A Survey of Adaptive Optimization in Virtual Machines

Arnold, Matthew; Fink, Stephen J.; Grove, David; Hind, Michael; Sweeney, Peter F.

doi:10.1109/jproc.2004.840305

Cited by 141 publications

(83 citation statements)

References 114 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In the context of Virtual Machines used heavily today, this has taken hold [3]. Here, the question is again how to make intelligent use of resources.…”

Section: Review Of Self-modificationmentioning

confidence: 99%

Developments in Cartesian Genetic Programming: self-modifying CGP

Harding

Miller

Banzhaf

2010

Genet Program Evolvable Mach

View full text Add to dashboard Cite

Self-modifying Cartesian Genetic Programming (SMCGP) is a general purpose, graph-based, developmental form of Genetic Programming founded on Cartesian Genetic Programming. In addition to the usual computational functions, it includes functions that can modify the program encoded in the genotype. This means that programs can be iterated to produce an infinite sequence of programs (phenotypes) from a single evolved genotype. It also allows programs to acquire more inputs and produce more outputs during this iteration. We discuss how SMCGP can be used and the results obtained in several different problem domains, including digital circuits, generation of patterns and sequences, and mathematical problems. We find that SMCGP can efficiently solve all the problems studied. In addition, we prove mathematically that evolved programs can provide general solutions to a number of problems: n-input even-parity, n-input adder, and sequence approximation to p.

show abstract

“…In the context of Virtual Machines used heavily today, this has taken hold [3]. Here, the question is again how to make intelligent use of resources.…”

Section: Review Of Self-modificationmentioning

confidence: 99%

Developments in Cartesian Genetic Programming: self-modifying CGP

Harding

Miller

Banzhaf

2010

Genet Program Evolvable Mach

View full text Add to dashboard Cite

show abstract

“…A complete survey can be found in [2], while further considerations are presented by Kulkarny et al [15]. This approach exploits a dedicated thread to detect hot spots and optimize them.…”

Section: Related Workmentioning

confidence: 99%

Dynamic Look Ahead Compilation: A Technique to Hide JIT Compilation Latencies in Multicore Environment

Campanoni

Sykora

Agosta

et al. 2009

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Object-code virtualization, commonly used to achieve software portability, relies on a virtual execution environment, typically comprising an interpreter used for initial execution of methods, and a JIT for native code generation. The availability of multiple processors on current architectures makes it attractive to perform dynamic compilation in parallel with application execution. The major issue is to decide at runtime which methods to compile ahead of execution, and how much time to invest in their optimization. This research introduces an abstract model, termed Dynamic Look Ahead (DLA) compilation, which represents the available information on method calls and computational weight as a weighted graph. The graph dynamically evolves as computation proceeds. The model is then instantiated by specifying criteria for adaptively choosing the method compilation order. The DLA approach has been applied within our dynamic compiler for .NET. Experimental results are reported and analyzed, for both synthetic programs and benchmarks. The main finding is that a careful choice of method-selection criteria, based on light-weight program analysis and execution tracing, is essential to mask compilation times and to achieve higher overall performances. On multi-processors, the DLA approach is expected to challenge the traditional virtualization environments based on bytecode interpretation and JITing, thus bridging the gap between ahead-of-time and just-in-time translation.

show abstract

“…Smith defines feedback-directed optimizations (FDO) [24][25][26][27]13] as a family of techniques that alter a program's execution based on tendencies observed in its present or past runs. FDO is orthogonal to convergent compilation since FDO typically uses dynamic edge, path, or method profiling to select which code to optimize, and to guide heuristics.…”

Section: Feedback-directed Optimizationmentioning

confidence: 99%

Convergent Compilation Applied to Loop Unrolling

Nethercote

Burger

McKinley

2007

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Well-engineered compilers use a carefully selected set of optimizations, heuristic optimization policies, and a phase ordering. Designing a single optimization heuristic that works well with other optimization phases is a challenging task. Although compiler designers evaluate heuristics and phase orderings before deployment, compilers typically do not statically evaluate nor refine the quality of their optimization decisions during a specific compilation. This paper identifies a class of optimizations for which the compiler can statically evaluate the effectiveness of its heuristics and phase interactions. When necessary, it then modifies and reapplies its optimization policies. We call this approach convergent compilation, since it iterates to converge on high quality code. This model incurs additional compilation time to avoid some of the difficulties of predicting phase interactions and perfecting heuristics This work was motivated by the TRIPS architecture which has resource constraints that have conflicting phase order requirements. For example, each atomic execution unit (a TRIPS block) has a maximum number of instructions (128) and a fixed minimum execution time cost. Loop unrolling and other optimizations thus seek to maximize the number of mostly full blocks. Because unrolling enables many downstream optimizations, it needs to occur well before code generation, but this position makes it impossible to accurately predict the final number of instructions. After the compiler generates code, it knows the exact instruction count and consequently if it unrolled too much or too little or just right. If necessary, convergent unrolling then goes back and adjusts the unroll amount accordingly and reapplies subsequent optimization phases. We implement convergent unrolling which automatically matches the best hand unrolled version for a set of microbenchmarks on the TRIPS architectural simulator. Convergent compilation can help solve other phase ordering and heuristic tuning compilation challenges. It is particularly well suited for resource constraints that the compiler can statically evaluate such as register usage, instruction level parallelism, and code size. More importantly, these resource constraints are key performance indicators in embedded, VLIW, and partitioned hardware and indicate that convergent compilation should be broadly applicable.

show abstract

A Survey of Adaptive Optimization in Virtual Machines

Cited by 141 publications

References 114 publications

Developments in Cartesian Genetic Programming: self-modifying CGP

Developments in Cartesian Genetic Programming: self-modifying CGP

Dynamic Look Ahead Compilation: A Technique to Hide JIT Compilation Latencies in Multicore Environment

Convergent Compilation Applied to Loop Unrolling

Contact Info

Product

Resources

About