We design the next generation of computer systems. Working at the intersection of hardware and software, our research studies how to best implement computation in the physical world. We design processors that are faster, more efficient, easier to program, and secure. Our research covers systems of all scales, from tiny Internet-of-Things devices with ultra-low-power consumption to high-performance servers and datacenters that power planet-scale online services. We design both general-purpose processors and accelerators that are specialized to particular application domains, like machine learning and storage. We also design Electronic Design Automation (EDA) tools to facilitate the development of such systems.

    Advances in computer architecture create quantum leaps in the capabilities of computers, enabling new applications and driving the creation of entirely new classes of computer systems. For example, deep learning, which has transformed many areas of computer science, was made practical by hardware accelerators (initially GPUs and later more specialized designs); and advances in computer performance have also made self-driving cars and autonomous drones possible.

    Computer architecture spans many layers of the hardware and software stack, and as a result we collaborate with researchers in many other areas. For example, several of our current projects focus on the design of domain-specific architectures, and involve researchers in programming languages and compilers to ensure that our systems are broadly useful, as well as domain experts. In addition, the waning of Moore’s Law is making emerging technologies, like CN-FETs, photonics, or resistive memories, an attractive way to implement computation, sparking collaborations with experts in these areas.

    Latest news in computer architecture

    New work on linear-probing hash tables from MIT CSAIL could lead to more efficient data storage and retrieval in computers.

    A long-time department head who established the MEng degree for EECS undergraduates, Penfield developed courses illuminating the equivalence of information and thermodynamic entropy.

    Two research projects on the design of state-of-the-art hardware could one day power next-generation 5G and 6G mobile networks.

    Image: Adam Glanzman MIT has again been named the world’s top university by the QS World University Rankings, which were announced today. This is the 10th year in

    MIT researchers have developed a deep learning neural network to aid the design of soft-bodied robots, such as these iterations of a robotic elephant. Image courtesy of the

    Upcoming events