Code motion and code placement: Just synonyms?

Abstract

Abstract. We prove that there is no difference between code motion (CM) and code placement (CP) in the traditional syntactic setting, however, a dramatic difference in the semantic setting. We demonstrate this by re-investigating semantic CM under the perspective of the recent development of syntactic CM. Besides clarifying and highlighting the analogies and essential differences between the syntactic and the semantic approach, this leads as a side-effect to a drastical reduction of the conceptual complexity of the value-flow based procedure for semantic CM of [20], as the original bidirectional analysis is decomposed into purely unidirectional components. On the theoretical side, this establishes a natural semantical understanding in terms of the Herbrand interpretation (transparent equivalence), and thus eases the proof of correctness; moreover, it shows the frontier of semantic CM, and gives reason for the lack of algorithms going beyond. On the practical side, it simplifies the implementation and increases the efficiency, which, like for its syntactic counterpart, can be the catalyst for its migration from academia into industrial practice.

Value-Based Partial Redundancy Elimination

“…Certain extensions must be made to this algorithm for theoretical optimality (such as code placement in Knoop et al [21]). In practice, we have found them to yield no benefit, but they are described in a technical report [35].…”

Expansion-Based Removal of Semantic Partial Redundancies

“…This was discussed in detail in [12], and here we only recall an example for illustration. In the program fragment of Figure 3 in Figure 3(c).…”

“…This is because neither a+ b nor c + b are anticipable (down-safe) at the join point of control, and hence none of them can be hoisted across this point. In [12] we gave reasons why, under the constraint of structural invariance, the problem of designing a general placementbased SPRE-technique cannot satisfactorily be solved: even in acyclic programs there are redundancies which can only be eliminated at the cost of introducing other ones, which excludes optimality in general. Though the heuristically based extension of a motion-based SPRE-algorithm proposed by Bodík and Anik can be considered a step towards placement-based algorithms (cf.…”

“…These redun- State-of-the-art algorithms for SPRE (cf. [3,12,15,18,19]) are characterised by two common major design decisions made on the conceptual and technical side:…”

Interprocedural Herbrand Equalities

“…Knowledge about definite equalities can be exploited for performing and enhancing powerful optimizing program transformations. Examples include constant propagation, common subexpression elimination, and branch elimination [3,8], partial redundancy elimination and loopinvariant code motion [18,22,12], and strength reduction [23]. Clearly, it is undecidable whether two variables always have the same value at a program point even without interpreting conditionals [17].…”