Safe and efficient hybrid memory management for Java

Abstract

Professor Michael Franz, Chair Java uses automatic memory management, usually implemented as a garbage-collected heap.That lifts the burden of manually allocating and deallocating memory, but it can incur significant runtime overhead and increase the memory footprint of applications. We propose a hybrid memory management scheme that utilizes region-based memory management to deallocate objects automatically on region exits. Static program analysis detects allocation sites that are safe for region allocation, i.e., the static analysis proves that the objects allocated at such a site are not reachable after the region exit. A regular garbage-collected heap is used for objects that are not region allocatable.The region allocation exploits the temporal locality of object allocation. Our analysis uses coarse-grain source code annotations to disambiguate objects with non-overlapping lifetimes, and maps them to different memory regions. Region-allocated memory does not require garbage collection as the regions are simply deallocated when they go out of scope. The region allocation technique is backed by a garbage collector that manages memory that is not region allocated.We provide a detailed description of the analysis, provide experimental results showing that as much as 78% of the memory is region allocatable and discuss how our hybrid memory management system can be implemented efficiently with respect to both space and time.xi

Project snowflake: non-blocking safe manual memory management in .NET

“…Scala off-heap [github.com/densh 2017] provides a mechanism to offload all allocations in a given scope onto the unmanaged heap but no full temporal and thread safety. Stancu et al [2015] introduce a static analysis that can infer a hierarchy of regions and annotate allocation sites so that at runtime a stack-like discipline of regions can be enforced. The analysis ensures that pointers only exist from newer regions to older regions or the GC heap but not vice verca.…”

HPTA: High-performance text analytics

“…Alternatively, systems using garbage collection will incur significant garbage collection overheads when all text fragments are separately allocated. The garbage collector must analyze these objects upon each collection pass, adding to the overhead of garbage collection [33]. However, it can be expected that large groups of text fragments have equal life-times in text analytics applications.…”

“…Region-based memory allocators aim to maximize performance by eradicating internal fragmentation and by efficiently de-allocating a large number of items in bulk [6,12,33]. Region-based memory allocation is effective when individually allocated items go out of scope at the same time, implying that their memory can be reclaimed at the same time.…”

Fast and Lean Immutable Multi-Maps on the JVM based on Heterogeneous Hash-Array Mapped Tries

“…Escape analysis enables compilers to improve the run-time performance of programs: it determines whether an object is accessible outside its allocating method or thread. Subsequently this information is used to apply optimizations such as stack allocation (in contrast to heap allocation), scalar replacements, lock elision, or regionbased memory management [16]. Current JVMs use partial escape analysis [15], which is a control-flow sensitive and practical variant tailored toward JIT compilers.…”