Integrated Hardware Garbage Collection

Garbage collected programming languages, such as Python and C#, have accelerated software development. These modern languages increase productivity and software reliability as they provide high-level data representation and control structures. Modern languages are widely used in software development for mobile, desktop, and server devices, but their adoption is limited in real-time embedded systems. There is clear interest in supporting modern languages in embedded devices as emerging markets, like the Internet of Things, demand ever smarter and more reliable products. Multiple commercial and open-source projects, such as Zerynth and MicroPython, are attempting to provide support. But these projects rely on software garbage collectors that impose high overheads and introduce unpredictable pauses, preventing their use in many embedded applications. These limitations arise from the unsuitability of conventional processors for performing efficient, predictable garbage collection. We propose the Integrated Hardware Garbage Collector (IHGC); a garbage collector tightly coupled with the processor that runs continuously in the background. Further, we introduce a static analysis technique to guarantee that real-time programs are never paused by the collector. Our design allocates a memory cycle to the collector when the processor is not using the memory. The IHGC achieves this by careful division of collection work into single-memory-access steps that are interleaved with the processor's memory accesses. As a result, our collector eliminates run-time overheads and enables real-time program analysis. The principles behind the IHGC can be used in conjunction with existing architectures. For example, we simulated the IHGC alongside the ARMv6-M architecture. Compared to a conventional processor, our experiments indicate that the IHGC offers 1.5-7 times better performance for programs that rely on garbage collection. The IHGC delivers the benefits of garbage-collected languages with real-time performance but without the complexity and overheads inherent in software collectors.