 |
|
Success Stories
|
Econometric Modeling
Scientists at Columbia University's Earth Institute are
using supercomputers to better understand the economic
impact of climate change on crops throughout the world,
particularly in developing countries. The team is developing
a complex statistical model that combines climate data,
and crop statistics, and economic factors, to accurate
assess how farming communities make risk assessments and
economic choices. These insights can help drive policy
decisions, and design of relief and insurance programs
spanning large land regions, potentially entire countries.
The team developed the models on desktop computers using
MATLAB®, but were unable to scale beyond a small geographic
area. |
|
|
Star-P enabled the researchers to tap the power of parallel
high-performance computers to develop and test statistical
models involving enormous volumes of data interactively.
They were able to run models too large for serial desktops,
without having to reprogram them in C, Fortran, and MPI.
And, they could continue working in their familiar and
interactive MATLAB® environment, exploring a broad set
of algorithms and model parameters.
|
|
 |
 Summary
& Metrics
- Development of complex
statistical model with large and
growing data sets
- No need to re-write
MATLAB® code in C, Fortran, MPI
- Models that were
impossible to run on the desktop can now be
run in under 2 minutes with an 8-processor SGI
Altix server
|
Back to Success Stories >
|
|
 |
| |
|
 |
|
|
|
Star-P Support
Blog
Email
Page
Star-P Brochure (PDF - 3MB)
