| | GSRC Student Profile:
Research Overview: MPSoc Emulation
As we begin to cross the threshold into the billion transistor era Multiprocessor Systems-on-Chips (MPSoC) are becoming ever-present, driving advancement in multiple application domains ranging from entertainment to medical devices. Consequently, MPSoC designers must be able to optimize a multi-constrained set comprised of performance, power consumption, reliability and cost. At the same time, as feature sizes shrink into the sub-micron regime, physical limitations and manufacturing variability are resulting in unreliable, power-hungry chips. To combat this, system simulations which take into consideration the interplay of performance, thermal characteristics, variability, and reliability effects from real-life workloads are becoming valuable when performing design space explorations of these next-generation architectures. However, the increase in complexity stemming from the need to profile complex modern processors, over extended periods of time, is placing a significant burden on the speed of traditional software simulators. My work has been motivated by these challenges and has led me to begin investigating a Virtual Platform (ViP) framework on FPGAs which can significantly speed up system profiling through system emulation. In addition, 3-D technology has emerged as a viable solution for increasing chip densities and improving performance. Currently, my work involves emulating 3-D microarchitectures on these FPGA platforms to investigate design alternatives. My research interest also include:
(*) FPGA accelerators for Scientific Computing Applications
(*) FPGA System-on-Chip (SoC) solutions
(*) ASIC design for low power, low cost, and high performance
(*) Embedded Systems
(*) Statistical Learning theory and Machine Learning
(*) Computer Architecture
| |
|
