FLUPS - A Flexible and Performant Massively Parallel Fourier Transform Library

نویسندگان

چکیده

Massively parallel Fourier transforms are widely used in computational sciences, and specifically fluid dynamics which involves unbounded Poisson problems. In practice the latter is usually most time-consuming operation due to its inescapable all-to-all communication pattern. The original flups library tackles that issue with an implementation of distributed transform tailor-made for successive resolutions However proposed lacks flexibility as it only supports cell-centered data layout features a plain strategy. This work extends along two directions. First, ' generalized support node-centered layout. Second, three distinct approaches provided handle communications: one all-to-all, non-blocking implementations relying on manual packing xmlns:xlink="http://www.w3.org/1999/xlink">MPI_Datatype communicate over network. software validated against analytical solutions unbounded, semi-unbounded, periodic domains. performance then compared xmlns:xlink="http://www.w3.org/1999/xlink">accFFT , another FFT implementation, using case. Finally metrics each analyzed detailed various top-tier European facilities up 49,152 cores. brings fully production-ready performant library, featuring all possible types FFTs data-layout. code available under BSD-3 license at github.com/vortexlab-uclouvain/flups .

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

COSMOS: Python library for massively parallel workflows

SUMMARY Efficient workflows to shepherd clinically generated genomic data through the multiple stages of a next-generation sequencing pipeline are of critical importance in translational biomedical science. Here we present COSMOS, a Python library for workflow management that allows formal description of pipelines and partitioning of jobs. In addition, it includes a user interface for tracking ...

متن کامل

Continuous Wavelet Transform on Massively Parallel Arrays

Strategies for computing the continuous wavelet transform on massively parallel SIMD computers are introduced. The diierent approaches are evaluated from the theoretical as well as from the experimental point of view.

متن کامل

Ordered Fast Fourier Transforms on a Massively Parallel Hypercube Multiprocessor

We examine design alternatives for ordered FFT algorithms on massively parallel hypercube multiprocessors such as the Connection Machine. Particular emphasis is placed on reducing communication which is known to dominate the overall computing time. To this end we combine the order and computational phases of the FFT and also use sequence to processor maps that reduce communication. The class of...

متن کامل

PARMONC - A Software Library for Massively Parallel Stochastic Simulation

We assume that a functional of interest ϕ ∈ R is represented as an expectation of some random variable ζ: ϕ ≈ Eζ One evaluates the value of Eζ using a sample mean Eζ ≈ ¯ ζ = L −1 L i=1 ζ i where sample values ζ i are independent and identically distributed random variables having the same distribution as ζ. One also needs to evaluate a second moment Eζ 2 of the random variable Eζ 2 ≈ ¯ ξ = L −1...

متن کامل

Parallel Fast Fourier Transform Literature Review

The fast Fourier transform (FFT), developed in 1965, is widely considered one of the most prolific and significant algorithms of the 20th century. While the classic Fourier Transform algorithm converts data from timedomain to frequency domain in O(N), the FFT reduces the complexity to O(NlgN). This performance improvement has had a significant impact on many fields and has revolutionized the ar...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

ژورنال

عنوان ژورنال: IEEE Transactions on Parallel and Distributed Systems

سال: 2023

ISSN: ['1045-9219', '1558-2183', '2161-9883']

DOI: https://doi.org/10.1109/tpds.2023.3254302