How Hard is Partitioning for the Sporadic Task Model?

Partitioning n independent sporadic real-time tasks among m identical processors is known to be an NP-hard in the strong sense (by transformation from bin-packing). Therefore, current research on partitioning has focused on developing and analyzing various heuristics and approximation algorithms. However, currently only "loose\´\´ fundamental limits of approximation (trivially based on the known limits for bin packing and partitioning periodic tasks) are known for partitioning sporadic tasks. In this position paper, we briefly summarize known resource-augmentation approximation ratio results for sporadic task systems and argue for further theoretic investigation of approximation schemes and lower bounds for this problem. We believe the results of such an investigation will be invaluable (beyond their theoretic implications) to the real-time multicore system designer; such results will inform the system designer on the relative benefits and disadvantages of various task allocation strategies.