Adaptive multi-level compilation in a trace-based Java JIT compiler

Inoue, Hiroshi; Hayashizaki, Hiroshige; Wu, Peng; Nakatani, Toshio

doi:10.1145/2384616.2384630

Cited by 10 publications

(5 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recently, trace-based compilation has gained popularity in dynamic scripting languages [5,10] and high level language virtual machines [11,16,17,26]. Wu et al [26] and Inoue et al [16,17] investigate the performance of several variations of NET on tracebased Java virtual machines.…”

Section: Trace-based Language Virtual Machinesmentioning

confidence: 99%

See 1 more Smart Citation

Improving dynamic binary optimization through early-exit guided code region formation

et al. 2013

View full text Add to dashboard Cite

Most dynamic binary translators (DBT) and optimizers (DBO) target binary traces, i.e. frequently executed paths, as code regions to be translated and optimized. Code region formation is the most important first step in all DBTs and DBOs. The quality of the dynamically formed code regions determines the extent and the types of optimization opportunities that can be exposed to DBTs and DBOs, and thus, determines the ultimate quality of the final optimized code. The Next-Executing-Tail (NET) trace formation method used in HP Dynamo is an early example of such techniques. Many existing trace formation schemes are variants of NET. They work very well for most binary traces, but they also suffer a major problem: the formed traces may contain a large number of early exits that could be branched out during the execution. If this happens frequently, the program execution will spend more time in the slow binary interpreter or in the unoptimized code regions than in the optimized traces in code cache. The benefit of the trace optimization is thus lost. Traces/regions with frequently taken early-exits are called delinquent traces/regions. Our empirical study shows that at least 8 of the 12 SPEC CPU2006 integer benchmarks have delinquent traces.In this paper, we propose a light-weight region formation technique called Early-Exit Guided Region Formation (EEG) to improve the quality of the formed traces/regions. It iteratively identifies and merges delinquent regions into larger code regions. We have implemented our EEG algorithm in two LLVM-based multithreaded DBTs targeting ARM and IA32 instruction set architecture (ISA), respectively. Using SPEC CPU2006 benchmark suite with reference inputs, our results show that compared to an NETvariant currently used in QEMU, a state-of-the-art retargetable DBT, EEG can achieve a significant performance improvement of Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. . . . $15.00 up to 72% (27% on average), and to 49% (23% on average) for IA32 and ARM, respectively.

show abstract

Section: Trace-based Language Virtual Machinesmentioning

confidence: 99%

“…Wu et al [26] and Inoue et al [16,17] investigate the performance of several variations of NET on tracebased Java virtual machines.…”

Section: Trace-based Language Virtual Machinesmentioning

confidence: 99%

Improving dynamic binary optimization through early-exit guided code region formation

et al. 2013

View full text Add to dashboard Cite

show abstract

“…Traditional program representations, such as control flow graphs, cannot handle the effects of runtime changes to the code, which require accommodating the possibility of some memory locations having different instructions at different times during execution. JIT compilers [16,22] and dynamic binary translators [31] maintain representations of the code being dynamically modified, but not together with that of the code that performs code modification. Whole-system analyses [11,14,21,45,46] perform dynamic taint propagation, taking into account explicit information flows via data dependencies but not implicit flows via control dependencies.…”

Section: Introductionmentioning

confidence: 99%

Representing and reasoning about dynamic code

Bartels

Stephens²,

Debray

2020

Proceedings of the 35th IEEE/ACM International Conference on Automated Software Engineering

View full text Add to dashboard Cite

Dynamic code, i.e., code that is created or modified at runtime, is ubiquitous in today's world. The behavior of dynamic code can depend on the logic of the dynamic code generator in subtle and non-obvious ways, with significant security implications, e.g., JIT compiler bugs can lead to exploitable vulnerabilities in the resulting JIT-compiled code. Existing approaches to program analysis do not provide adequate support for reasoning about such behavioral relationships. This paper takes a first step in addressing this problem by describing a program representation and a new notion of dependency that allows us to reason about dependency and information flow relationships between the dynamic code generator and the generated dynamic code. Experimental results show that analyses based on these concepts are able to capture properties of dynamic code that cannot be identified using traditional program analyses.

show abstract

“…This paper adopts hot-trace compilation technology to compile and optimize it. The so-called hot trace compilation [5] translates frequently executed program fragments (instruction sequences) into target code (Java bytecode) to compile and execute instead of interpretive execution in order to improve the execution efficiency of frequently executed program fragments and avoid the overhead of compiling without frequent program execution.…”

Section: Introductionmentioning

confidence: 99%

An Optimization Method for XML Twig Query

He¹

2018

IJPE

View full text Add to dashboard Cite

XML tree pattern query, also known as Twig query, is the core operation in XML query processing. In the research of the Twig query algorithm, TreeMatch is considered to be one of the best algorithms because it reduces the generation of intermediate results. However, in the core operation getNext of the TreeMatch algorithm, there are many calculations that depend only on Twig mode. This redundant duplicate calculation affects the performance of the TreeMatch algorithm when there are many getNext calls. In order to further improve the algorithm, this paper proposes a Twig query optimization method based on partial evaluation and hot-trace compilation. This method takes Twig mode as an invariant to perform partial evaluation and translates query requests into a Twig query machine instruction sequence. The duplication calculation of the Twig pattern during the query process is avoided, and the process of interpretation of the instruction sequence of the query machine is optimized by using the hot trace compilation technique. The comparison experiment shows that the optimization method based on partial evaluation and hot-trace compilation can increase the efficiency of twig query by 20% to 60%.

show abstract

Adaptive multi-level compilation in a trace-based Java JIT compiler

Cited by 10 publications

References 17 publications

Improving dynamic binary optimization through early-exit guided code region formation

Improving dynamic binary optimization through early-exit guided code region formation

Representing and reasoning about dynamic code

An Optimization Method for XML Twig Query

Contact Info

Product

Resources

About