NSERC USRA Topics

I am very interested in talking to prospective NSERC USRA students. Below is a list of possible topics. Send me email and we can arrange to meet in my office or by phone.
  1. Cloud Computing
  2. Programming on Graphical Processing Units (GPUs)

High Performance Computing on the Cloud

Description: Recent advances in Cloud computing, particularly Infrastructure as a service, offer an intreging alternaitve computing platform for high performance (parallel) sceintific applications. In this project we will explore using Amazon EC2 Cloud infrastructure as a platform for computational sciences applications in catastrope modeling and risk analytics.

Students need not start with any backgound in cloud computing, but should be good programmings and have an interest in systems and performance issues. They will have a chance to learn about the EC2 cloud, programming in Hadoop, and build applications on the Amazon Cloud.

Type: Mix of literature review, algorithm design, challenging systems oriented implementation, extensive performance evaluation.

Links:

  1. Amazon Cloud Services - http://aws.amazon.com/ec2/
  2. Apache Hadoop - http://wiki.apache.org/hadoop/
  3. Map reduce - http://www.willowgarage.com/map-reduce-machine-learning-multicore
  4. HPC and the Cloud - http://www.hpcinthecloud.com/

Programming on Graphical Processing Units (GPUs)

Description: While a performance oriented CPU like the Core i7 975 @ 4.0 GHz is only capable of about 60-70 Gigaflops, today's high-end GPU's are capable of multiple-teraflop performance (1 Teraflop = 1,000,000,000,000 (Trillion) floating point operations per second). The challenge is designing anglorithms and applications that can effective expoit this computing power. This project concerns the design and implementation of risk management simulation on GPUs.

Students need not start with any backgound in GPU computing, but should be good programmings and have an interest in low level programming (in C) and performance issues. They will have a chance to learn about the GPUs, CUDA, and risk simulations.

Type: Mix of literature review, algorithm design, challenging systems oriented implementation, extensive performance evaluation.

Links:

  1. What is GPU Computing? - http://www.nvidia.com/object/GPU_Computing.html
  2. GPU Programming Course - http://courses.cms.caltech.edu/cs101gpu/
  3. CUDA Zone - http://www.nvidia.com/object/cuda_home_new.html
  4. R + GPU - http://brainarray.mbni.med.umich.edu/Brainarray/Rgpgpu/