BASS Research Group

Our lab targets high-performance computer systems, at the intersection of Computer Architecture and Data Systems. In particular, we design fast algorithms and build efficient software/hardware systems, for AI, big data, S&P, etc. We are looking for grad and undergrad students to work on our research projects. If you are interested in joining our lab, please fill in the recruiting form.

Our recent focus is scalable computing. The rapid escalating complexity of emerging applications and the lagging speed of computer systems creates a huge gap. This gap largely limits the problem scale that emerging technology, e.g. AI, can be applied to, known as the scalability wall.

We use performance engineering to provide scalable computing power in the post-Moore’s Law era. It covers three aspects: new algorithms, software performance engineering and novel hardware architectures. More specifically, application domains like ML and databases continually create a need for new algorithms. We explore algorithm innovations in new problem domains, and under new machine models, e.g. accelerator offloading. Beyond algorithms, software performance engineering removes software bloat and tailors software to hardware. For instance, data tiling improves cache/memory efficiency. In addition to software, hardware architecture redesign, like processor simplification and domain specialization, is inevitable. One example is to build dedicated AI/ML and big-data accelerators.