The Georgia Tech Computer Architecture group ("Comparch," for short) conducts research that is both diverse and leading edge. We strive to answer questions such as:
- How can we design future multi- and manycore architectures to have better performance, spend less power, and be easier to program—all at a lower cost?
- How can we design processors and memory systems for new technologies, especially since the traditional cadence of CMOS scaling is slowing down?
- How do we restructure the memory hierarchy to meet the demands of “big data” applications such as dynamically changing graphs, machine intelligence, and search?
- What should be the architectures of future GPUs?
- How do we help programmers efficiently create and debug their software for parallel architectures?
- How do we protect computer hardware against snooping attacks? And how do we monitor running programs to prevent attacks and ensure correctness?
- How to holistically embrace error and unreliability in the computing fabric and design general-purpose approximate computers?
- How to utilize characteristic of the emerging applications to specialize current and design next generation computing systems?
- How do we quickly simulate future computer architectures with current computers?
Because of this diversity of research topics and the quality of our faculty and students, virtually every top-tier computer architecture conference today has one or more papers from Georgia Tech Comparch. Our funding comes from both government sponsors—such as the National Science Foundation, Air Force Office of Sponsored Research, and Department of Energy—and industry sponsors like IBM, Intel, AMD, Google, Qualcomm, Microsoft, and Samsung.
At the undergraduate level, we offer standard courses such as Processor Design and Advanced Computer Architecture. At the graduate level, our classes include High Performance Computer Architecture, Advanced Microarchitecture, Reliability and Security in Computer Architecture, and Alternative Computing Technologies. This listing updates continuously, reflecting the rapidly changing nature of computer architecture.