SC19 Proceedings

The International Conference for High Performance Computing, Networking, Storage, and Analysis

Poster 117: A New Polymorphic Computing Architecture Based on Fine-Grained Instruction Mobility

Authors: David Hentrich (Illinois Institute of Technology), Erdal Oruklu (Illinois Institute of Technology), Jafar Saniie (Illinois Institute of Technology)

Abstract: This is a summary of the base concepts behind David Hentrich’s May 2018 Ph.D. dissertation in Polymorphic Computing. Polymorphic Computing is the emerging field of changing the computer architecture around the software, rather than vice versa. The main contribution is a new polymorphic computing architecture. The key idea behind the architecture is to create an array of processors where a program’s instructions can be individually and arbitrarily assigned/mobilized to any processor, even during runtime. The key enablers of this architecture are a dataflow instruction set that is conducive to instruction migration, a microarchitectual block called an “operation cell” (“op-cell”), a processor built around the instruction set and the “op-cells”, and arrays of these processors.

Best Poster Finalist (BP): no

Poster: PDF
Poster summary: PDF

Back to Poster Archive Listing