Latency Reduction in Software–dsms by Means of Dynamic Function Splicing

We present a compiler–based technique to automatically identify and extract Remote Procedure Calls, so–called Function Splices, out of potentially arbitrary sequences of Java code compiled for a software DSM. The goal is to lower communication latencies and message traffic by replacing data shipping by function shipping. Dynamic Function Splicing dynamically decides at runtime whether to invoke a function splice on the local machine or to execute it remotely on the home node of the requested data. On proof–of–concept micro–benchmarks Dynamic Function Splicing reduces the execution wall time by approximately 29 %; about 25 % of the messages can be saved.