EPiQC: https://www.epiqc.cs.uchicago.edu/

Dissertation:The clock is fundamental to synchronous computing because it influences the entire system stack - from circuits and micro-architecture to the OS and applications. However, the clocking system is still somewhat of a black box across various layers of computing and this abstraction limits the capability of synchronous systems. My dissertation shows that breaking the clock abstraction and analyzing its implications can have significant benefits across the computing stack. Breaking the clock abstraction has implications across the following categories: 1. Functional: The relationship between clock cycle utilization and any functional unit, its computations and the operands involved.2. Structural: The association between hardware structure and the clock. This can be further sub-divided as: a) the relationship between clock cycle utilization, the topology of the system, and the distance between system nodes and b) the power consumed by the clocking system and its unequal distribution across the clock hierarchy.3. Differential: The effect of variation on the clock period. My dissertation exploits these implications to optimize diverse parts of the computer system stack and benefits multiple domains of applications.

Heterogeneous architectures and their technology optimization under Prof. Nam Sung Kim.

Task scheduling, synchronization, sharing - GPUs, coherence, consistency etc.

GPU Design team. 1. Debug bus, Perf Counters, , Scan chains for the Adreno GPU2. Socket communication between CPU/GPU

Full chip performance verification for Tegra SOCs. Worked on SD/MMC, Security, Video Decode engine, GPU modules. Also developed performance infrastructure for the entire SOC.