HIGH-PERFORMANCE

COMPUTING
• In high-performance computing systems, a pool of
processors (processor machines or central processing
units [CPUs]) connected (networked) with other
resources like memory, storage, and input and output
devices, and the deployed
software is enabled to run in the entire system of
connected components.
The processor machines can be of homogeneous
or heterogeneous type.
• Thus, examples of HPC include a small cluster of
desktop computers or personal computers (PCs) to
the fastest supercomputers.
 PARALLEL COMPUTING

• Parallel computing is also one of the facets of

HPC. Here, a set of processors work cooperatively
to solve a computational problem.
• These processor machines or CPUs are mostly of
homogeneous type. Therefore, this definition is
the same as that of HPC and is broad enough to
include supercomputers that have hundreds or
thousands of processors interconnected with
other
resources.
• One can distinguish between conventional (also
known as serial or sequential or Von Neumann)
computers and parallel computers in the way the
applications are executed.
• In serial or sequential computers, the following
apply:
• • It runs on a single computer/processor machine
having a single CPU.
• • A problem is broken down into a discrete series
of instructions.
• • Instructions are executed one after another
• In parallel computing, since there is simultaneous
use of multiple processor machines, the following
apply:
• It is run using multiple processors (multiple
CPUs).
• A problem is broken down into discrete parts that
can be solved concurrently.
• Each part is further broken down into a series of
instructions.
• Instructions from each part are executed
simultaneously on different processors.
• An overall control/coordination mechanism is
employed.
 DISTRIBUTED COMPUTING

• Distributed computing is also a computing system

that consists of multiple computers or processor
machines connected through a network, which can
be homogeneous or heterogeneous, but run as a
single system.
• The connectivity can be such that the CPUs in a
distributed system can be physically close together
and connected by a local network, or they can be
geographically distant and connected by a wide
area network.
• The goal of distributed computing is to make such
a network work as a single computer.
• Distributed computing systems are advantageous
over centralized systems, because there is a
support for the following characteristic features:
• 1. Scalability: It is the ability of the system to be
easily expanded by adding more machines as
needed, and vice versa, without affecting the
existing setup.
• 2. Redundancy or replication: Here, several
machines can provide the same services, so that
even if one is unavailable (or failed), work does
not stop because other similar computing
supports will be available.
 CLUSTER COMPUTING
• A cluster computing system consists of a set of the
same or similar type of processor machines connected
using a dedicated network infrastructure. All processor
machines share resources such as a common home
directory and have a software such as a message
passing interface (MPI) implementation installed to
allow programs to be run across all nodes
simultaneously.
• This is also a kind of HPC category
• The individual computers in a cluster can be referred
to as nodes.
• The reason to realize a cluster as HPC is due to the fact
that the individual nodes can work together to solve a
problem larger than any computer can easily solve.
 GRID COMPUTING
• The idea of grid computing is to make use of such
nonutilized computing power by the needy
organizations, and thereby the return on investment
(ROI) on computing investments can be increased.
• Thus, grid computing is a network of computing or
processor machines managed with a kind of software
such as middleware, in order to access and use the
resources remotely.
• The managing activity of grid resources through the
middleware is called grid services.
• Grid services provide access control, security, access
to data including digital libraries and databases, and
access to large-scale interactive and long-term
storage facilities
• Grid computing is more popular due to the
following reasons:
• • Its ability to make use of unused computing
power, and thus, it is a cost-effective solution
(reducing investments, only recurring costs)
• • As a way to solve problems in line with any
HPC-based application
• • Enables heterogeneous resources of computers
to work cooperatively and collaboratively to solve
a scientific problem
 CLOUD COMPUTING

• The computing trend moved toward cloud from

the concept of grid computing, particularly when
large computing resources are required to solve a
single
problem, using the ideas of computing power as
a
utility and other allied concepts.
• In cloud computing, the IT and business
resources, such as servers, storage, network,
applications, and processes, can be dynamically
provisioned to the user needs and workload.
 BIOCOMPUTING
• Biocomputing systems use the concepts of
biologically derived or simulated molecules (or
models) that perform computational processes in
order to solve a problem. The biologically derived
models aid in structuring the computer programs
that become part of the application. Biocomputing
provides the theoretical background and practical
tools for scientists to explore proteins and DNA
• Biocomputing shall, therefore, lead to a better
understanding of life and the molecular causes of
certain diseases.
 MOBILE COMPUTING
• In mobile computing, the processing (or computing)
elements are small (i.e., handheld devices) and the
communication between various resources is taking
place using wireless media.
• Mobile communication for voice applications (e.g.,
cellular phone) is widely established throughout the
world and witnesses a very rapid growth in all its
dimensions including the increase in the number of
subscribers of various cellular networks.
• Mobile computing–based applications are becoming very
important and rapidly evolving with various technological
advancements as it allows users to transmit data from
remote locations to other remote or fixed locations.
 QUANTUM COMPUTING

• Manufacturers of computing systems say that

there is a limit for cramming more and more
transistors into smaller and smaller spaces of
integrated circuits (ICs) and thereby doubling the
processing power about every 18 months
• This problem will have to be overcome by a new
quantum computing–based solution, wherein the
dependence is on quantum information, the rules
that govern the subatomic world.
• Quantum computers are millions of times faster
than even our most powerful supercomputers
today.
 OPTICAL COMPUTING
• Optical computing system uses the photons in
visible light or infrared beams, rather than electric
current, to perform digital computations.
• An electric current flows at only about 10% of the
speed of light. This limits the rate at which data can
be exchanged over long distances and is one of the
factors that led to the evolution of optical fiber.
• By applying some of the advantages of visible
and/or IR networks at the device and component
scale, a computer can be developed that can
perform operations 10 or more times faster than
conventional electronic computer.
 NANOCOMPUTING

• Nanocomputing refers to computing systems that

are constructed from nanoscale components. The
silicon transistors in traditional computers maybe
replaced by transistors based on carbon
nanotubes
• The successful realization of nanocomputers
relates to the scale and integration of these
nanotubes or components
• Researchers are working on all these issues to
bring nanocomputing a reality.
 NETWORK COMPUTING

• Network computing is a way of designing systems to

take advantage of the latest technology and maximize
its positive impact on business solutions and their ability
to serve their customers using a strong underlying
network of computing resources
• In any network computing solution, the client component
of a networked architecture or application will be with
the customer or client or end user, and in modern days,
they provide an essential set of functionality necessary
to support the appropriate client functions at minimum
cost and maximum simplicity
• Almost all the computing paradigms that were discussed
earlier are of this nature.