Research

• Overview
• Computer Arithmetic
  • Debugging and verification of arithmetic operations.
• Digital Typography
  • AlQalam project.
• Publications
• Supervised theses

Connecting the parts of the puzzle to see the big picture

If we contemplate the design of large digital systems we find that it is mostly composed of four main parts:

Memories

are used for temporary storage of results (registers), for reducing the time delay of retrieving the information (caches), or as the main store of information (main memories).

Control logic blocks

handle the flow of information and assure that the circuit performs what is desired by the user.

Datapath blocks

are the real engine that performs the work. These are mainly circuits performing either some arithmetic or logic operation on the data.

Communication channels

are usually wires inside the chips, dedicated or shared buses between the chips, and networks between systems.

Our research focuses mainly on the third category and how it interacts with the remaining parts.

A good optimization of the arithmetic blocks results in an improved datapath which directly leads to a better overall design. However, in order to decide on the figure of merit for the optimization a good understanding of the whole system is necessary. Some system level research is necessary to guide the work at the arithmetic units level.

Embedded systems have some constraints that might be different from those in general purpose computers. The problems arising from those constraints must be contrasted to the large body of research previously developed for computer architecture. We should not re-invet the wheel if someone already did!

However, future circuits will be quite different from what we used to handle in the past few decades. The International Technology Roadmap for Semiconductors predicts that CMOS technology is coming to an end. We must be ready when this happens. We study how to implement the basic blocks in future technologies broadly known as Nanoelectronics to present possible solutions.

As a researcher, I publish my results in a variety of forms. It could be in journals, on a poster, as a computer presentation in a conference session, on a web page,... In all of these forms what is really important for me and my readers is the content itself. The content is my main concern when I write. The way it will look on the screen or on paper is a different matter that changes from one media to the other and that evolves as the technology matures.

I became interested in text processing and typesetting versus WYSIWYG (What You See Is What You Get) word processing because of this specific need: I want to concentrate on the message I am conveying. A WYSIWYG software is just a stumbling block forcing me to constantly think about the appearance rather than the message. On the other hand, in the text processing paradigm, the author writes and the typesetter makes it look professional regardless of the media used for the output.

So, to sum up, my focus is on the datapath of digital designs from the devices that might be needed to the systems that perform the users' requirements. In addition to that, I have a deep interest in how the written information is presented.