Jung Ho Ahn

Jung Ho has published over 60 peer-reviewed papers at prestigious venues, such as ACM/IEEE International Symposium on Computer Architecture (ISCA), IEEE/ACM International Symposium on Microarchitecture (MICRO), and IEEE International Symposium on High Performance Computer Architecture (HPCA).

Recently, he has focused on studies improving the performance, energy efficiency, and reliability of memory subsystems, exemplified by ‘Row-Buffer Decoupling: A Case for Low-latency DRAM Microarchitecture’ (ISCA 2014), ‘Microbank: Architecting Through-Silicon Interposer-Based Main Memory Systems’ (SC 2014), and ‘CiDRA: A Cache-inspired DRAM Resilience Architecture’ (HPCA 2015).

He is a member of HPCA hall of fame, and received multiple best paper awards from International Conference for High Performance Computing, Networking, Storage and Analysis (SC) and International Conference on Parallel Architectures and Compilation Techniques (PACT).

Dr. Ahn’s research goals are to improve the performance, energy efficiency, and reliability of computer systems through proposing hardware and software changes derived from an in-depth analysis of the characteristics of both emerging real workloads and the strength/weakness of modern technology and their alternatives. Dr. Ahn has been working primarily with the platforms team at Google but also collaborating with other groups to understand and improve server designs. A key initiative has been to understand how large online data-intensive workloads like Google Search use the memory subsystems in existing processors. This work has identified several interesting new insights about how current processor offerings in the industry, while well-optimized for previous-generation scientific and engineering workloads, are often mismatched with large web workloads. Using these insights, the team is researching several new optimizations to the system architecture to improve performance.

By working alongside our engineers on an efficient warehouse-scale computing infrastructure, Dr. Ahn hopes to leverage what he has learned about Google infrastructure and open source tools (such as TensorFlow, gRPC, and Protobuf) to his classes and students, in order to help them prepare and take advantage of emerging research trends.