Publications by authors named "Hung-Hsi Hsu"
Article Synopsis
- - The text discusses a new AI edge processor that uses a combination of memristor and SRAM technologies to enhance energy efficiency, rapid response time, and accuracy in compute-in-memory applications.
- - Traditional memristor-based systems struggle with accuracy and training capabilities, while SRAM-based systems have large area requirements and volatility; the new processor addresses these issues.
- - The fusion processor achieved impressive performance metrics, including a quick wakeup response of 392 microseconds and high energy efficiency, proving memristor technology is now viable for practical use in AI applications.
Article Synopsis
- Artificial Intelligence (AI) is getting much better thanks to deep learning, which uses lots of simple computer units working together.
- Traditional computers have trouble moving data quickly, so new methods like using memristors as memory devices can help solve this problem by being more efficient and powerful.
- This work explains how memristive neural networks work, their design options, and offers guidance for those interested in studying or improving these new technologies.
