Title
A Reconfigurable Memory Management Core For Java Applications
Abstract
Shortly after the introduction of Java technology, numerous research efforts went into overcoming the performance shortcomings of the Java Virtual Machine (JVM) by mapping it onto hardware. It seems that most works addressing the implementation of a JVM in hardware often excludes the mapping of compute-intensive tasks such as memory allocation, garbage collection, and exception handling. In this paper, we propose a hardware architecture to support dynamic memory allocation and deallocation operations in a JVM. Such hardware architectures are becoming attractive alternatives when compared to their software implementation counterparts. This architecture, called Memory Management Core (MMC), (i) supports a memory map that is independent of the physical memory size, (ii) allocates blocks with exact size to minimize fragmentation, (iii) performs in-place coalescing and splitting of the memory blocks, (iv) services concurrent multiple allocation and de-allocation requests to achieve maximum performance, and (iv) attempts to achieve a high utilization of all its hardware components. A prototype of this architecture is implemented and mapped on a suitable FPGA device to analyze its performance. The obtained results show that the FPGA implementation of this architecture can service memory allocation and deallocation many order of magnitude faster than software implementations for memory-intensive applications.
Publication Date
9-24-2004
Publication Title
Proceedings - IEEE Computer Society Annual Symposium on VLSI: Emerging Trends in VLSI Systems Design
Number of Pages
309-312
Document Type
Article; Proceedings Paper
Personal Identifier
scopus
DOI Link
https://doi.org/10.1109/ISVLSI.2004.1339568
Copyright Status
Unknown
Socpus ID
4544383753 (Scopus)
Source API URL
https://api.elsevier.com/content/abstract/scopus_id/4544383753
STARS Citation
Ejnioui, Abdel and Rhiati, Abdelkader, "A Reconfigurable Memory Management Core For Java Applications" (2004). Scopus Export 2000s. 5008.
https://stars.library.ucf.edu/scopus2000/5008