Detecting equality of variables in programs
POPL '88 Proceedings of the 15th ACM SIGPLAN-SIGACT symposium on Principles of programming languages
An efficient method of computing static single assignment form
POPL '89 Proceedings of the 16th ACM SIGPLAN-SIGACT symposium on Principles of programming languages
The value flow graph: a program representation for optimal program transformations
Proceedings of the third European symposium on programming on ESOP '90
Efficiently computing static single assignment form and the control dependence graph
ACM Transactions on Programming Languages and Systems (TOPLAS)
Effective partial redundancy elimination
PLDI '94 Proceedings of the ACM SIGPLAN 1994 conference on Programming language design and implementation
Translation validation for an optimizing compiler
PLDI '00 Proceedings of the ACM SIGPLAN 2000 conference on Programming language design and implementation
A sparse algorithm for predicated global value numbering
PLDI '02 Proceedings of the ACM SIGPLAN 2002 Conference on Programming language design and implementation
A unified approach to global program optimization
POPL '73 Proceedings of the 1st annual ACM SIGACT-SIGPLAN symposium on Principles of programming languages
Symbolic evaluation and the global value graph
POPL '77 Proceedings of the 4th ACM SIGACT-SIGPLAN symposium on Principles of programming languages
POPL '77 Proceedings of the 4th ACM SIGACT-SIGPLAN symposium on Principles of programming languages
Value-driven redundancy elimination
Value-driven redundancy elimination
Global value numbering using random interpretation
Proceedings of the 31st ACM SIGPLAN-SIGACT symposium on Principles of programming languages
Partial value number redundancy elimination
LCPC'04 Proceedings of the 17th international conference on Languages and Compilers for High Performance Computing
| Hi-index | 0.00 |
Global value numbering (GVN) is an important static analysis technique both for optimizing compilers and program verification tools. Existing complete GVN algorithms discovering all Herbrand equivalences are all inefficient. One reason of this is the intrinsic exponential complexity of the problem, but in practice, since the exponential case is quite rare, the more important reason is the huge data structures annotated to every program point and slow abstract evaluations on them site by site. In this paper, we present an SSA-based algorithm for complete GVN, which uses just one global graph to represent all equivalences at different program points and performs fast abstract evaluations on it. This can be achieved because in SSA form, interferences among equivalence relations at different program points can be entirely resolved with dominance information. We implement the new algorithm in GCC. The average proportion of execution time of the new algorithm in the total compilation time is only 0.36%. To the best of our knowledge, this is the first practical complete GVN algorithm.