Register Allocation for Programs in SSA-Form

Hack, Sebastian; Grund, Daniel; Goos, Gerhard

doi:10.1007/11688839_20

Cited by 91 publications

(134 citation statements)

References 17 publications

Supporting

Mentioning

133

Contrasting

Unclassified

Order By: Relevance

“…However, maintaining SSA form provides an advantage for register allocation: Due to the live-range splits that are induced by the φ-functions, the interference graph of programs in SSA form is chordal [4,1,13], which means every induced subgraph that is a cycle, has length three. For chordal graphs the chromatic number is determined by the size of the largest clique.…”

Section: Register Allocation On Ssa Formmentioning

confidence: 99%

“…This means that the interference graph is k-colorable, if the spiller has reduced the register pressure to at most k at each program point. Thus, spilling can be decoupled from assignment [13], which means that the process is not iterated as with Graph Coloring.…”

Section: Register Allocation On Ssa Formmentioning

confidence: 99%

“…A PEO defines an ordering < of nodes of a graph G, such that each successively removed node is simplicial, which means it forms a clique with all remaining neighbors in G. After spilling, assigning the registers in reverse PEO ensures that each node is simplicial and thus has a free register available. Following Hack, we obtain a PEO by a postorder walk on the dominance tree [11].…”

Section: Register Allocation On Ssa Formmentioning

confidence: 99%

“…Since both constraints are contradictory, there is no coloring of the interference graph. To prevent such situations, Hack splits all live-ranges before the constrained instruction [11]. For our example, this allows to fulfill the constraints by inserting a copy from register c to register a or b.…”

Section: Register Constraintsmentioning

confidence: 99%

“…Common register allocation algorithms like Graph Coloring [3,6] spill on demand during the allocation. This can result in an increased number of spills and reloads [11]. In contrast, register allocation based on static single assignment (SSA) form allows to completely decouple the spilling phase.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

SSA-Based Register Allocation with PBQP

Buchwald

Zwinkau

Bersch

2011

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Recent research shows that maintaining SSA form allows to split register allocation into separate phases: spilling, register assignment and copy coalescing. After spilling, register assignment can be done in polynomial time, but copy coalescing is NP-complete. In this paper we present an assignment approach with integrated copy coalescing, which maps the problem to the Partitioned Boolean Quadratic Problem (PBQP). Compared to the state-of-the-art recoloring approach, this reduces the relative number of swap and copy instructions for the SPEC CINT2000 benchmark to 99.6% and 95.2%, respectively, while taking 19% less time for assignment and coalescing.

show abstract

Section: Register Allocation On Ssa Formmentioning

confidence: 99%

Section: Register Allocation On Ssa Formmentioning

confidence: 99%

Section: Register Allocation On Ssa Formmentioning

confidence: 99%

Section: Register Constraintsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

SSA-Based Register Allocation with PBQP

Buchwald

Zwinkau

Bersch

2011

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

Register allocation and spilling using the expected distance heuristic

Burroughs

2016

Softw Pract Exp

View full text Add to dashboard Cite

Summary The primary goal of the register allocation phase in a compiler is to minimize register spills to memory. Spill decisions by the allocator are often made based on the costs of spilling a virtual register and, therefore, on an assumed placement of spill instructions. However, because most allocators make these decisions incrementally, placement opportunities can change as allocation proceeds, calling into question the basis for the original spill decision. An alternative heuristic to placement costs for spill decisions focuses on where program execution will lead. Spilling the virtual register with the Furthest Next Use is known to lead to the minimum number of loads under certain conditions in straight‐line code. While it has been implemented in register allocation in different forms, none of these implementations fully exploits profiling information. We present a register allocator that can adapt to improved profiling information, using branch probabilities to compute an Expected Distance to Next Use for making spill decisions and block frequency information to optimize post‐allocation spill instruction placement. Spill placement is optimized after allocation using a novel method for minimizing spill instruction costs on the control flow graph. Our evaluation of the allocator compared with LLVM recognizes more than 36% and 50% reductions, on average, in the number of dynamically executed store and load instructions, respectively, when using statically derived profiling information. When using dynamically gathered profiling, these improvements increase to 50% and 60% reductions, on average, for stores and loads, respectively. Copyright © 2016 John Wiley & Sons, Ltd.

show abstract

Register Allocation for Programs in SSA-Form

Hack

Grund

Goos

2006

Lecture Notes in Computer Science

133

View full text Add to dashboard Cite

Abstract. As register allocation is one of the most important phases in optimizing compilers, much work has been done to improve its quality and speed. We present a novel register allocation architecture for programs in SSA-form which simplifies register allocation significantly. We investigate certain properties of SSA-programs and their interference graphs, showing that they belong to the class of chordal graphs. This leads to a quadratic-time optimal coloring algorithm and allows for decoupling the tasks of coloring, spilling and coalescing completely. After presenting heuristic methods for spilling and coalescing, we compare our coalescing heuristic to an optimal method based on integer linear programming.

show abstract

Register Allocation for Programs in SSA-Form

Cited by 91 publications

References 17 publications

SSA-Based Register Allocation with PBQP

SSA-Based Register Allocation with PBQP

Register allocation and spilling using the expected distance heuristic

Register Allocation for Programs in SSA-Form

Contact Info

Product

Resources

About