MemryX, Cachengo partner to deliver high-performance, scalable analytics, AI platform - The EE

MemryX, Cachengo partner to deliver high-performance, scalable analytics, AI platform

Ann Arbor, United States – MemryX Inc., a pioneering startup focused on accelerating artificial intelligence (AI) processing for edge devices, announced it has formed a partnership with Cachengo, an AI hardware and software provider, to be the provider of artificial intelligence processors for their modular compute and storage servers.

Cachengo’s modular edge servers are optimised for local computer vision and AI-centric workloads that benefit from massive parallelism, low power, and security. Coupled with MemryX’s low-power, on-board screaming AI accelerator and scalability of up to 8 parallel processors, the combination unleashes unlimited opportunity for AI at the edge.

“We are thrilled to partner with Cachengo to bring the power of MemryX’s innovative Edge AI processing to their compute vision processing systems,” says Keith Kressin, CEO of MemryX. ” By combining our expertise, we are able to offer a unique solution that delivers exceptional AI performance in small form factors for edge-based data intensive applications. Together, MemryX and Cachengo are revolutionising the way businesses can analyse, process, and store their own data.”

Benefits of Edge AI

Artificial intelligence processing of local data on the edge has several significant benefits over cloud-based AI, including:

  • Lower latency: processes data locally on the device, eliminating the need for significant data transfers.
  • Improved privacy and security: processes data locally on the device, which means that sensitive data does not need to be transmitted to the cloud, reducing the risk of data breaches and maintaining data privacy as data moves across the internet.
  • Reduced network bandwidth: processes data locally on the device, which reduces the amount of data that needs to be transmitted over the network, reducing the bandwidth required.
  • Improved reliability: continues to function even when the network connection is lost, ensuring that critical tasks are completed regardless of network availability.
  • Real-time decision-making: makes real-time decisions on the device without sending data to the cloud and waiting for a response, enabling faster decision-making.
  • Cost savings: reduces the cost of data storage and transmission, as less data needs to be sent to the cloud for processing, reducing the cost of networking, cloud computing and storage.
  • Better user experience: reduces latency and enables faster response times, improving user satisfaction and engagement.
  • Improved compliance: help organisations meet regulatory compliance requirements by keeping sensitive data on the device and reducing the risk of data breaches.
  • Scalability: scales more easily than cloud-based AI, as additional devices can be added to the network without additional cloud resources.
  • Offline operation: operates offline without requiring a network connection, which is particularly useful in environments with poor or unreliable network connectivity or no network connection available.

Edge AI opens the door to new, AI-enhanced products and services. The partnership with Cachengo allows AI developers to deploy edge AI models on a proven platform leveraging MemryX’s high-performance and low-latency AI processing with proprietary at-memory computing, parallelism and dataflow architecture.

Broad deployment and use cases

MemryX AI accelerators support all popular AI software frameworks, including PyTorch, ONNX, TensorFlow and Keras; all popular host processors, including x86, Arm and RISC-V; and all modern operating systems, such as Linux, Windows, and Android. Hundreds of AI models have been verified on MemryX silicon which is sampling now to numerous customers with production planned Q4’23.

The partnership provides developers with a high performance, scalable, and locally deployed Edge AI platform which can support a variety of workloads across sectors, including:

  • Industrial & building automation: AI-enabled industrial systems provide remote asset video monitoring, including predictive maintenance, quality control and automation monitoring with high video processing and AI demands, while AI-enabled buildings can optimise comfort and water, power, heating and lighting based on occupancy and environmental conditions.
  • Smart city infrastructure: Examples include smart traffic management systems that can analyse, reroute and predict traffic flow, public safety systems that can monitor video and alert authorities to developing public safety situations, smart environmental monitoring systems, and smart parking systems that can manage and optimise public parking.
  • Healthcare monitoring: Monitor patients in real time, processing video and sensor data from wearables and other sensing devices to detect anomalies and notify healthcare professionals when intervention is necessary.
  • Retail analytics: Analyse customer behavior in stores and generate recommendations for product placement and pricing.
  • Agriculture monitoring: Monitor crop health and identify pests and diseases, helping farmers optimise their crop yields and reduce costs.
  • Financial services: Fraud detection, credit risk analysis, and processing data in real-time to detect anomalies and identify potential risks.
  • Sports analytics: Track player movement and generate real-time insights for coaches and players.

“Our partnership with MemryX creates new opportunities for AI developers to create analytics and AI solutions, across a wide range of industries, that are highly performant and meet the power, networking, and size constraints of edge AI,” says Ash Young, co-founder and CEO of Cachengo. “We’re excited to see the innovative new solutions that are brought to market as a result of our partnership with MemryX.”

Co-founded in 2019 by Wei Lu, MemryX is working to develop an AI accelerator for edge devices. Lu, an electrical engineering and computer science professor at the University of Michigan, worked over ten years on memory-centric computing architectures and spent over three years at MemryX developing and proving the company’s approach. His expertise in memory systems and neuromorphic computing is internationally recognised.

Follow us and Comment on Twitter @TheEE_io

By continuing to use the site, you agree to the use of cookies. more information

The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.