Optimizing translation out of SSA using renaming constraints

Rastello, Fabrice; Ferrière, François de; Guillon, Christophe

doi:10.1109/cgo.2004.1281680

Cited by 11 publications

(15 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Aggressive coalescing can also minimize the number of copy instructions required to eliminate ϕ-functions during translation out of SSA Form. Heuristics for aggressive coalescing have been proposed by Sreedhar et al [37], Budimlic et al [11], Rastello et al [35] and Boissinot el al. [1].…”

Section: Related Workmentioning

confidence: 99%

Optimistic chordal coloring: a coalescing heuristic for SSA form programs

Brisk

Verma

Ienne

2008

Des Autom Embed Syst

View full text Add to dashboard Cite

The interference graph for a procedure in Static Single Assignment (SSA) Form is chordal. Since the k-colorability problem can be solved in polynomial-time for chordal graphs, this result has generated interest in SSA-based heuristics for spilling and coalescing. Since copies can be folded during SSA construction, instances of the coalescing problem under SSA have fewer affinities than traditional methods. This paper presents Optimistic Chordal Coloring (OCC), a coalescing heuristic for chordal graphs. OCC was evaluated on interference graphs from embedded/multimedia benchmarks: in all cases, OCC found the optimal solution, and ran, on average, 2.30× faster than Iterated Register Coalescing.

show abstract

Section: Related Workmentioning

confidence: 99%

Optimistic chordal coloring: a coalescing heuristic for SSA form programs

Brisk

Verma

Ienne

2008

Des Autom Embed Syst

View full text Add to dashboard Cite

show abstract

“…Actually, once the code is in CSSA, the optimization problem is a standard aggressive coalescing problem (i.e., with no constraints on the number of target variables) and heuristics exist for this NP-complete problem [19], [20]. The fact that the code is in SSA does not make it simpler or special.…”

Section: B Going Out Of Cssa: a Coalescing Problemmentioning

confidence: 99%

“…A pinned variable is a SSA variable pre-coalesced to another variable or pre-allocated to an architectural register [19]. To avoid interferences, we first ensure that a pinned variable has a short live range spanning no more than the constraining instruction.…”

Section: Handling Register Renaming Constraintsmentioning

confidence: 99%

Revisiting Out-of-SSA Translation for Correctness, Code Quality and Efficiency

Boissinot

Darte

Rastello

et al. 2009

2009 International Symposium on Code Generation and Optimization

View full text Add to dashboard Cite

Abstract-Static single assignment (SSA) form is an intermediate program representation in which many code optimizations can be performed with fast and easy-to-implement algorithms. However, some of these optimizations create situations where the SSA variables arising from the same original variable now have overlapping live ranges. This complicates the translation out of SSA code into standard code. There are three issues to consider: correctness, code quality (elimination of copies), and algorithm efficiency (speed and memory footprint). Briggs et al. proposed patches to correct the initial approach of Cytron et al. A cleaner and more general approach was proposed by Sreedhar et al., along with techniques to reduce the number of generated copies. We propose a new approach based on coalescing and a precise view of interferences, in which correctness and optimizations are separated. Our approach is provably correct and simpler to implement, with no patches or particular cases as in previous solutions, while reducing the number of generated copies. Also, experiments with SPEC CINT2000 show that it is 2x faster and 10x less memory-consuming than the Method III of Sreedhar et al., which makes it suitable for just-in-time compilation.

show abstract

“…Rastello et al [24] showed that translation out of an SSA form is NP-Complete and tried to minimize the number of copies inserted while also satisfying architecture-imposed naming constraints on variables in the post-SSA program. Budimlić et al [8] used Theorems 1 and 2 to develop a heuristic with an almost-linear time complexity that tried to minimize the number of copies inserted during translation out of SSA.…”

Section: B Translation Out Of Ssa Formmentioning

confidence: 99%

“…Moreover, once spilling decisions have been made, the placement of spill code is NP-Complete [26]. Finally, minimizing the number of copies inserted during SSA deconstruction is NP-Complete as well [24].…”

Section: Implications Of Theorem 3 For Software Compilersmentioning

confidence: 99%