embedded world | Optimium

Hall 5 / Booth Number 5-473

Optimium

Key Facts

  • Optimium has been developed based on our proprietary programming language, Nadya. Through metaprogramming, Nadya generates optimized code for target hardware based on countless combinations of layer-specific optimization parameters. Additionally, it utilizes tens of unique optimization passes to compile the generated code into executable forms, maximizing inference speed on the target hardware.
  • Optimium can maximize AI model inference on target hardware while maintaining accuracy. Based on our proprietary optimization techniques such as model tuning, layer fusion, and memory planning, it has demonstrated superior performance over existing engines.
  • Optimium enables convenient deployment across various hardware platforms with a single tool. In the case of existing engines, different engines should be used for each target hardware, which hinders swift AI model deployment. With Optimium, substantial time and costs associated with AI model deployment can be effectively reduced.

Categories

  • Embedded Vision
  • Artificial Intelligence
  • Other Tools and Software

Key Facts

  • Optimium has been developed based on our proprietary programming language, Nadya. Through metaprogramming, Nadya generates optimized code for target hardware based on countless combinations of layer-specific optimization parameters. Additionally, it utilizes tens of unique optimization passes to compile the generated code into executable forms, maximizing inference speed on the target hardware.
  • Optimium can maximize AI model inference on target hardware while maintaining accuracy. Based on our proprietary optimization techniques such as model tuning, layer fusion, and memory planning, it has demonstrated superior performance over existing engines.
  • Optimium enables convenient deployment across various hardware platforms with a single tool. In the case of existing engines, different engines should be used for each target hardware, which hinders swift AI model deployment. With Optimium, substantial time and costs associated with AI model deployment can be effectively reduced.

Categories

  • Embedded Vision
  • Artificial Intelligence
  • Other Tools and Software
Show More

Product information

Optimium is a next-generation AI inference engine that overcomes functional limitations of existing engines and enables convenient deployment of an AI model with the highest performance. It not only improves inference speed of the AI model on target hardware without sacrificing accuracy, but also facilitates convenient deployment across various hardware using a single tool.

Based on our proprietary optimization techniques, Optimium has demonstrated superior performance over existing engines. AI models deployed using Optimium achieved much faster inference speeds than those deployed using existing engines on various hardware platforms such as Arm, Intel, and AMD. In fact, it is the fastest inference engine for deploying computer vision models on CPUs. Since it performs hardware-aware inference optimization tailored to the target device, Optimium is the ideal tool for implementing high-performing and power-efficient AI solutions for resource-constrained embedded systems.

... read more

Watch our product video

blue graphical background