For prospective graduate/undergradeuate students:

You are more than welcome to drop by our lab and discuss about the research related topics after you read the brief introduction of our projects and still be interested in it.

We are currently devoted ourselves into the following two research topics:

1. Performance Improvement

The major goal of this project is to improve the performance of the storage system. We try to accelerate the speed of data transmission and reduce the miss rate of picking data.

To achieve this goal, we will outline the overall performance characterization. After that, we try to identify where the performance bottlenecks are and reduce the harm that the bottlenecks cause as possible as we can. Just like the graph below, we try and do the experiment iteratively to arrive optimal situation that the storage system can run more fast and more stable.

Additionally, because of the use of the widespread hardware now, we will port the driver on X86 and XScale and do the same test on these two platform.

Moreover, we will also attempt to improve the design for the next-generation products by simulating. Like maybe we can simulate that there are more CPUs on system and still ensure the correctness and reliability.

:public:circle.jpg

2. Software Modeling

We play an important agent role between hardware providers and TeleComs, in order to convince TeleComs that they could buy the hardware machine. Our goal for this hardware project is to develop a tool that could model all the TeleComs Apps in the world, so that the TeleComs would use the tool to generate relative code, they will be able to understand the situation that the relative code run on the hardware machine without their source code.

Since the behavior of the TeleComs Apps is very large, we decide to choose one instance of behavior of TeleComs Apps to survey it. Now we start to survey the behavior of SIP application to profile the performance metrics and find out basic operations, then using the tool, Metropolis, to model the code behavior of SIP. So we can get the systemC code generated by Metropolis to run on the hardware machine. If the result is as more similar as the result that the SIP run on the hardware machine, then the tool has very high reliability. This is the short-term goal of the hardware project.

In the future, we will accede to more instances of behavior of the TeleComs Apps as possible as we can, so that we could make the tool be more complete. Therefore the tool will have more high reliability to increase the motivation of the TeleComs to buy the hardware machine.

:public:chart_.jpg

 
public/research.txt · Last modified: 2010/05/22 09:20 (external edit)
 
Recent changes RSS feed Creative Commons License Donate Powered by PHP Valid XHTML 1.0 Valid CSS Driven by DokuWiki