Aizip Announces PIMNet for Processing-In-Memory Applications

Saratoga, CA, June 22, 2021 – Aizip, Inc., a Silicon Valley startup company providing best-in-class tiny artificial intelligence (AI) models for the Internet of Things (IoT), the so-called AIoT, announced its tiny machine learning (TinyML) design services for semiconductor chips based on Processing-In-Memory (PIM) technology.

PIM, also referred to as Computing-In-Memory (CIM) or In-Memory Computing (IMC), offers significant advantages in computing density and energy efficiency. It is generally believed to be the choice for next-generation AI accelerators, especially for the IoT market where low power operation is strongly desired. PIM has attracted tremendous attention and investment in both academia and industry in recent years. Multiple types of PIM technologies have been investigated, and much progress has been made to make PIM a reality. As the technology matures, several companies have started to offer PIM-based IC devices for AI applications, and more semiconductor companies are moving into the PIM space.

The design of deep neural networks (DNN) for PIM-based AI accelerators is very different from the design methodology for traditional digital AI accelerators. DNN performance may experience a large drop with existing design tools, and the application can be severely impacted. One fundamental difference is that the analog nature of PIM, such as analog to digital conversion, must be addressed. Another important parameter is maintaining efficiency with the built-in large throughput feature of PIM. Aizip is pioneering in investigating innovative neural networks and design tools for PIM devices and has built extensive IP and technical differentiation in the field. Leveraging its breakthrough neural network, PIMNet, Aizip has achieved DNN performance in PIM AI accelerators on par with their digital counterparts while maintaining PIM advantages in high-throughput and low-power operation.

“We’re very excited to offer this new design service to our customers,” said Hanlin Sun, a Product Manager at Aizip. “We believe Aizip is the first company to launch a PIM AI design service, and we’ve received strong interest and excellent feedback from our customers.” Sun further commented, “This pioneering release demonstrates Aizip’s technology leadership in the fast-growing market of TinyML.”

The general availability of the Aizip PIMNet should allow its customers to take advantage of PIM AI accelerators without compromising DNN performance. PIM-based intelligent solutions can reduce AIoT power consumption by more than 100x, reducing operation and service cost significantly with battery-powered systems. Aizip is rapidly expanding its offerings and customer base in the AIoT market. For more information, please contact info@aizip.ai.

* * *

About Aizip, Inc.

Aizip, Inc., develops AI models for edge applications. Based in Silicon Valley, Aizip provides models with superior performance, quick turnaround time, and excellent ROI. These models can be used in a wide range of applications for an intelligent, automated, and connected world. For more information, visit www.aizip.ai.