Fast, Faster, Julia: High Performance Implementation of the NFFT | Tobias Knopp | JuliaCon 2022

[nHackel]

Speaker: Tobias Knopp Video url: https://www.youtube.com/watch?v=o9yF7HjkNG4&t=1303s

Content: 00:00 Welcome! 00:10 Background: FFT 02:27 Background: Limits of FFT 03:16 Non-Equidistant Fast Fourier Transform 05:03 Basic Idea of NFFT 06:39 Formula of the NFFT 08:15 Design Space 09:46 Introduction of NFFT.jl 11:25 Example of NFFT.jl 12:35 Dimension-Agnostic Implementations 14:55 Implementation Overview 16:15 Convolution: Idea 18:05 Convolution: Hot-Code Path 19:30 Adjoint Convolution 22:16 Performance and Accuracy Evaluation 22:52 Benchmark Results 22:45 Benchmark Results: Mutli-Threading 23:30 Summary and Outlook

[KZiemian]

I allowed myself to teak a little your work.

Content 00:00 Welcome! 00:10 Background: FFT 02:27 Background: Limits of FFT 03:16 Non-Equidistant Fast Fourier Transform 05:03 Basic Idea of NFFT 06:39 Formula of the NFFT 08:15 Design Space 09:46 Introduction of NFFT.jl 11:25 Example of NFFT.jl 12:35 Dimension-Agnostic Implementations 14:55 Implementation Overview 16:15 Convolution: Idea 18:05 Convolution: Hot-Code Path 19:30 Adjoint Convolution 22:16 Performance and Accuracy Evaluation 22:52 Benchmark Results 22:45 Benchmark Results: Mutli-Threading 23:30 Summary and Outlook

Resources NFFT.jl

[KZiemian]

Thank you. Timestamps were added to the video, we will close this issue, when we get access rights to this GitHub repository.

[KZiemian]

We can now close the issue.