Understanding soft errors in uncore components

Abstract

The effects of soft errors in processor cores have been widely studied. However, little has been published about soft errors in uncore components, such as memory subsystem and I/O controllers, of a System-on-a-Chip (SoC). In this work, we study how soft errors in uncore components affect system-level behaviors. We have created a new mixed-mode simulation platform that combines simulators at two different levels of abstraction, and achieves 20,000× speedup over RTL-only simulation. Using this platform, we present the first study of the system-level impact of soft errors inside various uncore components of a large-scale, multi-core SoC using the industrial-grade, open-source OpenSPARC T2 SoC design. Our results show that soft errors in uncore components can significantly impact system-level reliability. We also demonstrate that uncore soft errors can create major challenges for traditional system-level checkpoint recovery techniques. To overcome such recovery challenges, we present a new replay recovery technique for uncore components belonging to the memory subsystem. For the L2 cache controller and the DRAM controller components of OpenSPARC T2, our new technique reduces the probability that an application run fails to produce correct results due to soft errors by more than 100× with 3.32% and 6.09% chip-level area and power impact, respectively.

Artificial Neural Networks for Space and Safety-Critical Applications: Reliability Issues and Potential Solutions

“…Recent works combined an extremely detailed gate level fault injection in tandem with a faster (but still impracticable for complex devices) RTL evaluation [79], [103]. Cho et al used high level simulation (not using real hardware) triggering a RTL model when the fault needs to be injected [104]. Subasi et.…”

On-Chip Bus Protection against Soft Errors

“…It was examined that transient faults on the bus matrix could have severe consequences, possibly leading to deadlock, memory corruption, or undefined behavior [40,41]. A simulation study [42] observed the highest rate of erroneous application outcomes due to soft errors in the crossbar interconnect (only components external to the core underwent fault injection). This section provides a review of proposed and currently used bus protection approaches.…”

Design for manufacturability and reliability in extreme-scaling VLSI

“…Multi-objective genetic algorithm, based on gate sizing, is developed in [162] to optimize SER of standard cell circuits with marginal delay overhead. Recently in [163], soft error in non-core components such as memory and I/O subsystems is studied. A mixed mode simulation platform has been proposed to estimate how soft error in non-core components can affect the system level reliability.…”