1.
Dhammpal Ramtake
SoS in Computer sciences & IT
Pt. R.S.U. Raipur, (C.G)
Guided by
Dr. Sanjay Kumar
Dr. Vinod Kumar Patle
2.
Contents
INTRODUCTION OF MATLAB
INTRODUCTION OF PARALLEL COMPUTING
PARALLEL COMPUTING WITH MATLAB
MATLAB FOR MULTI CORE SYSTEM
MATLAB FOR DISTRIBUTED COMPUTING SERVER
PARALLEL COMANDS IN MATLAB
TEST THE EFFICINCY OF PARALLEL CODE
CONCLUSION
3.
Introduction
• MATLAB is a high-level technical computing language and interactive
environment for algorithm development, data visualization, data analysis, and
numeric computation. Using the MATLAB we can solve computing problems
faster than with traditional programming languages, such as C, C++, and
Fortran.
• We can use MATLAB in a wide range of applications, including signal and
image processing, communications, control design, test and
measurement, financial modeling and analysis ,computational biology and
parallel computing.
5.
Generally, computer code is written in serial
One task completed after another until the script is finished with
only one task completing at each time
Because computer only has single processing unit .
What is Parallel Computing?
6.
What is Parallel Computing? (cont.)
Parallel Computing:
Using multiple computer processing units (CPUs) to solve a problem at the same
time.
computer with multiple processors or networked computers(Distributed
computing )
7.
PARALLEL MATLAB
Single Multi Core
CPU
Distributed Computing
Server
Number of core on
single machine as a
worker to execute
a task parallel like
OpenMP
Client machine having
there core which is take as
workers in network with
central control of server .
PARALLEL COMPUTING WITH MATLAB
MATLAB provide Parallel computing tool for Distributed Computing Server as
well as Single desktop .
8.
MATLAB FOR MULTI CORE SYSTEM
MATLAB Provides workers (MATLAB computational engines) to
execute applications on a multi core system.
We can write a commands that will be executed in parallel or call
an MATLAB script file that will run in parallel
Fig:- MATLAB workers for multi core processor
9.
MATLAB FOR DISTRIBUTED COMPUTING SERVER
The Distributed Computing Server controls parallel execution of MATLAB on a
cluster with tens or hundreds of cores
With a cluster running parallel MATLAB, we can submit an Matlab file from a client, to
run on the cluster or we can submit an Matlab file to be executed in “batch"
Fig:- MATLAB Distributed Computing Server
10.
PARALLEL COMMANDS IN MATLAB
findResource
Code performaces commands
tic & toc , cputime,profile viewer etc.
Matlabpool
parfor (for loop)
pmode
spmd (distributed computing for datasets)
batch jobs (run job in background)
11.
1. FIND RESOURCE
This command give available parallel computing resources
Syntax
out = findResource()
out =findResource('scheduler','configuration','ConfigurationName')
12.
Code Performance Commands
Use MATLAB’s tic & toc functions tic starts a timer toc tells you the number of
seconds since the tic function was called.
1. TIC & TOC
Syntax
tic
ticID = tic
toc
toc(ticID)
elapsedTime = toc
Example :-
13.
2. CPU TIME
This command returns the total CPU time (in seconds) used by MATLAB
application from the time it was started
Syntax
cputime
Out= cputime
Example
14.
1. PROFILE VIEWER
This command gives profile records information about execution time, number of
calls, parent functions, child functions, code line hit count, and code line
execution time.
Syntax: -
profile on;
profile off;
profile resume;
profile clear;
profile viewer;
Example:-
16.
MATLABPOOL
This command open or close pool of MATLAB sessions for parallel computation .It
starts a worker pool using the default parallel configuration, with the pool size
specified by configuration.
Syntax
matlabpool
matlabpool open
matlabpool open poolsize
matlabpool open configname
matlabpool close
matlabpool close force
matlabpool close force configname
Example:-
17.
Request for too many workers, get an error
only request 2 workers on this machine!
18.
Matlabpool Close
Use matlabpool close to end parallel session
Options
matlabpool close force
deletes all pool jobs for current user in the cluster specified by default
profile (including running jobs)
matlabpool close force <profilename>
deletes all pool jobs run in the specified profile
19.
Parallel for Loops (parfor)
parfor loops can execute for loop like code in parallel to significantly improve
performance
Must consist of code broken into discrete parts that can be solved simultaneously (i.e. it
can’t be serial)
Matlab workers evaluate iterations in no particular order and independently
of each other.
Syntax
parfor loopvar = initval:endval; statements; end
parfor (loopvar = initval:endval, M); statements; end
20.
Parfor example
work in parallel loop increments are not dependent on each
other:
Makes the loop run in
parallel
21.
Test the efficiency of parallel code
Observed speedup of a code which has been parallelized
defined as :-
Execution time of Serial code
Execution time of parallel code
Speedup =
One of the simplest and widely used speedup formula for parallel programs
performances.
For the testing the efficiency of parfor loop we considered prime number
finding program .which limits is increasing ordered like 10 ,100,1000,10000
and 100000 iteration
22.
Simple for loop program :-
function [ total ] = simpleprime( n )
%SIMPLEPRIME Summary of this
function goes here
% Detailed explanation goes here
total = 0 ;
tic
for i = 2 : n
prime = 1 ;
for j = 2 : i-1
if( mod ( i , j ) == 0 )
prime = 0 ;
end
end
total = total + prime ;
end
toc
return
end
n=10;
while ( n <= 10000)
primes = newprime( n ) ;
fprintf( 1 , ' %8d %8d n ' , n
, primes ) ;
n = n * 10 ;
Function call
24.
parallel for (parfor)loop program :-
matlabpool open local 2
n=10;
while ( n <= 100000)
primes = newprime( n ) ;
fprintf( 1 , ' %8d %8d n ' , n
, primes ) ;
n = n * 10 ;
end
matlabpool close
function [ total ] = newprime( n )
%NEWPRIME Summary of this
function goes here
% Detailed explanation goes here
total = 0 ;
tic
parfor i = 2 : n
prime = 1 ;
for j = 2 : i-1
if( mod ( i , j ) == 0 )
prime = 0 ;
end
end
total = total + prime ;
end
toc
return
end
Function call
26.
Problem size Serial
execution
time
Parallel
execution
time
speedup
10 0.529376 0.320184 1.6533
100 0.009848 0.063059 0.1561
1000 0.023988 0.067858 0.3535
10000 2.009835 1.168925 1.7193
100000 200.9593 109.9775 1.8272
Result Table :-
27.
Pmode
pmode allows the interactive parallel execution of MATLAB commands. pmode
achieves this by defining and submitting a parallel job, and opening a Parallel
Command Window connected to the labs running the job.
Syntax:-
pmode start
pmode start numlabs
pmode start conf numlabs
pmode quit
pmode exit
Example:-
28.
Pmode with labindex
while ( n <= 10000)
primes = newprime( n ) ;
fprintf( 1 , ' %8d %8d n ' , n
, primes ) ;
if labindex==1
n = n * 10 ;
else
n =n*20;
end
end
function [ total ] = newprime( n )
%NEWPRIME Summary of this function
goes here
% Detailed explanation goes here
total = 0 ;
tic
Parfor i = 2 : n
prime = 1 ;
for j = 2 : i-1
if( mod ( i , j ) == 0 )
prime = 0 ;
end
end
total = total + prime ;
end
toc
return
end
Function call
30.
Spmd command
Spmd “Single program multiple data” execute code in parallel on MATLAB pool
The "single program" aspect of spmd means that the identical code runs on
multiple labs.
The "multiple data" aspect means that even though the spmd statement runs
identical code on all labs, each lab can have different, unique data for that code.
Syntax
spmd, statements, end
spmd(n), statements, end
spmd(m, n), statements, end
For example, create a random matrix on three labs:
matlabpool open
spmd (2)
R = rand(4,4);
end
matlabpool close
32.
Conclusion
MATLAB Parallel computing toolbox has the large number of
functionality .Which is essay to understand and solve the complex
parallel computing problems .
Clipping is a handy way to collect and organize the most important slides from a presentation. You can keep your great finds in clipboards organized around topics.