Título: Multicore Fault Tolerant Systems: From Principles to Challenges
Recent integrated circuit (IC) technologies enable the integration of billions of transistors into a single chip, which allows the implementation of complex parallel systems. These systems are very close to physical limitations increasing the likelihood of manufacturing failures and failures during system operation. In this regards, it is essential to have a multicore fault tolerant system. This talk addresses the beginning of multicore systems, as well as discusses the limitations and challenges that must be overcome to design reliable multicore systems. In addition, this talk also details some works that aim to produce more resilient architectures.
Rafael Garibotti received the Ph.D. and MSc. Degree in Microelectronics, respectively from University of Montpellier and EMSE, France and the BSc.
Degree in Computer Engineering from PUCRS, Brazil. Former Postdoctoral Fellow at Harvard University. Now he has a Postdoctoral Fellow at UFRGS. His research activity focuses on programmability of multi/many-core systems and design of parallel architectures, such as accelerators.