Fault-aware Communication Mapping for NoCs with Guaranteed Latency

Abstract

As feature sizes shrink, transient failures of on-chip network links become a critical problem. At the same time, many applications require guarantees on both message arrival probability and response time. We address the problem of transient link failures by means of temporally and spatially redundant transmission of messages, such that designer-imposed message arrival probabilities are guaranteed. Response time minimisation is achieved by a heuristic that statically assigns multiple copies of each message to network links, intelligently combining temporal and spatial redundancy. Concerns regarding energy consumption are addressed in two ways. First, we reduce the total amount of transmitted messages, and, second, we minimise the application response time such that the resulted time slack can be exploited for energy savings through voltage reduction. The advantages of the proposed approach are guaranteed message arrival probability and guaranteed worst case application response time.

σnLBDR: generic congestion handling routing implementation for two‐dimensional mesh network‐on‐chip

“…It is proposed only for deterministic routing, and the delivery of packets is not guaranteed. Reliability issues have been addressed at bus level [18] as well as application level [19].…”

System Reliability-Aware Energy Optimization for Network-on-Chip with Voltage-Frequency Islands

“…At the high-integration level, the transient failures rates of devices are affected by such as power supply noise and capacitive cross-talk [13]. Reliability becomes a critical requirement for most systems.…”

On bottleneck analysis in stochastic stream processing

“…For example, audio frames with a busy signal can be decoded faster than normal voice frames; some video frames exhibit less motion than others. A real-life audio-video decoder with stochastic stage-processing times is studied in Manolache et al [2007a]. In a different domain, the performance of GPGPU programs processing irregular data is hard to predict accurately due to (i) long graphics pipelines and (ii) increasing user-hardware separation encouraged by CUDA programming.…”