List scheduling: alone, with foresight, and with lookahead

Beaty, Steven J.

doi:10.1109/mpcs.1994.367017

Cited by 5 publications

(4 citation statements)

References 11 publications

(1 reference statement)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally the pipeline is drained (the epilogue). List scheduling (Beaty, 1994) is a common approach used to solve the scheduling problem. Some of the approaches carried out in modulo scheduling of the loop body are discussed below.…”

Section: Related Workmentioning

confidence: 99%

Analysis of Inner-Loop Mapping onto Coarse-Grained Reconfigurable Architectures Using Hybrid Particle Swarm Optimization

Gnanaolivu

Norvell

Venkatesan

2011

International Journal of Organizational and Collective Intelligence

View full text Add to dashboard Cite

Coarse-Grained Reconfigurable Architectures (CGRAs) have gained currency in recent years due to their abundant parallelism and flexibility. To utilize the parallelism found in CGRAs, this paper proposes a fast and efficient Modulo-Constrained Hybrid Particle Swarm Optimization (MCHPSO) scheduling algorithm to exploit loop-level parallelism in applications. This paper shows that Particle Swarm Optimization (PSO) is capable of software pipelining loops by overlapping placement, scheduling and routing of successive loop iterations and executing them in parallel. The proposed algorithm has been experimentally validated on various DSP benchmarks under two different architecture configurations. These experiments indicate that the proposed MCHPSO algorithm can find schedules with small initiation intervals within a reasonable amount of time. The MCHPSO scheduling algorithm was analyzed with different topologies and Functional Unit (FU) configurations. The authors have tested the parallelizability of the algorithm and found that it exhibits a nearly linear speedup on a multi-core CPU.

show abstract

Section: Related Workmentioning

confidence: 99%

Analysis of Inner-Loop Mapping onto Coarse-Grained Reconfigurable Architectures Using Hybrid Particle Swarm Optimization

Gnanaolivu

Norvell

Venkatesan

2011

International Journal of Organizational and Collective Intelligence

View full text Add to dashboard Cite

show abstract

“…Another study of scheduling alone, with foresight, and with lookahead scheduling 38 (using the IBM RS/6000 as a target) compares the effectiveness of plain list scheduling versus list scheduling with foresight and list scheduling with lookahead scheduling. The IBM RISC RS6000 39 allows an integer operation, a floating point operation and two branch unit operations to be issued in a single cycle.…”

Section: Effectiveness Of Lookahead Schedulingmentioning

confidence: 99%

“…Lookahead also schedules nodes without having to topologically sort them. By doing so it reduces the number of nodes LS must deal with and thereby increases the speed of scheduling [9]. Also note that naïve lookahead places nodes in the final schedule non-heuristically.…”

Section: E Lookaheadmentioning

confidence: 99%

Building a retargetable local instruction scheduler

Allan

Beaty

et al. 1998

Softw: Pract. Exper.

View full text Add to dashboard Cite

SUMMARYWhile high-performance architectures have included some Instruction-Level Parallelism (ILP) for at least 25 years, recent computer designs have exploited ILP to a significant degree. Although a local scheduler is not sufficient for generation of excellent ILP code, it is necessary as many global scheduling and software pipelining techniques rely on a local scheduler. Global scheduling techniques are well-documented, yet practical discussions of local schedulers are notable in their absence. This paper strives to remedy that disparity by describing a list scheduling framework and several important practical details that, taken together, allow implementation of an efficient local instruction scheduler that is easily retargetable for ILP architectures.The foundation of our machine-independent instruction scheduler is a timing model that allows easy retargetability to a wide range of architectures. In addition to describing how a general listscheduler can be implemented within the framework of our timing model, experimental results indicate that lookahead scheduling can profoundly improve a scheduler's ability to produce a legal schedule. Further experimental data shows that deciding to schedule a data dependence DAG (DDD) in forward or reverse order depends significantly upon that target architecture, suggesting the possibility of scheduling in each direction and using the best of the two schedules.In contrast, experiments demonstrate little difference in code quality for schedules generated by either instruction-driven or operation-driven schedulers. Thus, the inherent flexibility of operationdriven methods suggests including that approach in a retargetable instruction scheduler.List scheduling is, of course, a heuristic scheduling method. A variety of scheduling heuristics are presented. In addition, the paper describes a method, using a genetic algorithm search, to 'fine-tune' the weights of twenty-four individual heuristics to form a DDD-node heuristic tuned to a specific architecture.

show abstract

“…This greatly increases the likelihood of generating valid schedules, but breaks down when a more constraining operation is scheduled after an unconstrained one, due to factors such as having a higher priority. Beaty introduced a technique called Lookahead Scheduling [6,7], which not only looked at whether the placement of an operation would constrain future placements, but also placed those future operations so other unconstrained operations would not prevent a valid schedule from forming. Both the foresighted and lookahead techniques differ from our study in that they have a notion of valid and invalid schedules; their focus is mainly to place operations in any way possible to find a valid schedule.…”

Section: Related Workmentioning

confidence: 99%