A Complete Weather Forecasting Suite On Param 10000

A COMPLETE WEATHER
FORECASTING SUITE ON PARAM

10000
Presented by-
Sukanta pal
Rgdn. No-0701106280
branch-IT
1
Agenda
 Introduction.
 Need for Weather Forecasting.
 PARAM 10000.
 Design Approach.
 Numerical algorithm.
 The Weather Forecasting Capabilities.
 Accuracy Issue And Fault Tolerance
 Conclusion.
 References
2
INTRODUCTION
 What is weather forecasting?
Prediction of atmospheric state for a given time and location
 Why to forecast?
 Air traffic
 Agriculture
 Marine
 Private sector
 Forestry
 Military application
3
Methods of weather forecast
Synoptic
Statistical
Numerical
4
Numerical Forecasting
 Computer simulation of atmospheric condition.
 It includes
i. Data collection
ii. Data assimilation and analysis
iii. Numerical weather prediction
iv. Model output post processing
5
PARAM 10000
 PARAM is a series of supercomputers developed by CDAC.

 It is India’s first Teraflop Supercomputer.
 The major applications of PARAM 10000 includes
1. long-range weather forecasting
2. remote sensing
3. drug design and
4. molecular modelling.
6
Design approach
7
 Based on cluster models of SMPs
 Has 1gb of RAM per node.
 Message passing communication software like MPI and PVM.
 Low latency, high bandwidth supported communication software
KSHIPRA.
 Lightweight communication protocols on the PARAMNET
and MYRINET
8
9
Numerical Algorithm
 Derived from conservation laws of mass, momentum, and energy.
 Two conditions .
 If it is a linear quantity, then the representations are straight
forward.
 If it is a non-linear term, spectral transform methods have to be used.
The computations for spectral algorithms consists of following steps
10
u    u  z  p  
M m
m m
j , Zk n k j
n
n  | m|
u  l ,  j , z k    
M
 u
n M
 j , z k e iml
11
u m
 j , zk  
1

k 2 1
l 0
u  l ,  j , z k e iml
k2
 zk    j 1   j u   j , zk  pn   j 
k1 m
m m
u n
12
Weather Forecasting
capabilities
1. Parallelisation Strategy
 Involves of Identification of code and subsequent compute intensive

parts and their data dependencies
 Latitude is taken as the most obvious and clear independent data for
computation work in this algorithm.
 Major computation is done through the use of Fourier and Legendre
transform.
13
 A pair of latitudes is placed on one processor
 Each processor computes Fourier coefficients using the inverse

Legendre transform and the gaussian grid point values using the inverse
Fourier transform
 Partial sum of the Legendre transform is performed on each

processor for given latitudes.
 Then these sums are circulated among all the processors to have
global sum on each of the processor
14
2 Implementation
 CRAY assembly language was replaced by the Temperton's Fortran

algorithm.
 The important components of this code are the
I. I/O, the forecast loop(GLOOPA), the radiation loop (GLOOPR) which is
called once in twelve hours.
II. Physics loop (GLOOPB), FFT and Legendre transform. GLOOPA and
GLOOPB .
 The data distribution of parallel T80 is along the latitudes distributed
over the processors. This is a master-workers model
15
16
Accuracy Issue
 Performance strongly linked to the architectural features like clock
speed, main memory bandwidth.
 Message sending proportional to message length but independent of

node location and network traffic.
 Four cpus in a single node responsible for low latency of message

passing between the processors.
17
Fault Tolerance
 Independent Processing can be run independentently using an
interface.
 Fault tolerance save re-running of the codes from system failures.
 System failure recovered by restarting JCL.
18
Conclusion
 Parallel Climate models T2l/T42 and Forecast models T80/T126
have been developed.
 Achieve fault tolerance and accuracy.
 Access to greater number of cpus thereby enhancing scalability.
 Give better performances than before.
19
References
 http://www.iitr.ac.in/centers/CC/pages/Resources+Pa
ram_1000.html
 http://www.cdac.in/html/press/3q09/spot687.aspx
 http://pune.cdac.in/html/secg/cas.aspx
 http://pune.cdac.in/html/npsf/npsfidx.aspx
 http://pune.cdac.in/html/npsf/param10k/benchmp.as
px
 http://www.buyya.com/papers/cdacrep.txt
 http://en.wikipedia.org/wiki/Remote_direct_memory
_access
20
THANK YOU
21

A Complete Weather Forecasting Suite On Param 10000

Uploaded by

Copyright:

Available Formats

A Complete Weather Forecasting Suite On Param 10000

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Complete Weather Forecasting Suite On Param 10000

Uploaded by

Copyright:

Available Formats

A COMPLETE WEATHER

FORECASTING SUITE ON PARAM

 PARAM is a series of supercomputers developed by CDAC.

 Low latency, high bandwidth supported communication software

 Lightweight communication protocols on the PARAMNET

 Involves of Identification of code and subsequent compute intensive

 Each processor computes Fourier coefficients using the inverse

 Partial sum of the Legendre transform is performed on each

 CRAY assembly language was replaced by the Temperton's Fortran

 Message sending proportional to message length but independent of

 Four cpus in a single node responsible for low latency of message

 Fault tolerance save re-running of the codes from system failures.

 System failure recovered by restarting JCL.

 Achieve fault tolerance and accuracy.

 Access to greater number of cpus thereby enhancing scalability.

 Give better performances than before.

You might also like