Parakeet: A just-in-time parallel accelerator for python

Alex Rubinsteyn; Eric Hielscher; Nathaniel Weinman; Dennis Shasha

Parakeet: A just-in-time parallel accelerator for python

Alex Rubinsteyn, Eric Hielscher, Nathaniel Weinman, Dennis Shasha

Research output: Contribution to conference › Paper › peer-review

Abstract

High level productivity languages such as Python or Matlab enable the use of computational resources by nonexpert programmers. However, these languages often sacrifice program speed for ease of use. This paper proposes Parakeet, a library which provides a just-in-time (JIT) parallel accelerator for Python. Parakeet bridges the gap between the usability of Python and the speed of code written in efficiency languages such as C++ or CUDA. Parakeet accelerates data-parallel sections of Python that use the standard NumPy scientific computing library. Parakeet JIT compiles efficient versions of Python functions and automatically manages their execution on both GPUs and CPUs. We assess Parakeet on a pair of benchmarks and achieve significant speedups.

Original language	English (US)
State	Published - 2012
Event	4th USENIX Workshop on Hot Topics in Parallelism, HotPar 2012 - Berkeley, United States Duration: Jun 7 2012 → Jun 8 2012

Conference

Conference	4th USENIX Workshop on Hot Topics in Parallelism, HotPar 2012
Country/Territory	United States
City	Berkeley
Period	6/7/12 → 6/8/12

ASJC Scopus subject areas

Software
Theoretical Computer Science
Computational Theory and Mathematics

Cite this

@conference{5daf8c75a062461190a0136ec1d0a23d,

title = "Parakeet: A just-in-time parallel accelerator for python",

abstract = "High level productivity languages such as Python or Matlab enable the use of computational resources by nonexpert programmers. However, these languages often sacrifice program speed for ease of use. This paper proposes Parakeet, a library which provides a just-in-time (JIT) parallel accelerator for Python. Parakeet bridges the gap between the usability of Python and the speed of code written in efficiency languages such as C++ or CUDA. Parakeet accelerates data-parallel sections of Python that use the standard NumPy scientific computing library. Parakeet JIT compiles efficient versions of Python functions and automatically manages their execution on both GPUs and CPUs. We assess Parakeet on a pair of benchmarks and achieve significant speedups.",

author = "Alex Rubinsteyn and Eric Hielscher and Nathaniel Weinman and Dennis Shasha",

note = "Publisher Copyright: {\textcopyright} HotPar 2012.; 4th USENIX Workshop on Hot Topics in Parallelism, HotPar 2012 ; Conference date: 07-06-2012 Through 08-06-2012",

year = "2012",

language = "English (US)",

}

TY - CONF

T1 - Parakeet

T2 - 4th USENIX Workshop on Hot Topics in Parallelism, HotPar 2012

AU - Rubinsteyn, Alex

AU - Hielscher, Eric

AU - Weinman, Nathaniel

AU - Shasha, Dennis

PY - 2012

Y1 - 2012

N2 - High level productivity languages such as Python or Matlab enable the use of computational resources by nonexpert programmers. However, these languages often sacrifice program speed for ease of use. This paper proposes Parakeet, a library which provides a just-in-time (JIT) parallel accelerator for Python. Parakeet bridges the gap between the usability of Python and the speed of code written in efficiency languages such as C++ or CUDA. Parakeet accelerates data-parallel sections of Python that use the standard NumPy scientific computing library. Parakeet JIT compiles efficient versions of Python functions and automatically manages their execution on both GPUs and CPUs. We assess Parakeet on a pair of benchmarks and achieve significant speedups.

AB - High level productivity languages such as Python or Matlab enable the use of computational resources by nonexpert programmers. However, these languages often sacrifice program speed for ease of use. This paper proposes Parakeet, a library which provides a just-in-time (JIT) parallel accelerator for Python. Parakeet bridges the gap between the usability of Python and the speed of code written in efficiency languages such as C++ or CUDA. Parakeet accelerates data-parallel sections of Python that use the standard NumPy scientific computing library. Parakeet JIT compiles efficient versions of Python functions and automatically manages their execution on both GPUs and CPUs. We assess Parakeet on a pair of benchmarks and achieve significant speedups.

UR - http://www.scopus.com/inward/record.url?scp=84907078956&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84907078956&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:84907078956

Y2 - 7 June 2012 through 8 June 2012

ER -

Parakeet: A just-in-time parallel accelerator for python

Abstract

Conference

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this