You want a cheap high performance GPU for deep learning? In this blog post I will guide through the choices, so you can find the GPU which is best for you.
gpu.js is a single-file JavaScript library for GPGPU in the browser. gpu.js will automatically compile specially written JavaScript functions into shader language and run them on the GPU using the WebGL API. In the case where WebGL is not available, the functions will still run in regular JavaScript.
Marvin is a deep learning framework designed first and foremost to be hackable. It is naively simple for fast prototyping, uses only basic C/C++, and only calls CUDA and cuDNN as dependencies.
Y. Li, Y. Sun, and A. Jog. Proceedings of the 56th Annual IEEE/ACM International Symposium on Microarchitecture, MICRO 2023, Toronto, ON, Canada, 28 October 2023 - 1 November 2023, page 380--394. ACM, (2023)
E. Berger, S. Stern, and J. Pizzorno. 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23), Boston, MA, USENIX Association, (July 2023)
R. Okuta, Y. Unno, D. Nishino, S. Hido, and C. Loomis. Proceedings of Workshop on Machine Learning Systems (LearningSys) in The Thirty-first Annual Conference on Neural Information Processing Systems (NIPS), (2017)
M. Qunaibit, S. Brunthaler, Y. Na, S. Volckaert, and M. Franz. 32nd European Conference on Object-Oriented Programming (ECOOP 2018), volume 109 of Leibniz International Proceedings in Informatics (LIPIcs), page 16:1--16:29. Dagstuhl, Germany, Schloss Dagstuhl--Leibniz-Zentrum fuer Informatik, (2018)