Saratoga, CA, March 29, 2022 – Aizip, Inc., a Silicon Valley startup company providing leading artificial intelligence (AI) solutions for the Internet of Things (IoT), or the so-called tiny machine learning (TinyML), announced that its full-stack design services for Processing-In-Memory (PIM) AI chips are ready for its worldwide customers.
Processing-in-memory (PIM) AI accelerator performs efficient analog computation directly within dense memory elements, leading to significant power and cost savings over conventional hardware. As a result, PIM has been widely regarded as the technology of choice to realize next-generation AI processors for edge/IoT applications.
After years of research, PIM is now undergoing a transition into productization. Aizip is at the forefront to enable this transition by offering full-stack PIM system design services to IC companies. The list of services includes silicon-verified PIM IPs, chip-architecture designs, PIM-optimized neural networks (PIMNet), and PIM-aware training frameworks. Aizip further tailors its design services to specific customer needs at each stage of IC product development. Aizip’s full-stack co-design ensures PIM products support desired AI applications while delivering superior energy-efficiency and DNN performance.
“Productizing PIM for AI inference at the edge is the logical next step” said H.-S. Philip Wong, a professor at Stanford University and a fellow of the IEEE. His group at Stanford is pioneering PIM research on multiple fronts. Wong recently joined Aizip to serve on its board of advisors: “I’m excited to join Aizip. Making PIM a reality requires thinking top-down from a system perspective. Aizip is taking the right approach and is well positioned to take leadership in the space.”
“Co-optimizing hardware, software, and model design is the key to unleash the full power of PIM,” said Weier Wan, who has led multiple successful tape-outs of PIM chips and is currently leading the PIM hardware-software co-design effort at Aizip. According to Wan, who worked in Professor Wong’s group for his Ph.D. at Stanford University and is a founding member of Aizip, the company has developed full-stack solution that allows DNN inference accuracy on PIM hardware to be on par with that achieved by high-precision digital hardware, while not trading-off DNN model size. Aizip has been working with several IC partners and has recently signed IP sub-licensing and design service agreements with excellent feedback.
With a team of top researchers and engineers bringing full-stack know-how from data and model to system and hardware, Aizip’s PIM solution has been well developed to create an immediate differentiation in the edge/AIoT market. Aizip is dedicated to helping its IC partners to develop next-generation low-power and low-cost edge intelligence platforms. 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.