Download SPAA: Symposium on Parallelism in Algorithms and Architectures

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Artificial intelligence wikipedia , lookup

Post-quantum cryptography wikipedia , lookup

Computational complexity theory wikipedia , lookup

Expectation–maximization algorithm wikipedia , lookup

Machine learning wikipedia , lookup

Cluster analysis wikipedia , lookup

Selection algorithm wikipedia , lookup

Algorithm characterizations wikipedia , lookup

Lateral computing wikipedia , lookup

General-purpose computing on graphics processing units wikipedia , lookup

Factorization of polynomials over finite fields wikipedia , lookup

Sorting algorithm wikipedia , lookup

Stream processing wikipedia , lookup

Fast Fourier transform wikipedia , lookup

Pattern recognition wikipedia , lookup

Natural computing wikipedia , lookup

Time complexity wikipedia , lookup

Algorithm wikipedia , lookup

Genetic algorithm wikipedia , lookup

Theoretical computer science wikipedia , lookup

Transcript
SPAA: Symposium on Parallelism in Algorithms
and Architectures
Petr Barta
March 26, 2017
1
Summary
The ACM Symposium on Parallelism in Algorithms and Architectures is an
academic conference in the fields of parallel computing and distributed computing. It was first organised in 1989, so in 2016 there was 28th edition of this
conference. It took place at the Asilomar Conference Grounds near Moneterey,
California, USA on July 2016. During the conference 103 papers were presented.
1.1
Topics of interest
• Parallel and Distributed Algorithms
• Parallel and Distributed Architectures
• Parallel and Distributed Data Structures
• Multiprocessor and Multicore Architectures
• Parallel Complexity Theory
• Transactional Memory Hardware and Software
• Scheduling in Parallel Systems
• Instruction Level Parallelism and VLSI
• Specification and Verification of Concurrent Systems
• Compilers and Tools for Concurrent Programming
• Mobile, Ad-Hoc, Wireless and Sensor Networks
• Algorithms for GPUs and Other Alternative Parallel Architectures
• Fault-tolerance and Reliability
• High-Performance Parallel Computing and Architectures
1
• Self-stabilization and Self-organization
• Green & Power-Efficient Algorithms and Architectures
• Security and Privacy in Distributed and Parallel Systems
• Algorithms for Social Networks
• Parallel/Distributed Computational Learning
• Biological Distributed Algorithms
• Game Theory and Collaborative Learning
• Network Algorithms
• Parallel/Distributed issues in Big Data
• Algorithms for Routing and Information Dissemination
• Resource Management and Awareness
• Peer-to-Peer Systems
2
2.1
Papers
Parallel Algorithms for Summing Floating-Point Numbers
In this paper Ahmed Eldawy and Michael T. Goodrich presented their algotitmhs for summing up floating point numbers on (EREW) PRAM, External
memory and MapReduce models. Their PRAM algorithm, is (as far as the authors know) very first alogithm solving this problem with time complexity O(log
n) using n processors. Algorithm for extrnal memory model is able to sum up
n numbers in O((n/B) logM/B(n/B)) I/O operations, where B is the block
ize and M is the size of internal memory. MapReduce models time complexity
O(o(n/p)p + n/p), where p < n is the number of processors.
2.2
Provably Good and Practically Efficient Parallel Race
Detection for Fork-Join Programs
When it comes to parallel programming, the data races is pretty common problem we have to deal with. For detecting these bugs, there are several race
detectors, which key component is a series-parallel maintenance algorithm. In
this paper Robert Utterback, Kunal Agrawal, Jeremy T. Fineman and I-Ting
Angelina Lee introduce the asymptotically optimal WSP-Order algorithm. The
second part of the paper describes C-RACER, a race detector using WSP-Order
algorithm, and its performance.
2
2.3
Parallel Algorithms for Asymmetric Read-Write Costs
Because of significantly higher cost of writing than reading in nowadays, but
also in near future memories, Naama Ben-David, Guy E. Blelloch, Jeremy T.
Fineman, Phillip B. Gibbons, Yan Gu, Charles McGuffey and Julian Shun came
up with Asymmetric Nested-Parallel model of computation, that tries to keep
the number of writing operations as low as possible. This model adds a heap
allocated asymmetric large memory to the Nested-parallel (also known as Forkjoin) model. The paper also contains several algorithms which use this model.
References
[1] Michael Goodrich and Ahmed Eldawy. Parallel Algorithms for Summing
Floating-Point Numbers.
https://arxiv.org/pdf/1605.05436.pdf
[2] Robert Utterback, Kunal Agrawal, Jeremy Fineman and I-Ting Angelina
Lee. Provably good and practically efficient parallel race detection for forkjoin programs.
http://www.cse.wustl.edu/~angelee/papers/racer.pdf
[3] Naama Ben David, Guy Blelloch, Jeremy Fineman, Phillip Gibbons,
Yan Gu, Charles McGuffey and Julian Shun. Parallel Algorithms for
Asymmetric Read-Write Costs.
https://people.eecs.berkeley.edu/~jshun/parallel-readwrite.pdf
3