how inter –process communication is implemented. 2. Multiprocessor programming is based on the use of shared variables in a common memory for ipc (inter process communication). 3. Shared variable ipc demands the use of shared memory and mutual exclusion among multiple process accessing the same set of variable.
• 3.
  Block diagram ofshared varibale communication process A Shared variable in common memory Process B ProcessC
• 4.
  Issues with sharedvariable 1. Critical section problem 2. Memory consistency 3. Atomicity with memory operation 4. Fast synchronization 5. Shared data structure
• 5.
  MESSAGE PASSING MODEL Two process A and B communicate with each other by passing message through a direct network.  The messages may be data ,instruction , synchronization or interrupt signals etc.  delay cause by a message is much more longer than a shared variable model in a same memory. ProcessA Process B
• 6.
   Two messagepassing programming models are introduced here. 1. SYNCHRONUS MESSAGE PASSING 2. ASYNCHRONUS MESSAGE PASSING 3. synchronous message passing: it synchronizes the sender and receiver process with time and space just like telephone. 4. No shared memory 5. No need of mutual exclusion 6. No buffer is used in communication channel 7. It can be block by channel being busy or error 8. One message is allowed to be transmitted via a channel at a time
• 7.
   Sender andreceiver must be coupled in both time and space  Also called blocking communication scheme ASYNCHRONOUS MESSAGE PASSING: Does not need to synchronize the sender and receiver in time and space Non –blocking can be achieved Buffer are used to hold the message along the path of connecting channel
• 8.
  DATA PARALLEL MODEL The data parallel code is easier to write and debug because the parallelism explicitly is handled by hardware synchronization and control flow  Data parallel languages can be directly modified from standard serial programming languages  Data parallel programming emphasizes local computation and data routing , replication ,reduction and parallel prefix
• 9.
   It isapplied to fine grain problems using regular grids ,stencils and multidimensional signal /image data sets.  In data parallel model tasks are assigned to the process and each task perform similar type of operation on different data  Data parallel model can be applied on shared- address space , message passing paradigms  The primary characteristic of data parallel model problems is that the intensity of data parallelism increases with the size of problem which in turn makes it possible to use more processes to solve larger.
• 10.
  Example: dense matrix multip. Processor1 Processor 2 Processor 4 Processor 3 Memory unit Control unit
• 11.
  Object oriented model In this model objects are dynamically created and manipulated  Processing Is performed by sending and receiving messages among objects  concurrent programming models are built up from low level objects such as queues , semaphores and process into high level object like monitor and program modules  Program abstraction lead to program modularity and software reusability as is often found in oop
• 12.
   Other areashave encouraged the growth of oop include the CAD tools and word processors with graphics capability  The idea behind parallel object oriented programming is to provide suitable abstractions and software engineering methods for structured application design  Processing is performed using object
• 13.
  FUNCTONAL AND LOGICMODELS  Functional programming languages such as pure , lisp , sisal , strand 88.  A functional programming language emphasizes the functionality of program and should not produce side affects after execution  There is no concept of storage ,assignment and branching programs  It is used in artificial intelligence where parallel processing is verymuch in demand
• 14.
  Logic models  Logicprogramming languages such as concurrent space prolog and parlog  Based on predicate logic , logic programming is suitable for knowledge processing deal with large database