On-Demand Redundancy Grouping: Selectable Soft-Error Tolerance for a Multicore Cluster
TimeWednesday, July 13th6pm - 7pm PDT
LocationLevel 2 Lobby
DescriptionWith the shrinking of technology nodes and the use of parallel processors in hostile environments, runtime faults caused by radiation are a serious cross-cutting concern. This work introduces an architectural approach to run-time configurable soft-error tolerance, augmenting a six-core RISC-V cluster with a novel On-Demand Redundancy Grouping (ODRG) scheme. ODRG allows the cluster to operate either as two parallel fault tolerant cores, or six parallel cores for high-performance, with limited overhead to switch between these modes. The ODRG unit adds around 11% of a core’s area for a 3-core group, shows negligible timing increase, and has no impact on performance in a non-redundancy mode.