ARM Unveils Project Trillium Machine Learning Processor
One of the giants in the field of CPU and GPU processors is the ARM. It has made its own name in coming up with computer and graphics processing units with the best features. Despite its success in the said field, the ARM does not hesitate to show what it can do in other fields like in the creation of a machine learning processor.
Introducing ARM's Project Trillium MLP
Based on William Judd's article on Developer, ARM recently unveiled its Project Trillium machine learning processor (MLP). This MLP makes it possible for smartphones and tablets to do machine learning on its own. The engineers from the CPU and GPU teams of ARMS are the ones behind this new MLP.
3 Major Points Highlighted by Engineers in Building this MLP...
1. Efficient Convolutions
2. Efficient Data Movement
3. Sufficient Programmability
What are the advantages of this Project Trillium MLP?
- ARM's Project Trillium MLP makes it possible for smartphones and tables to do machine learning even if they are not connected to the server.
- Less bandwidth and power are needed for smartphones to do machine learning.
- This MLP masks repeating zeroes in a certain block achieving a 4:1 compression ratio.
- Codes written using ARM NN for Android and Embedded Linux can work on this MLP.
Schedule of Release of ARM's Project Trillium MLP
This machine learning processor is scheduled to be released in the middle of the year 2018. Based on the article, this MLP is going to be available on 16nm and 7nm fabrication nodes.
What can you say about it? Feel free to drop your comments below. Thank you for reading this article. You may read more blog posts here in sourcecodester.com.