Playing with Polynomials and Precision

Add to your list(s) Download to your calendar using vCal

• David P Boland (Imperial College)
• Wednesday 12 January 2011, 16:00-17:00
• Level 9 Mahanakorn Lab., EE Dept..

If you have a question about this talk, please contact George A Constantinides.

Abstract: When accelerating an algorithm in hardware, the choice precision is often overlooked. Whilst this is traditionally done to allow a direct comparison with software, this significantly reduces the potential performance improvement. An alternative metric for fair comparison with software would be that the final output satisfies the same criterion as the software design, such as lying within a desired range or guaranteeing that the potential floating point error lying below some threshold; designing with this in mind would allow a hardware designer to explore the silicon area, performance, power and error trade-offs of using variable precision. In order to achieve this, a tool is needed to provide a designer a means to verify whether the chosen precision will satisfy the design criteria.

This talk will examine such tools, notably the more modern of these tools which make use of polynomials for range and precision analysis. It will then highlight problems with existing methods via the use of several simple examples, before demonstrating how simple techniques can solve many of these issues and lead to superior hardware designs.

This talk is part of the CAS Talks series.

This talk is included in these lists:

• Andrea Picciau's list
• CAS Talks
• Circuits and Systems Group: Internal Seminars
• Level 9 Mahanakorn Lab., EE Dept.

Note that ex-directory lists are not shown.