Support of Probabilistic Pointer Analysis in the SSA Form

Abstract

Probabilistic pointer analysis (PPA) is a compile-time analysis method that estimates the probability that a points-to relationship will hold at a particular program point. The results are useful for optimizing and parallelizing compilers, which need to quantitatively assess the profitability of transformations when performing aggressive optimizations and parallelization. This paper presents a PPA technique using the static single assignment (SSA) form. When computing the probabilistic points-to relationships of a specific pointer, a pointer relation graph (PRG) is first built to represent all of the possible points-to relationships of the pointer. The PRG is transformed by a sequence of reduction operations into a compact graph, from which the probabilistic points-to relationships of the pointer can be determined. In addition, PPA is further extended to interprocedural cases by considering function related statements. We have implemented our proposed scheme including static and profiling versions in the Open64 compiler, and performed experiments to obtain the accuracy and scalability. The static version estimates branch probabilities by assuming that every conditional is equally likely to be true or false, and that every loop executes 10 times before terminating. The profiling version measures branch probabilities dynamically from past program executions using a default workload provided with the benchmark. The average errors for selected benchmarks were 3.80 percent in the profiling version and 9.13 percent in the static version. Finally, SPEC CPU2006 is used to evaluate the scalability, and the result indicates that our scheme is sufficiently efficient in practical use. The average analysis time was 35.59 seconds for an average of 98,696 lines of code.

Accelerating Dynamic Data Race Detection Using Static Thread Interference Analysis

“…Then, an iterative flow-and context-sensitive analysis is performed on the inter-procedural control-flow by matching calls and returns (line [33][34]. If the interleaving information I of a context-and thread-sensitive statement changes, it will be added into the work list for fixed point resolution (line [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36]. For a fork site during resolution, the spawner thread t is added to the interleaving set of the entry statement in a spawnee's start routine procedure (line [19][20].…”

“…Compared to pointer analysis for sequential programs [17,45,43,44,51,19], flow-sensitive analysis for multithreaded programs is relatively unexplored. Earlier, Rugina and Rinard [35] introduced a pointer analysis for Clik programs with structured parallelism.…”

Probabilistic Alias Analysis for Parallel Programming in SSA Forms

“…Our proposed technique assumes that the given program, that is to be be analyzed for its probabilistic alias competent, has the static single assignments form. Therefore the input program would contain annotations (added by any efficient SSA such as algorithm [11]). The program annotations will have the form of new statements added to the original program.…”

“…Aiming at evaluating, among array references, the probabilities of aliases, [18], [19], [20] presents a probabilistic technique for memory disambiguation. A probabilistic, interprocedural, contextsensitive, and flow-sensitive techniques for alias analysis were proposed in [3], [11], [21]. On alias relationships, these technique evaluate measurable information.…”

Probabilistic data flow analysis: a linear equational approach

“…Speculative optimisation [15,2] has been an element of hardware design for some time, in particular to branch prediction [16] or for cache optimisation [17]. More recently, related ideas have also been discussed in the context of speculative multi-threading [4] or probabilistic pointer analysis [9,13].…”