Loading...
Please wait, while we are loading the content...
Similar Documents
Static task partitioning for locked caches in multi-core real- time systems (2011)
| Content Provider | CiteSeerX |
|---|---|
| Author | Sarkar, Abhik Mueller, Frank Ramaprasad, Harini |
| Abstract | Massive multi-core architectures with tens of cores are becoming more prevalent in embedded systems. However, their acceptance in the realtime systems domain is rather low due to challenges in system analysis and predictability. Recent real-time systems research has focused on shared cache architectures. In such systems, tasks across all cores share the same cache (typically at level 2). In contrast, tile-based architectures with massive numbers of cores have private caches, only. At larger numbers of cores, it becomes important to utilize all resources efficiently and share cores among hard, soft and non real-time tasks. In such a scenario, it becomes difficult to analyze cache behavior statically. Consequently, hard real-time systems become subject to conservative analysis including timing bounds based on the assumption that data references default to lower levels of caches/memory. Locking cache lines in hard real-time systems is a common means to ensure timing predictability of data references and to lower bounds on worstcase execution time, especially in a multi-tasking environment. However, cache locking poses a challenge to hard real-time systems on multi-core architectures. Static scheduling on hard real-time tasks does not consider conflicts among locked regions within caches. This work proposes three static scheduling schemes as a remedy: (1) Greedy First Fit Decreasing (GFFD) and (2) Colored First Fit Decreasing (CoFFD). This work also adapts these algorithms for conflict resolution partially locked regions. Experiments indicate that CoFFD consistently outperforms GFFD for lower number of cores and lower system utilization. With partial locking, the number of cores in some cases is reduced by almost 50 % with an increase in system utilization of 10%. Overall, this work is unique in considering the challenges of future multi-core architectures for real-time systems and provides key insights into task partitioning with locked caches for architectures with private caches. 1. |
| File Format | |
| Language | English |
| Publisher Date | 2011-01-01 |
| Access Restriction | Open |
| Subject Keyword | Locked Cache Static Task Multi-core Real Time System Hard Real-time System Real-time System Private Cache System Utilization Multi-tasking Environment Cache Memory Static Scheduling Partial Locking Recent Real-time System Research Data Reference First Fit Decreasing Cache Architecture Worstcase Execution Time Massive Number Data Reference Default Cache Behavior Tile-based Architecture Realtime System Cache Line System Analysis Hard Real-time Task Share Core Core Share Future Multi-core Architecture Key Insight Conflict Resolution Non Real-time Task Multi-core Architecture Conservative Analysis Cache Locking Locked Region Embedded System Massive Multi-core Architecture Greedy First Fit Decreasing |
| Content Type | Text |
| Resource Type | Technical Report |
