Bruce Jacob - - http://www.ece.umd.edu/~blj/

Overview of Our Work (and FAQ)

A surprising number of researchers and designers pay little attention to the memory system, despite the fact that the memory system is typically the dominant source of both performance loss and power dissipation within a given computer system. Most academic research in computer-systems design still focuses on the microprocessor, even though the processor has not actually been a significant bottleneck for over a decade. Our research group is arguably the best academic group in the world looking at memory systems: it is our primary focus. We have developed one of the world's most accurate and modular memory-system design frameworks, currently in wide use in both industry and academia ... as a result we have deep insight into the workings of the memory system, arguably better than any other group in the world, save for a handful of very specialized groups in industry. The few companies who pay significant attention to memory-systems performance come to our group: they hire our students, they ask our advice, they use our software to design their systems, and one uses our memory-controller design in their next-generation supercomputer. Our goals include detailed understanding of the memory system, deep characterization of its behavior, exploration of novel mechanisms and technologies, and development of extremely efficient next-generation systems from very large scale to memory-systems-on-chip.

Computational Artifacts

Workloads and Their Characterization

The Studies

Brought to you by The Memory-Systems Research Consortium