Question

I’m reading about program specialization – specifically java and I don’t think I quite understand it to be honest. So far what I understand is that it is a method for optimizing efficiency of programs by constraining parameters or inputs? How is that actually done? Can someone maybe explain to me how it helps, and maybe an example of what it actually does and how its done?

Thanks

I have been reading:

Program Specialization – java

Answer 1

Program specialization is the process of specializing a program when you know in advance what are the arguments you’re going to have.

One example is if you have a test and you know that with your arguments, you’re never going to enter the block, you can eliminate the test.

You create a specialized version of the program for a certain kind of input.

Basically, it helps to get rid off useless with your input. However, with the modern architectures and compilers (at least in C), you’re not going to win a lot in terms of performance.

From the same authors, i would recommend the Tempo work.

EDIT

From the Toplas paper:

Program specialization is a program
transformation technique that
optimizes a pro- gram fragment with
respect to information about a context
in which it is used, by generating an
implementation dedicated to this usage
context. One approach to automatic
program specialization is partial
evaluation, which performs aggressive
inter-procedural constant propagation
of values of all data types, and
performs constant folding and
control-flow simplifications based on
this information [Jones et al. 1993].
Partial evaluation thus adapts a
program to known (static) informa-
tion about its execution context, as
supplied by the user (the programmer).
Only the program parts controlled by
unknown (dynamic) data are
reconstructed. Par- tial evaluation
has been extensively investigated for
functional languages [Bondorf 1990;
Consel 1993], logic languages [Lloyd
and Shepherdson 1991], and imperative
languages [Andersen 1994; Baier et al.
1994; Consel et al. 1996].