SRAM-based in-memory computing for machine learning applications
In memory computing has become popular recently. It not only could accelerate the AI application on hardware, but also could solve the Neumann problem. In this field, digital SRAM design for machine learning has received a lot of attention due to its easy design and high accuracy characteristics. In...
محفوظ في:
| المؤلف الرئيسي: | |
|---|---|
| مؤلفون آخرون: | |
| التنسيق: | Thesis-Master by Coursework |
| اللغة: | English |
| منشور في: |
Nanyang Technological University
2022
|
| الموضوعات: | |
| الوصول للمادة أونلاين: | https://hdl.handle.net/10356/156761 |
| الوسوم: |
إضافة وسم
لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!
|
| الملخص: | In memory computing has become popular recently. It not only could accelerate the AI application on hardware, but also could solve the Neumann problem. In this field, digital SRAM design for machine learning has received a lot of attention due to its easy design and high accuracy characteristics. In this paper, a 4k weight-selective digital SRAM design is implemented with improvements on Bitcell and Adder Tree. It uses TG logic in the design to improve the speed and eliminate power consumption. The weight-Selective function is used to adapt to the different complexity of the calculation. The simulation is done by using the TSMC65LP process. The Bitcell Array is 64x64, GOPS is 409.6 and the frequency is 200MHz. |
|---|