| | |
Summary: Partial Evaluation for
ConstraintBased Program Analyses
Torben Amtoft #
March 1, 2000
Abstract
We report on a case study in the application of partial evaluation, ini
tiated by the desire to speed up a constraintbased algorithm for control
flow analysis. We designed and implemented a dedicated partial evaluator,
able to specialize the analysis wrt. a given constraint graph and thus re
move the interpretive overhead, and measured it with Feeley's Scheme
benchmarks. Even though the gain turned out to be rather limited, our
investigation yielded valuable feed back in that it provided a better under
standing of the analysis, leading us to (re)invent an incremental version.
We believe this phenomenon to be a quite frequent spino# from using par
tial evaluation, since the removal of interpretive overhead makes the flow
of control more explicit and hence pinpoints sources of ine#ciency. Fi
nally, we observed that partial evaluation in our case yields such regular,
lowlevel specialized programs that it begs for runtime code generation.
1 Introduction
O#ine partial evaluation benefits much from static program analysis [5, 9],
|