Mahlke’s research focuses on novel methods for designing energy-efficient high performance computer systems. Hardware-software codesign is used to create specialized computer systems that are customized to application domains or even individual applications. Efficiency gains are realized through specialization and optimization of computation, memory systems, connectivity, etc. Application domains of interest include machine learning, augmented/virtual reality, and scientific simulation. Tools employed include compilers such as LLVM, program instrumentation/analysis tools such as Pin, and architectural simulators such as Gem5.
Faculty
Scott Mahlke
Professor, Electrical Engineering and Computer Science
Contact
734-936-1602
[email protected]
Website
Research
Research Areas
AI; ML and Statistical InferenceComputational and Informational Processing
Computer Architecture; Optimization; Control and HPC