Banca de QUALIFICAÇÃO: Johannes Warwick da Costa Farias

Uma banca de QUALIFICAÇÃO de DOUTORADO foi cadastrada pelo programa.
STUDENT : Johannes Warwick da Costa Farias
DATE: 31/08/2022
TIME: 13:30
LOCAL: Auditório nPITI
TITLE:

SPECULATIVE GUARDBAND: EXPLOITING CRITICAL-DELAY VARIATIONS ACROSS CACHED INSTRUCTIONS

KEY WORDS:

voltage margins, guardband, energy efficiency, computer architecture, reliability.

PAGES: 25
BIG AREA: Outra
AREA: Microeletrônica
SUMMARY:

Moore’s Law states that the number of transistors in an integrated circuit will double every two years. Both performance and the number of functionalities per circuit have significantly increased as a result of the ongoing development in transistor density. As a result, the evolution of mobile devices like laptops, tablets, and smartphones has been accelerated. However, as technology scales down into the nanoscale range, CMOS (Complementary metal-oxide-semiconductor) devices are experiencing a number of issues including higher leakage currents, wide parameter variability, poor reliability, and low yield. Therefore, Parameter variations have become a challenge for maintaining the reliability of processors. Processor designers traditionally rely on voltage guardband to offset the effect of these parameter variations. The added voltage margin (guardband) must be great enough to ensure the reliable execution of processors even with the worst-case parameter variations. Using a static value of guardband makes the production process faster and less costly but trades off performance and power efficiency for reliability. Therefore, it is interesting to explore this scenario to achieve energy efficiency through active guardband management, where a processor operates at a lower voltage because worst-case conditions rarely occur. This work proposes the development of a machine learning solution capable of reducing average supply voltage levels by capturing the intrinsic critical path variations according to which internal circuits are used by different sets of instructions.

COMMITTEE MEMBERS:
Interno - 1673543 - SAMUEL XAVIER DE SOUZA
Externo ao Programa - 1721835 - KAYO GONCALVES E SILVA - UFRNExterno ao Programa - 3216921 - TIAGO TAVARES LEITE BARROS - UFRN