Skip to content

Latest commit

 

History

History
35 lines (21 loc) · 1.01 KB

Cambricon-F Machine Learning Computers with Fractal von Neumann Architecture.md

File metadata and controls

35 lines (21 loc) · 1.01 KB
Raw

Cambricon-F: Machine Learning Computers with Fractal von Neumann Architecture

Corresponding author

Yunji Chen, SKL of Computer Architecture

Keywords

Machine learning hardware, programming productivity, model support

ML通用架构，流片验证

Summary

Challenge

With the fast development in silicon technology, programming productivity, including programming itself and software stack development, becomes the vital reason instead of performance and power efficiency that hinders the application of machine learning computer.

如何支持各种机器学习算法

Contribution

  1. Propose a series of homogeneous, sequential, multi-layer, layer-similar, machine learning computers with the same ISA. (架构设计)

  2. Implement two instances at different scales. (流片验证)

Result

Compared to GPU based machines (DGX-1 and 1080Ti), Cambricon-F instances achieve 2.82x, 5.14x better performance, 8.37x, 11.39x better efficiency on average, with 74.5%, 93.8% smaller area costs, respectively