1. XwattPilot: A Full-stack Cloud System Enabling Agile Development of Transprecision Software for Low-power SoCs
- Author
-
Fabian Schuiki, A. Cristiano I. Malossi, Germain Haugou, Frank K. Gurkaynak, Michael Schaffner, Florian Scheidegger, Luca Benini, Christoph Hagleitner, Stefan Mach, Dionysios Diamantopoulos, Diamantopoulos, Dionysio, Scheidegger, Florian, Mach, Stefan, Schuiki, Fabian, Haugou, Germain, Schaffner, Michael, Gurkaynak, Frank K., Hagleitner, Christoph, Malossi, A. Cristiano I., and Benini, Luca
- Subjects
010302 applied physics ,Transprecision ,RISC-V ,FPGA ,Cache-Coheren ,CAPI ,Virtual prototyping ,Claud ,Energy efficiency ,business.industry ,Computer science ,Transprecision, RISC-V, FPGA, Cache-Coherent, CAPI, Virtual Prototyping, Cloud, Energy Efficiency ,Software development ,Cloud computing ,02 engineering and technology ,01 natural sciences ,020202 computer hardware & architecture ,symbols.namesake ,Software ,Embedded system ,0103 physical sciences ,0202 electrical engineering, electronic engineering, information engineering ,symbols ,Performance improvement ,business ,Von Neumann architecture ,Agile software development - Abstract
The performance improvement rate of conventional von Neumann processors has slowed as Moore's Law grinds to an economic halt, giving rise to a new age of heterogeneity for energy-efficient computing. Extending processors with finely tunable precision instructions have emerged as a form of heterogeneity that tradeoffs computation precision with power consumption. However, the prolonged design time due to customization of the supported framework for a system-on-a-chip may counteract the advantages of transprecision computing. We propose XwattPilot, a system aiming at accelerating the transprecision software development of low-power processors using cloud technology. We show that the total energy-to-solution can be significantly decreased by using transprecision computations, whereas the proposed system can accelerate tie energy-efficiency evaluation runtime by 10.3x.
- Published
- 2020