JITProf: pinpointing JIT-unfriendly JavaScript code

Abstract

Most modern JavaScript engines use just-in-time (JIT) compilation to translate parts of JavaScript code into efficient machine code at runtime. Despite the overall success of JIT compilers, programmers may still write code that uses the dynamic features of JavaScript in a way that prohibits profitable optimizations. Unfortunately, there currently is no technique that helps developers to identify such JIT-unfriendly code. This paper presents JITPROF, a profiling framework to dynamically identify code locations that prohibit profitable JIT optimizations. The basic idea is to associate execution counters with potentially JIT-unfriendly code locations and to use these counters to report code locations that match code patterns known to prohibit optimizations. We instantiate the idea for six JIT-unfriendly code patterns that cause performance problems in the Firefox and Chrome browsers, and we apply the approach to popular benchmark programs. Our results show that refactoring these programs to avoid performance problems identified by JITPROF leads to performance improvements of up to 26.3% in 12 benchmarks.

TypeDevil: Dynamic Type Inconsistency Analysis for JavaScript

“…First, the implementation deviates from the standard. Second, the implementation results in arrays that contain undefined values ("arrays with holes"), which is known to cause suboptimal performance because the JIT compiler cannot apply particular optimizations [17], [20]. TypeDevil detects these problems because functions that implement parts of the cryptographic algorithms have arguments that contain both number and undefined values.…”

“…3d -cube 289 104 43 1 1 3d-morph 24 3 3 0 0 3d-raytrace 346 103 72 3 0 access-binary-trees 40 9 9 0 0 access-fannkuch 52 3 3 0 0 access-nbody 143 29 25 0 0 access-nsieve 29 6 6 0 0 bitops-3bit-bits-in-byte 17 …”

DLint: dynamically checking bad coding practices in JavaScript

“…1 Note that it is legitimate for performance reasons to call these constructors with arguments [24]. Table 3: Code quality rules and runtime patterns related to language misuse (all are single-event patterns).…”

“…JITProf is a profiler that identifies performance bottlenecks caused by code that JIT engines cannot effectively optimize [24]. Each of these approaches addresses a particular kind of programming error, whereas DLint is a generic framework for checking code quality rules.…”

SJS: A Type System for JavaScript with Fixed Object Layout

“…Modern JavaScript engines implement just-in-time (JIT) compilation techniques combined with inline caching, which rely, among other things, on the fact that the layouts of most JavaScript objects do not change often. These optimization heuristics are ineffective when new fields and method are added to an object [16].…”