Dr. Kogge was with
His current research areas include massively parallel processing architectures, advanced VLSI and nano technologies and their relationship to computing systems architectures, non von Neumann models of programming and execution, parallel algorithms and applications, and their impact on computer architecture. Since the late 1980s' this has focused on scalable single VLSI chip designs integrating both dense memory and logic into "Processing In Memory" (PIM) architectures, efficient execution models to support them, and scaling multiple chips to complete systems. This includes not only efficient parallel processing topologies, control strategies, and chip floor plans, but doing so with inherently low power CPU architectures, and for a range of real system applications from highly scalable deep space exploration to exa-scale level supercomputing. Special emphasis has been on alternative models of massive light and ultra-light weight multi-threading.
Other recent work has focused on how PIM-like ideas may port into nanotechnology such as quantum dot cellular array (QCA), where instead of "Processing-In-Memory" we have opportunities for "Processing-In-Wire" and similar paradigm shifts. A key part of this is focusing on the interchange between the underlying device physics, the design rules and metaphors best suited to using such devices, and how to recast "conventional" computing structures into such design metaphors in ways that optimize the overall system.
Most recently, Dr. Kogge led a DARPA-sponsored group of industry and academic technical experts in developing a detailed projection of the technical challenges needed to advance the art of supercomputing to the exascale level (1000 times today's emerging petascale computers). Both a report and an IEEE radio interview are available.
Since coming to Notre Dame, Dr. Kogge has participated on a long series of
projects aimed at achieving new levels of supercomputing performance by
leveraging both technology and architecture. This includes proposing one of the
three "petascale" architectures that emerged from the "Enabling
Technologies for Petascale Computing" workshop, held in Feb. of 1994.
These PIM-based concepts were developed through a long series of petascale
development projects including HTMT and Cray's Cascade project, and in
collaboration with Dr. Jay Brockman, PIM
Earlier work with a student of his Victor Zyuban
His Ph.D. thesis on the parallel solution of recurrence equations was one of the early works on what is now called parallel prefix operations, and applications of those results are still acknowledged as defining the fastest possible implementations of circuits such as adders with limited fan-in logic (known as the Kogge-Stone adder). His 1982 book, “The Architecture of Pipelined Computers” is widely regarded as perhaps the first formal treatise on this now-ubiquitous technique.