Introduction to parallel computing

PARALLELCOMPUTING
Table of Contents
1.Overview
1.What is Parallel Computing?
2.Why Use Parallel Computing?
2.Concepts and Terminology
1.von Neumann Computer Architecture
2.Flynn's Classical Taxonomy
3.Some General Parallel Terminology
3.Parallel Computer Memory Architectures
1.Shared Memory
2.Distributed Memory
3.Hybrid Distributed-Shared Memory
4.Parallel Programming Models
1.Overview
2.Shared Memory Model
3.Threads Model
4.Distributed Memory / Message Passing Model
5.Data Parallel Model
6.Hybrid Model
7.SPMD and MPMP
5.References

1.What is Parallel Computing?
 Traditionally, software has been written for serial computation:
o To be run on a single computer having a single Central Processing
Unit (CPU);
o A problem is broken into a discrete series of instructions.
o Instructions are executed one after another.
o Only one instruction may execute at any moment in time.

For example:
 In the simplest sense, parallel computing is the simultaneous use of
multiple compute resources to solve a computational problem:
o To be run using multiple CPUs
o A problem is broken into discrete parts that can be solved
concurrently

o Each part is further broken down to a series of instructions
o Instructions from each part execute simultaneously on different
CPUs
 The compute resources might be:
o A single computer with multiple processors;
o An arbitrary number of computers connected by a network;
o A combination of both.
 The computational problem should be able to:
o Be broken apart into discrete pieces of work that can be solved
simultaneously;
o Execute multiple program instructions at any moment in time;
o Be solved in less time with multiple compute resources than with a
single compute resource.

Why Use Parallel Computing?
Main Reasons:
 Save time and/or money:In theory, throwing more resources at a task will
shorten its time to completion, with potential cost savings. Parallel
computers can be built from cheap, commodity components.
 Solve larger problems: Many problems are so large and/or complex that it
is impracticalor impossible to solve them on a single computer, especially
given limited computer memory. For example:
o "Grand Challenge" (en.wikipedia.org/wiki/Grand_Challenge)
problems requiring PetaFLOPS and PetaBytes of computing
resources.
o Web search engines/databases processing millions of transactions
per second
 Provide concurrency: Asingle computeresourcecan only do one thing at a
time. Multiple computing resources can be doing many things
simultaneously. For example, the Access Grid (www.accessgrid.org)
provides a global collaboration network where people from around the
world can meet and conduct work "virtually".
 Use of non-local resources: Using compute resources on a wide area
network, or even the Internet when localcomputeresources arescarce. For
example:
o SETI@home (setiathome.berkeley.edu) over 1.3 million users, 3.2
million computers in nearly every country in the world.
Source:www.boincsynergy.com/stats/ (July, 2012).
o Folding@home (folding.stanford.edu) uses over 450,000 cpus
globally (July 2011)
 Limits to serial computing: Both physical and practical reasons pose
significant constraints to simply building ever faster serial computers:
o Transmission speeds - the speed of a serial computer is directly

dependent upon how fast data can move through hardware.
Absolute limits are the speed of light (30 cm/nanosecond) and the
transmission limit of copper wire (9 cm/nanosecond). Increasing
speeds necessitate increasing proximity of processing elements.
o Limits to miniaturization - processor technology is allowing an
increasing number of transistors to be placed on a chip. However,
even with molecular or atomic-level components, a limit will be
reached on how small components can be.
o Economic limitations - it is increasingly expensive to make a single
processor faster. Using a larger number of moderately fast
commodity processors to achievethe same (or better) performance is
less expensive.
o Current computer architectures are increasingly relying upon
hardware level parallelism to improve performance:
 Multiple execution units
 Pipelined instructions
 Multi-core

2.von Neumann Architecture
 Named after the Hungarian mathematician John von Neumann who first
authored thegeneralrequirements for an electronic computer in his 1945
papers.
 Since then, virtually all computers have followed this basic design,
differing fromearlier computers which were programmed through "hard
wiring".
o Comprised of four main components:
 Memory
 Control Unit
 Arithmetic Logic Unit
 Input/Output
o Read/write, random access memory is used to store both program
instructions and data
 Programinstructions are coded data which tell the computer
to do something
 Data is simply information to be used by the program
o Control unit fetches instructions/data from memory, decodes the
instructions and then sequentially coordinates operations to
accomplish the programmed task.
o Aritmetic Unit performs basic arithmetic operations
o Input/Output is the interface to the human operator

Flynn's Classical Taxonomy
 One of the more widely used classifications, in use since 1966, is called
Flynn's Taxonomy.
 Flynn's taxonomy distinguishes multi-processor computer architectures
according to how they can be classified along the two independent
dimensions of Instruction and Data. Each of these dimensions can have
only one of two possible states: Single orMultiple.
 The matrix below defines the4 possible classifications according to Flynn:
S I S D
Single Instruction, Single Data
S I M D
Single Instruction, Multiple Data
M I S D
Multiple Instruction, Single Data
M I M D
Multiple Instruction, Multiple Data
Single Instruction, Single Data (SISD):
 A serial (non-parallel) computer
 Single Instruction: Only oneinstruction streamis being acted on by the
CPU during any one clock cycle
 Single Data: Only one data stream is being used as input during any
one clock cycle
 Deterministic execution
 This is the oldest and even today, the most common type of computer

IBM 360
 Examples: older generation mainframes, minicomputers and
workstations; most modern day PCs.
PDP1 Dell Laptop
Single Instruction, Multiple Data (SIMD):
 A type of parallel computer
 Single Instruction: All processing unitsexecutethesame instruction at any
given clock cycle
 Multiple Data: Each processing unit can operate on a different data
element
 Best suited for specialized problems characterized by a high degree of

regularity, such as graphics/image processing.
 Synchronous (lockstep) and deterministic execution
 Two varieties: Processor Arrays and Vector Pipelines
 Examples:
o Processor Arrays: ConnectionMachine CM-2, MasPar MP-1 & MP-
2, ILLIAC IV
o Vector Pipelines: IBM 9000, Cray X-MP, Y-MP & C90, Fujitsu VP,
NEC SX-2, Hitachi S820, ETA10
 Most modern computers, particularly those with graphics processor units
(GPUs) employ SIMD instructions and execution units.
ILLIAC IV MasPar
Cray X-MP Cray Y-MP Thinking Machines CM-
Multiple Instruction, Single Data (MISD):

 Multiple Instruction: Each processing unit operates on the data
independently via separate instruction streams.
 Single Data: A single data stream is fed into multiple processing units.
 Few actual examples of this class of parallel computer have ever existed.
One is the experimental Carnegie-Mellon C.mmp computer (1971).
 Some conceivable uses might be:
o multiple frequency filters operating on a single signal stream
o multiple cryptography algorithms attemptingtocrack a single coded
message.
Multiple Instruction, Multiple Data (MIMD):
 Multiple Instruction: Every processor may be executing a different
instruction stream
 Multiple Data: Every processor may be working with a different data
stream
 Execution can be synchronous or asynchronous, deterministic or non-
deterministic
 Currently, the most common type of parallel computer - most modern
supercomputers fall into this category.
 Examples: most current supercomputers, networked parallel computer
clusters and "grids", multi-processor SMP computers, multi-core PCs.
 Note: many MIMD architectures also include SIMD execution sub-
components

IBM POWER5 HP/Compaq Alphaserver Intel
AMD Opteron Cray XT3 IBM

Some General Parallel Terminology
Like everything else, parallel computing has its own "jargon". Some of the more
commonly used terms associated with parallel computing are listed below. Most
of these will be discussed in more detail later.
Supercomputing / High Performance Computing (HPC)
Using the world's fastest and largest computers to solve large problems.
Node
A standalone "computer in a box". Usually comprised of multiple
CPUs/processors/cores. Nodes are networked together to comprise a
supercomputer.
CPU / Socket / Processor / Core
This varies, depending upon who you talk to. In the past, a CPU (Central
Processing Unit) was a singular execution component for a computer.
Then, multipleCPUs were incorporated intoa node. Then, individual CPUs
were subdivided intomultiple "cores", each being a unique execution unit.
CPUs with multiple cores are sometimes called "sockets" - vendor
dependent. The result is a node with multiple CPUs, each containing
multiple cores. The nomenclature is confused at times. Wonder why?
Task
A logically discrete section of computational work. A task is typically a
program or program-like set of instructions that is executed by a
processor. A parallel program consists of multiple tasks running on
multiple processors.
Pipelining
Breaking a task into steps performed by different processor units, with
inputs streaming through, much like an assembly line; a type of parallel
computing.
Shared Memory

From a strictly hardwarepoint of view, describes a computer architecture
where all processors have direct (usually bus based) access to common
physical memory. In a programming sense, it describes a model where
parallel tasks all have the same "picture" of memory and can directly
address and access the same logical memory locations regardless of where
the physical memory actually exists.
Symmetric Multi-Processor (SMP)
Hardware architecture where multiple processors share a single address
space and access to all resources; shared memory computing.
Distributed Memory
In hardware, refers tonetwork based memory access for physical memory
that is not common. As a programmingmodel, tasks can only logically "see"
local machinememory and must use communications toaccess memory on
other machines where other tasks are executing.
Communications
Parallel tasks typically need to exchange data. There are several ways this
can be accomplished, such as through a shared memory bus or over a
network, however theactualevent of data exchange is commonly referred
to as communications regardless of the method employed.
Synchronization
The coordination of parallel tasks in real time, very often associated with
communications. Often implemented by establishing a synchronization
point within an application where a task may not proceed further until
another task(s) reaches the same or logically equivalent point.
Synchronization usually involves waiting by at least one task, and can
therefore cause a parallel application's wall clock execution time to
increase.
Granularity

In parallel computing, granularity is a qualitative measure of the ratio of
computation to communication.
 Coarse: relatively large amounts of computational work are done
between communication events
 Fine: relatively small amounts of computational work are done
between communication events

3.Parallel Computer Memory Architectures
Shared Memory
General Characteristics:
 Shared memory parallel computers vary widely, but generally have
in common the ability for all processors toaccess all memory as
globaladdress space.
 Multipleprocessors can operateindependently but sharethesame
memory resources.
 Changes in a memory location effected by one processor are visible
to all other processors.
 Shared memory machines can be divided intotwo main classes
based upon memory access times: UMA and NUMA.
Shared
Memory
(UMA)
Shared
Memory
(NUMA)

Uniform Memory Access (UMA):
Most commonly represented today by Symmetric Multiprocessor
(SMP) machines
 Identical processors
 Equalaccess and access times to memory
 Sometimes called CC-UMA- CacheCoherent UMA. Cache
coherent means if one processor updates a location in shared
memory, all the other processors knowabout theupdate. Cache
coherency is accomplished at thehardwarelevel.
Non-Uniform Memory Access (NUMA):
 Often made by physically linking twoor more SMPs
 One SMP can directly access memory of another SMP
 Not all processors have equalaccess time to all memories
 Memory access across link is slower
 If cachecoherency is maintained, then may also be called CC-
NUMA - CacheCoherent NUMA
Advantages:

 Global address spaceprovides a user-friendly programmingperspectiveto
memory
 Data sharing between tasks is both fast and uniformdueto the proximity
of memory to CPUs
Disadvantages:
 Primary disadvantageis thelack of scalability between memory and CPUs.
Adding more CPUs can geometrically increases trafficon theshared
memory-CPU path, and for cache coherent systems, geometrically increase
traffic associated with cache/memory management.
 Programmer responsibility for synchronization constructs thatensure
"correct" access of global memory.
 Expense: it becomes increasingly difficult and expensiveto design and
produceshared memory machines with ever increasing numbers of
processors.
Distributed Memory
General Characteristics:
 Like shared memory systems, distributed memory systems vary widely but
share a common characteristic. Distributed memory systems require a
communication network to connect inter-processor memory.

 Processors have their own local memory. Memory addresses in one
processor do not map to another processor, sothereis no concept of global
address space across all processors.
 Because each processor has its own local memory, it operates
independently. Changes it makes to its local memory have no effect on the
memory of other processors. Hence, the concept of cache coherency does
not apply.
 When a processor needs access to data in another processor, it is usually
the task of the programmer to explicitly define how and when data is
communicated. Synchronization between tasks is likewise the
programmer's responsibility.
 The network "fabric" used for data transfer varies widely, though it can can
be as simple as Ethernet.
Advantages:
 Memory is scalablewith the number of processors. Increase the number of
processors and the size of memory increases proportionately.
 Each processor can rapidly access its own memory without interference
and without the overhead incurred with trying to maintain cache
coherency.
 Cost effectiveness: can use commodity, off-the-shelf processors and
networking.

Disadvantages:
 The programmer is responsiblefor many of the details associated with data
communication between processors.
 It may be difficult tomap existing data structures, based on globalmemory,
to this memory organization.
 Non-uniform memory access (NUMA) times
Hybrid Distributed-Shared Memory
 The largest and fastest computers in the world today employ both shared
and distributed memory architectures.
 The shared memory component can be a cache coherent SMP machine
and/or graphics processing units (GPU).
 The distributed memory component is the networking of multiple
SMP/GPU machines, which know only about their own memory - not the
memory on another machine. Therefore, network communications are
required to move data from one SMP/GPU to another.
 Current trends seemto indicatethat this type of memory architecture will
continue to prevail and increase at the high end of computing for the
foreseeable future.

4.Parallel programming model
Overview
 There are several parallel programming models in common use:
o Shared Memory (without threads)
o Threads
o Distributed Memory / Message Passing
o Data Parallel
o Hybrid
o Single Program Multiple Data (SPMD)
o Multiple Program Multiple Data (MPMD)
 Parallel programming models exist as an abstraction above hardware and
memory architectures.
 Although it might not seem apparent, these models are NOT specific to a
particular type of machine or memory architecture. In fact, any of these
models can (theoretically) be implemented on any underlying hardware.
Two examples from the past are discussed below.
o SHARED memory model on a
DISTRIBUTED memory machine: Kendall
Square Research (KSR) ALLCACHE
approach.
Machine memory was physically
distributed across networked machines,
but appeared to the user as a single shared
memory (global address space).
Generically, this approach is referred to as
"virtual shared memory".

o DISTRIBUTED memory model on a
SHARED memory machine: Message
Passing Interface (MPI) on SGI Origin
2000.
The SGI Origin 2000 employed the CC-
NUMA type of shared memory
architecture, where every task has direct
access to global address space spread
across all machines. However, the ability
to send and receive messages using MPI, as
is commonly done over a network of
distributed memory machines, was
implemented and commonly used.
 Which model to use? This is often a combination of what is available and
personal choice. There is no "best" model, although there certainly are
better implementations of some models over others.
 The following sections describe each of the models mentioned above, and
also discuss some of their actual implementations.
Shared Memory Model (without threads)
 In this programming model, tasks share a common address space, which
they read and write to asynchronously.
 Various mechanisms such as locks / semaphores may be used to control
access to the shared memory.
 An advantageof this model from the programmer's point of view is that the
notion of data "ownership" is lacking, so there is no need to specify

explicitly the communication of data between tasks. Program development
can often be simplified.
 An important disadvantage in terms of performance is that it becomes
more difficult to understand and manage data locality.
o Keeping data local to the processor that works on it conserves
memory accesses, cache refreshes and bus traffic that occurs when
multiple processors use the same data.
o Unfortunately, controlling data locality is hard to understand and
beyond the control of the average user.
Implementations:
 Native compilers and/or hardware translate user program variables into
actualmemory addresses, which areglobal. On stand-aloneSMP machines,
this is straightforward.
 On distributed shared memory machines, such as the SGI Origin, memory
is physically distributed across a network of machines, but made global
through specialized hardware and software.
Threads Model
 This programming model is a type of shared memory programming.
 In the threads model of parallel programming, a single process can have
multiple, concurrent execution paths.

 Perhaps the most simple
analogy that can be used to
describe threads is the
concept of a singleprogram
that includes a number of
subroutines:
o The main
program a.out is
scheduled to run by
the native operating
system. a.out loads and acquires allof the necessary system and user
resources to run.
o a.out performs some serial work, and then creates a number of tasks
(threads) that can be scheduled and run by the operating system
concurrently.
o Each thread has local data, but also, shares the entire resources
of a.out. This saves the overhead associated with replicating a
program's resources for each thread. Each threadalsobenefits from a
global memory view because it shares the memory space of a.out.
o A thread's work may best be described as a subroutine within the
main program. Any thread can execute any subroutine at the same
time as other threads.
o Threads communicate with each other through global memory
(updating address locations). This requires synchronization
constructs to ensure that more than one thread is not updating the
same global address at any time.
o Threads can come and go, but a.out remains present to provide the
necessary shared resources until the application has completed.
Implementations:

 From a programming perspective, threads implementations commonly
comprise:
o A library of subroutines that are called from within parallel source
code
o A set of compiler directives imbedded in either serial or parallel
source code
In both cases, the programmer is responsible for determining all
parallelism.
 Threaded implementations are not new in computing. Historically,
hardware vendors have implemented their own proprietary versions of
threads. These implementations differed substantially from each other
making it difficult for programmers to develop portable threaded
applications.
 Unrelated standardization efforts have resulted in two very different
implementations of threads: POSIX Threads and OpenMP.
 POSIX Threads
o Library based; requires parallel coding
o Specified by the IEEE POSIX 1003.1c standard (1995).
o C Language only
o Commonly referred to as Pthreads.
o Most hardware vendors now offer Pthreads in addition to their
proprietary threads implementations.
o Very explicit parallelism; requires significant programmer attention
to detail.
 OpenMP
o Compiler directive based; can use serial code
o Jointly defined and endorsed by a group of major computer
hardware and software vendors. The OpenMP Fortran API was
released October 28, 1997. TheC/C++ API was released in late 1998.

o Portable / multi-platform, including Unix and Windows NT
platforms
o Available in C/C++ and Fortran implementations
o Can be very easy and simple to use - provides for "incremental
parallelism"
 Microsoft has its own implementation for threads, which is not related to
the UNIX POSIX standard or OpenMP.
Distributed Memory / Message Passing Model
This model demonstrates the following characteristics:
o A set of tasks that use their own local memory during computation.
Multiple tasks can reside on the same physical machine and/or
across an arbitrary number of machines.
o Tasks exchange data through communications by sending and
receiving messages.
o Data transfer usually requires cooperative operations to be
performed by each process. For example, a send operation must have
a matching receive operation.
Implementations:
 From a programming perspective, message passing implementations
usually comprise a library of subroutines. Calls to these subroutines are
imbedded in source code. The programmer is responsible for determining
all parallelism.
 Historically, a variety of message passing libraries havebeen available since
the 1980s. These implementations differed substantially from each other
making it difficult for programmers to develop portable applications.

 In 1992, theMPI Forumwas formed with the primary goal of establishing
a standard interface for message passing implementations.
 Part 1 of the Message Passing Interface (MPI) was released in 1994. Part 2
(MPI-2) was released in 1996. Both MPI specifications are available on the
web at http://www-unix.mcs.anl.gov/mpi/.MPI is now the "de facto"
industry standard for message passing, replacing virtually all other
message passing implementations used for production work. MPI
implementations exist for virtually all popular parallel computing
platforms. Not all implementations include everything in both MPI1 and
MPI2.
Data Parallel
Model
 The data parallel
model demonstrates
the following
characteristics:
o Most of the parallel work focuses on performing operations on a
data set. The data set is typically organized intoa common structure,
such as an
array or
cube.
o A set of
tasks
work
collectiv
ely on
the same

data structure, however, each task works on a different partition of
the same data structure.
o Tasks perform the same operation on their partition of work, for
example, "add 4 to every array element".
 On shared memory architectures, all tasks may have access to the data
structure through global memory. On distributed memory architectures
thedata structureis split up and resides as "chunks" in the local memory of
each task.
Implementations:
 Programming with the data parallel model is usually accomplished by
writing a programwith data parallelconstructs. Theconstructs can becalls
to a data parallelsubroutinelibraryor, compiler directives recognized by a
data parallel compiler.
 Fortran 90 and 95 (F90, F95): ISO/ANSI standard extensions to Fortran 77.
o Contains everything that is in Fortran 77
o New source code format; additions to character set
o Additions to program structure and commands
o Variable additions - methods and arguments
o Pointers and dynamic memory allocation added
o Array processing (arrays treated as objects) added
o Recursive and new intrinsic functions added
o Many other new features
Implementations are available for most common parallel platforms.
 High Performance Fortran (HPF): Extensions to Fortran 90 to support data
parallel programming.
o Contains everything in Fortran 90

o Directives to tell compiler how to distribute data added
o Assertions that can improve optimization of generated code added
o Data parallel constructs added (now part of Fortran 95)
HPF compilers were relatively common in the 1990s, but are no longer
commonly implemented.
 Compiler Directives:Allow the programmer tospecify the distribution and
alignment of data. Fortran implementations areavailablefor most common
parallel platforms.
 Distributed memory implementations of this model usually require the
compiler to produce object code with calls to a message passing library
(MPI) for data distribution. All message passing is done invisibly to the
programmer.
Hybrid Model
 A hybrid
model
combines
more
than one
of the
previousl
y
described
program
ming models.

 Currently, a common example of a hybrid model is the combination of the
message passing model (MPI) with the threads model (OpenMP).
o Threads performcomputationally intensive kernels using local, on-
node data
o Communications between processes on different nodes occurs over
the network using MPI
 This hybrid model lends itself well to the increasingly common hardware
environment of clustered multi/many-core machines.
 Another similar and increasingly popular example of a hybrid model is
using MPI with GPU (Graphics Processing Unit) programming.
o GPUs perform computationally intensive kernels using local, on-
node data
o Communications between processes on different nodes occurs over
the network using MPI
SPMD and MPMD
Single Program Multiple Data (SPMD):
 SPMD is actually a "high level" programming modelthat can be built upon
any combination of
the previously
mentioned parallel
programming
models.
 SINGLE PROGRAM:

All tasks execute their copy of the same program simultaneously. This
program can be threads, message passing, data parallel or hybrid.
 MULTIPLE DATA: All tasks may use different data
 SPMD programs usually have thenecessary logic programmed intothemto
allow different tasks tobranch or conditionally execute only those parts of
theprogramthey are designed to execute. That is, tasks do not necessarily
have to execute the entire program - perhaps only a portion of it.
 The SPMD model, using message passing or hybrid programming, is
probably themost commonly used parallel programming model for multi-
node clusters.
Multiple Program Multiple Data (MPMD):
 Like SPMD, MPMD is actually a "high level" programming model that can
be built upon any combination of the previously mentioned parallel
programming models.
 MULTIPLE PROGRAM: Tasks may execute different programs
simultaneously. The programs can be threads, message passing, data
parallel or hybrid.
 MULTIPLE DATA: All tasks may use different data
 MPMD applications arenot as common as SPMD applications, but may be
better suited for certain types of problems, particularly those that lend
themselves better to functionaldecomposition than domain decomposition
(discussed later under Partioning).

5.References
 Author: Blaise Barney, Livermore Computing.
 A search on the WWW for "parallel programming" or "parallel computing"
will yield a wide variety of information.
 Recommended reading:
o "Designing and Building Parallel Programs". Ian Foster.
http://www-unix.mcs.anl.gov/dbpp/
o "IntroductiontoParallel Computing". Ananth Grama, Anshul Gupta,
George Karypis, Vipin Kumar.
http://www-users.cs.umn.edu/~karypis/parbook/
o "Overview of Recent Supercomputers". A.J. van der Steen, Jack
Dongarra.
OverviewRecentSupercomputers.2008.pdf
 Photos/Graphics have been created by the author, created by other LLNL
employees, obtained from non-copyrighted, government or public domain
(such as http://commons.wikimedia.org/) sources, or used with the
permission of authors from other presentations and web pages.
 History: These materials have evolved from the following sources, which
are no longer maintained or available.
o Tutorials located in the MauiHigh PerformanceComputing Center's
"SP Parallel Programming Workshop".
o Tutorials located at the Cornell Theory Center's "Education and
Training" web page.

Introduction to parallel computing

Introduction to parallel computing

More Related Content

What's hot (20)

Viewers also liked (8)

Similar to Introduction to parallel computing (20)

More from VIKAS SINGH BHADOURIA (20)

Recently uploaded (20)

Introduction to parallel computing