Download now Free registration required
Heterogeneous multi-core architectures promise greater energy/area efficiency than their homogeneous counterparts. This efficiency can only be realized, however, if the operating system assigns applications to appropriate cores based on their architectural properties. While several such heterogeneity-aware algorithms were proposed in the past, they were not meant to scale to a large number of cores and assumed long-lived threads due to reliance on continuous performance monitoring of threads for core assignment purposes. The authors propose a scheme that does not rely on dynamic performance monitoring.
- Format: PDF
- Size: 183 KB