What is it about?
Designing application-specific approximate arithmetic operators presents a large design space. The problem becomes more complex if the characteristics of the hardware platform have to be included in the analysis. This paper explores the use of state-of-the-art AI-based exploration using Monte Carlo Tree Search (MCTS) for traversing such a large design space effectively.
Featured Image
Photo by Google DeepMind on Unsplash
Why is it important?
We show how modern advancements in AI, such as MCTS and analysis from explainable AI, can be used to traverse large design spaces effectively. Our findings show that computing approaches such as approximate computing, which presents the scope for fine granularity application-specific optimizations but suffers from a large design space, can benefit from such DSE approaches.
Perspectives
I hope this article shows the scope of using cutting-edge AI methods in the design of electronic systems. Although the scope of the article is limited to synthesizing novel approximate arithmetic operators, it presents a more generic approach to leveraging AI for electronic design automation.
Dr. Siva Satyendra Sahoo
IMEC
Read the Original
This page is a summary of: AxOTreeS
: A
Tree
S
earch Approach to Synthesizing FPGA-based
A
ppro
x
imate
O
perators, ACM Transactions on Embedded Computing Systems, September 2023, ACM (Association for Computing Machinery),
DOI: 10.1145/3609096.
You can read the full text:
Contributors
The following have contributed to this page
