Scratchpad allocation for data aggregates in superperfect graphs

Existing methods place data or code in scratchpad memory, i.e., SPM by either relying on heuristics or resorting to integer programming or mapping it to a graph coloring problem. In this work, the SPM allocation problem is formulated as an interval coloring problem. The key observation is that in many embedded applications, arrays (including structs as a special case) are often related in the following way: For any two arrays, their live ranges are often such that one is either disjoint from or contains the other. As a result, array interference graphs are often superperfect graphs and optimal interval colorings for such array interference graphs are possible. This has led to the development of two new SPM allocation algorithms. While differing in whether live range splits and spills are done sequentially or together, both algorithms place arrays in SPM based on examining the cliques in an interference graph. In both cases, we guarantee optimally that all arrays in an interference gra...

• Type: Conference
• Year: 2007
• Where: LCTRTS
• Authors: Lian Li 0002, Quan Hoang Nguyen, Jingling Xue
• Keywords: Graph Coloring, Interference Graphs, LCTRTS 2007, Software Engineering, SPM Allocation