PROJECT TITLE :

RISO: Enforce Noninterfered Performance With Relaxed Network-on-Chip Isolation in Many-Core Cloud Processors

ABSTRACT:

Workload consolidation is widely utilized in modern cloud processors to cut back total cost of ownership. Performance isolation needs to be enforced between consolidated workloads to attain controllable quality of service. Networks-on-chip (NoCs), as a serious shared resource, often incur traffic interference and violate performance isolation criteria. Previous work resorts to strict isolation strategy that partitions NoC into independent regions to isolate core-to-core communication traffic. But, strict isolation either ends up in low consolidation density or degrades network performance, and more importantly, cannot be applied to memory access traffic. To address these weaknesses, we tend to propose a unique performance isolation strategy in NoC, known as relaxed isolation (RISO). It permits underutilized routers and links to be shared by multiple applications, and, at the same time, it keeps the aggregated traffic in check to enforce performance isolation. Experimental results show that RISO could effectively improve consolidation density and network performance in synergy.