Optimized interval splitting in a linear scan register allocator

DMCA

Optimized interval splitting in a linear scan register allocator (2005)

Cached

Download Links

Other Repositories/Bibliography

DBLP

by Christian Wimmer , Hanspeter Mössenböck

Venue:	In 1st ACM/USENIX International Conference on Virtual Execution Environments (VEE
Citations:	32 - 13 self

BibTeX

@INPROCEEDINGS{Wimmer05optimizedinterval,
    author = {Christian Wimmer and Hanspeter Mössenböck},
    title = {Optimized interval splitting in a linear scan register allocator},
    booktitle = {In 1st ACM/USENIX International Conference on Virtual Execution Environments (VEE},
    year = {2005},
    pages = {132--141}
}

OpenURL

Abstract

We present an optimized implementation of the linear scan register allocation algorithm for Sun Microsystems ’ Java HotSpot TM client compiler. Linear scan register allocation is especially suitable for just-in-time compilers because it is faster than the common graph-coloring approach and yields results of nearly the same quality. Our allocator improves the basic linear scan algorithm by adding more advanced optimizations: It makes use of lifetime holes, splits intervals if the register pressure is too high, and models register constraints of the target architecture with fixed intervals. Three additional optimizations move split positions out of loops, remove register-to-register moves and eliminate unnecessary spill stores. Interval splitting is based on use positions, which also capture the kind of use and whether an operand is needed in a register or not. This avoids the reservation of a scratch register. Benchmark results prove the efficiency of the linear scan algorithm: While the compilation speed is equal to the old local register allocator that is part of the Sun JDK 5.0, integer benchmarks execute about 15 % faster. Floating-point benchmarks show the high impact of the Intel SSE2 extensions on the speed of numeric Java applications: With the new SSE2 support enabled, SPECjvm98 executes 25 % faster compared with the current Sun JDK 5.0.

Keyphrases

interval splitting linear scan register allocator lifetime hole optimized implementation use position floating-point benchmark target architecture compilation speed sun jdk common graph-coloring approach basic linear scan algorithm linear scan algorithm old local register allocator new sse2 support register pressure benchmark result fixed interval unnecessary spill store register-to-register move split position current sun jdk yield result advanced optimization additional optimization high impact scratch register intel sse2 extension numeric java application linear scan register allocation just-in-time compiler linear scan register allocation algorithm split interval

DMCA