[Long Ver.] How to create a high-speed AI chip by connecting a handful of multipliers and adders скачать в хорошем качестве

[Long Ver.] How to create a high-speed AI chip by connecting a handful of multipliers and adders

Summary Presented an efficient AI hardware design • Base architecture: Neuron Machine • Paper: http://arxiv.org/pdf/2002.05526 Design binary available and can be run directly on an FPGA board. Hardware source code posted on Github • http://github.com/ijcnnanonymous/mult... Development platform available 00:00 intro 02:54 Table of Contents 03:44 Input/Output and Function 05:17 Convolution Neural Network 07:52 Efficiency in AI Hardware? 09:40 Our Design Goals 12:37 Our Approach and Solutions to the Problems 19:51 System Architecture 22:56 How It Works 25:48 Efficiency Measure 28:42 Source Code 29:14 Summary If you have any further questions, please leave a comment or contact us via the email and contact information below. Thank you. [email protected]