{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/fast-convolutional-nets-with-fbfft-a-gpu","title":"Fast Convolutional Nets With fbfft: A GPU Performance Evaluation","arxiv_id":"1412.7580","date":"2014-12-24","proceeding":null,"authors":["Nicolas Vasilache","Jeff Johnson","Michael Mathieu","Soumith Chintala","Serkan Piantino","Yann Lecun"],"abstract":"We examine the performance profile of Convolutional Neural Network training\non the current generation of NVIDIA Graphics Processing Units. We introduce two\nnew Fast Fourier Transform convolution implementations: one based on NVIDIA's\ncuFFT library, and another based on a Facebook authored FFT implementation,\nfbfft, that provides significant speedups over cuFFT (over 1.5x) for whole\nCNNs. Both of these convolution implementations are available in open source,\nand are faster than NVIDIA's cuDNN implementation for many common convolutional\nlayers (up to 23.5x for some synthetic kernel configurations). We discuss\ndifferent performance regimes of convolutions, comparing areas where\nstraightforward time domain convolutions outperform Fourier frequency domain\nconvolutions. Details on algorithmic applications of NVIDIA GPU hardware\nspecifics in the implementation of fbfft are also provided.","url_abs":"http://arxiv.org/abs/1412.7580v3","url_pdf":"http://arxiv.org/pdf/1412.7580v3.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"fast-convolutional-nets-with-fbfft-a-gpu","repo_url":"https://github.com/facebook/fbcunn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"torch","reach":{"status":"ok","spdx":"NOASSERTION"}},{"paper_slug":"fast-convolutional-nets-with-fbfft-a-gpu","repo_url":"https://github.com/facebookarchive/fbcunn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"torch","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":null,"task_name":"GPU"}],"methods":[{"method_slug":"convolution","method_name":"Convolution"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1412.7580","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}