The goal of this workshop is to encourage discussions about FFTs and to discover a benchmark suit that is widely accepted for computer co-design. This workshop is part of an initiative project that plans a series of meetings at different international venues with the aim of forming a community in order to collect the viewpoints of users, developers and vendors. The topics of these meetings can be found at http://www.fft.report/. FFT is an accurate low-cost algorithm which is extensively used in many parallel applications such as in signal processing, astronomy and fluid dynamics. However, it might not run well on a full parallel computer due to its Alltoall communication limitations. Consequently, it is of interest to the High Performance Computing community to address related challenges and seek solutions.
Keynote Speaker :Mahendra K. Verma, Dept of Physics, IIT Kanpur
Title : Challenges of Efficient Fast Fourier Transform |
Invited Speaker : Daisuke Takahashi, University of Tsukuba, Japan
Title : Automatic Tuning for Parallel FFTs on Cluster of Intel Xeon Phi Processors |
Invited Speaker : Anando G. Chatterjee, Dept of Physics, IIT Kanpur
Title : Development of an efficient FFT library and its scaling up to 65,536 cores of BlueGene/P and 196,608 cores of Cray
|
Time: December 17, 8:30 AM – 1:00 PM
Location: Arabica Hall
|
Time |
Event |
|
08:50-09:00 |
Opening Remarks by Workshop Chairs |
|
09:00-09:30 |
Keynote address: Mahendra K. Verma, Dept of Physics, IIT Kanpur “Challenges of efficient Fast Fourier Transform” |
|
09:30-10:00 |
Invited talk: Daisuke Takahashi, University of Tsukuba, Japan “Automatic Tuning for Parallel FFTs on Cluster of Intel Xeon Phi Processors” |
|
10:00-10:30 |
Contributed talk: Anumeena Sorna, Xiaohe Cheng, Eduardo D’azevedo, Kwai Wong and Stanimire Tomov. Optimizing the Fast Fourier Transform using Mixed Precision on Tensor Core Hardware |
|
10:30-11:00 |
Coffee Break |
|
11:00-11:30 |
Contributed talk: Semyon Khokhriakov, Ravi Reddy Manumachu and Alexey Lastovetsky. Performance Optimization of Multithreaded 2D FFT on Multicore Processors: Challenges and Solution Approaches |
|
11:30-12:00 |
Contributed talk: Franz Franchetti, Daniele Giuseppe Spampinato, Anuva Kulkarni, Doru Thom Popovici, Tze Meng Low, Michael Franusich, Andrew Canning, Peter McCorquodale, Brian Van Straalen and Phillip Colella. FFTX and SpectralPack: A First Look |
|
12:00-12:30 |
Invited Talk: Anando G. Chatterjee, Dept of Physics, IIT Kanpur “Development of an efficient FFT library and its scaling up to 65,536 cores of BlueGene/P and 196,608 cores of Cray XC40” |
|
12:30-13:00 |
Closing Remarks by Workshop Chairs |
|
13:00-14:00 |
Lunch Break |
CALL FOR PAPERS (CLOSED)
In supercomputing, FFT is an important and fundamental library for scientific and technical computing. Current hardware trends may make some Fourier transform algorithms somewhat inefficient. For papers, besides FFT applications we consider works on present status of the FFT libraries, alternative approximate Fourier transforms, alternate hardware designs and methods to compare performance of FFT, and alternative algorithms that will be useful in advancing the field.
We aim to explore how FFT could be implemented efficiently on large number of accelerators.
Manuscript Guidelines
Submitted manuscripts should be structured as technical papers and may not exceed ten (10) single-spaced double-column pages using 10-point size font on 8.5×11 inch pages (IEEE conference style), including figures, tables, and references. See style templates for details:
Electronic submissions must be in the form of a readable PDF file. All manuscripts will be reviewed by the Program Committee and evaluated on originality, relevance of the problem to the conference theme, technical strength, rigor in analysis, quality of results, and organization and clarity of presentation of the paper.
Submitted papers must represent original unpublished research that is not currently under review for any other conference or journal. Papers not following these guidelines will be rejected without review and further action may be taken, including (but not limited to) notifications sent to the heads of the institutions of the authors and sponsors of the conference.
Presentation of an accepted paper at the workshop is a requirement of publication. Any paper that is not presented at the conference will not be included in proceedings.
Submission Portal
Easy chair link: https://easychair.org/
Important Deadlines
August 15, 2018 September 7, 2018 September 30, 2018: Paper Submission deadline
September 7, 2018 September 17, 2018 October 17, 2018: Author Notification
October 3, 2018 October 29, 2018: Camera Ready Paper Submission deadline
Organising Committee
Samar Aseeri, King Abdullah University of Science and Technology, Saudi Arabia
Hari Sundar, University of Utah, USA
Daisuke Takahashi, University of Tsukuba, Japan
Mahendra K. Verma, Indian Institute of Technology Kanpur, India
Program Committee Members :
Samar Aseeri, King Abdullah University of Science and Technology, Saudi Arabia
Jens Henrik Gobbert, Julich Supercomputing Center, Germany
Ning Li, NAG, UK
Benson Muite, Tartu Ulikool, Astonia
Dmitry Pekurovsky, San Diego Supercomputing Center, USA
Hari Subramoni, Ohio State University, USA
Hari Sundar, University of Utah, USA
Daisuke Takahashi, University of Tsukuba, Japan
Ekaterina Tyutlyaeva, RSC Technologies, Russia
Mahendra K. Verma, Indian Institute of Technology Kanpur, India
Yu Feng, Berkeley Center for Cosmological Physics, USA