[FPBench] Community Meeting

[Ian Briggs]

Hi everyone!

We are happy to welcome Rémi Garcia this Thursday, February 1, to give a talk on the hardware side of floating point. He is a recent doctoral graduate of University of Nantes, and is now a postdoctoral researcher at the University of Rennes.

In his own words, this talk will be about:

Multiplication is a basic operator used in many applications. When implemented in embedded systems, e.g. FPGAs, these algorithms require highly optimized hardware. Improving the multiplication implementation is an important part of the final hardware cost reduction. This talk will expose the recent advances on the automatic design of multiplication of a variable with multiple a priori known constants, this problem is called the Multiple Constant Multiplication (MCM) problem. Using Integer Linear Programming (ILP) permits to significantly reduce the hardware cost when MCM is implemented using additions and bit-shifts. The ILP approach has been efficiently used for various hardware design problems but other approaches exists, such as SAT/SMT, Constraint Programming, etc. This presentation will introduce a few concepts relating to this ecosystem.

For those searching for the Zoom link, here it is: https://washington.zoom.us/j/9283133132

Hope to see you there,

Ian Briggs

[Ian Briggs]

Reminder for the community meeting!

We will be hearing from Rémi Garcia tomorrow at 9:00 AM pacific (thanks for the time reminder Martin!)

Thanks,

Zach and Ian