Browsing by Author "Graham, Aaron G."

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Just-In-Time compiler performance evaluation on the AArch64 platform
    (University of New Brunswick, 2021) Graham, Aaron G.; Kent, Kenneth
    The embedded computing market, which includes Internet-of-Things (IoT) and mobile computing devices, is a non-traditional computing market where computation resources—including CPU, memory, power—are more limited. Due to these limitations, software is required to be more compact and efficient. Providing a managed runtime, such as Eclipse OpenJ9 built on top of Eclipse OMR, in this climate differs from a cloud/desktop-based environment. This thesis focuses on porting the OpenJ9+OMR technology, which has a heritage of running in resource constrained systems, to a new environment while continuing to provide a generic run-time environment. The low-power AArch64 (ARMv8-A) platform—compatible with commonly used electronic devices—is becoming the answer for resource constrained environments of embedded systems. This thesis explores the performance of the Just-In-Time (JIT) compiler in OpenJ9, a Java® Virtual Machine (JVM), on an Instruction Set Architecture (ISA) appropriate for IoT and embedded devices. More specifically, we evaluate and validate the AArch64 implementation of OpenJ9’s JIT against more mature architectures currently available. The evaluation reveals performance discrepancies and necessary improvements, beyond those that are already known, by comparing the AArch64 implementation to another ISA, x86-64. Our work is an effort to template new architectural support and allow others to follow our model. We provide a baseline for future research on OpenJ9, OMR and the JIT on the AArch64 platform and outline some improvements as future work.