• Corpus ID: 113407064

Parallel AI Planning on the SCC

@inproceedings{Vidal2011ParallelAP,
  title={Parallel AI Planning on the SCC},
  author={Vincent Vidal and Simon Vernhes and Guillaume Infantes},
  year={2011}
}
We present in this paper a parallelized version of an existing Artificial Intelligence automated planner, implemented with standard programming models and tools (hybrid OpenMP/MPI). We then evaluate this planner with respect to its sequential version through extensive experiments over a wide range of academic benchmarks, on two different target architectures: a small standard cluster and the research processor SCC (“Single-chip Cloud Computer”) developed by Intel Labs and made available to the… 

Figures from this paper

Performance of RDF Query Processing on the Intel SCC
TLDR
This study uses the Intel SCC, an experimental manycore processor from Intel Labs, and concludes that the task parallelism model provides an immediate way to boost the performance of RDF query processing.
Landmark-based Meta Best-First Search Algorithm : First Parallelization Attempt and Evaluation
TLDR
A new method for problem splitting based on landmarks is proposed which has two advantages over the original technique: it is complete (if a solution exists, the algorithm finds it), and it uses the precedence relation over the landmarks in a more flexible way.
The Landmark-based Meta Best-First Search Algorithm for Classical Planning
TLDR
This paper lays the foundations of a meta best-first search algorithm, which explores the landmark orderings and can use any embedded planner to solve each subproblem, and opens up avenues for future research.
Transposition Driven Scheduling: Back to the Future? A study on vectorization and shared CPU programming
TLDR
Results show that it is plausible that TDS has the potential to hide the latency of locks by exchanging them for communication and can be rewritten in a form, called Batched TDS, that processes a set of states at once instead of one state at a time, making it a candidate for vectorization using SIMD instructions.
Transposition Driven Scheduling : Back to the Future ? A study on vectorization and shared memory CPU programming Name :
TLDR
Results show that it is plausible that TDS has the potential to hide the latency of locks by exchanging them for communication and can be rewritten in a form, called Batched TDS, that processes a set of states at once instead of one state at a time, making it a candidate for vectorization using SIMD instructions.
On Hash-Based Work Distribution Methods for Parallel Best-First Search
TLDR
This work proposes Abstract Zobrist hashing, a new work distribution method for parallel search which, instead of computing a hash value based on the raw features of a state, uses a feature projection function to generate a set of abstract features which results in a higher locality, resulting in reduced communications overhead.

References

SHOWING 1-10 OF 16 REFERENCES
Scalable, Parallel Best-First Search for Optimal Sequential Planning
TLDR
This work focuses on an approach which distributes and schedules work among processors based on a hash function of the search state, and uses this approach to parallelize the A* algorithm in the optimal sequential version of the Fast Downward planner.
Adaptive K-Parallel Best-First Search: A Simple but Efficient Algorithm for Multi-Core Domain-Independent Planning
TLDR
This work considers parallel versions of a best-first search algorithm that runK threads, each expanding the next best node from the open list, and shows that the approach is promising for parallel domain-independent, suboptimal planning.
AyAlsoPlan: Bitstate Pruning for State-Based Planning on Massively Parallel Compute Clusters
TLDR
This extended abstract proposes an answer to how to take advantage of modern multiple-processor computers and the proliferation of massively parallel compute clusters to address the memory problem.
ArvandHerd: Parallel Planning with a Portfolio
TLDR
It is shown that ArvandHerd can solve more IPC benchmark problems than even a perfect parallelization of LAMA-2011, which won the satisficing track at IPC 2011.
A Robust and Fast Action Selection Mechanism for Planning
TLDR
A variation of Korf's Learning Real Time A* algorithm together with a suitable heuristic function is developed by looking at planning as real time search and the resulting algorithm interleaves lookahead with execution and never builds a plan.
The FF Planning System: Fast Plan Generation Through Heuristic Search
TLDR
A novel search strategy is introduced that combines hill-climbing with systematic search, and it is shown how other powerful heuristic information can be extracted and used to prune the search space.
Transposition Table Driven Work Scheduling in Distributed Game-Tree Search
MTD(f) is a new variant of the ?s algorithm that has become popular amongst practitioners. TDS is a new parallel search algorithm that has proven to be effective in the single-agent domain. This
Sequential and Parallel Algorithms for Frontier A* with Delayed Duplicate Detection
TLDR
The results of an experimental evaluation of the implementation of PFA*-DDD are presented where it is used to solve instances of the multiple sequence alignment problem on a cluster of workstations running on top of a commodity network.
A Lookahead Strategy for Heuristic Search Planning
TLDR
This work presents a novel way for extracting information from the relaxed plan and for dealing with helpful actions, by considering the high quality of the relaxed plans in numerous domains, in a complete best-first search algorithm.
The Fast Downward Planning System
  • M. Helmert
  • Computer Science
    J. Artif. Intell. Res.
  • 2006
TLDR
A full account of Fast Downward's approach to solving multivalued planning tasks is given and a new non-heuristic search algorithm called focused iterative-broadening search, which utilizes the information encoded in causal graphs in a novel way is presented.
...
...