Artificial Intelligence (AI) is being applied to most aspects of daily life nowadays. Its use will only increase with the introduction of autonomous vehicles, robots, drones, etc., requiring more robust and memory-efficient AI applications. But general-purpose AI frameworks are heavy and slow to apply to production. Korean startup SOYNET has developed a general-purpose software accelerator designed to maximize GPU utilization using GPU acceleration.
SOYNET’s core technology can provide execution frameworks and development/consultation to companies that want to provide AI services, so they can easily enter the market. SoyNet’s high-speed inference engine is designed to bring speed 3x faster and 80% lighter to the AI Inference and help reduce time to market for AI.
Faster and affordable solution for AI
SOYNET can run any artificial intelligence models trained on Tensorflow/PyTorch/Caffe etc; making it faster consuming less memory. It has differentiated its technology to support Time to Market by shortening the development period of optimization/acceleration technology, which is required throughout the cycle from domain selection, data collection, model development and training services. SoyNet reduces the server cost required for service by accelerating the execution of AI models used by customers.
SOYNET solutions optimizes the processing speed by maximizing the usage rate of GPUs and minimizing the memory used. SOYNET is a general-purpose inference-only acceleration solution for the service of Artificial Intelligence models and was developed to provide faster inference execution in real-time production situations thereby, replacing existing artificial intelligence frameworks such as TensorFlow, PyTorch, and Caffe.
Its main products are SOYNET (GPU-based artificial intelligence deep learning processing acceleration technology), SoyFire, SoyLPR, and SoyEdge (artificial intelligence-based object detection and cognitive technology). The start-up business model serves as an annual support or premium customized plan. SOYNET introduced its AI execution accelerator at the CES 2022 in Las Vegas after being selected in the ‘Top 10 Korean Products of CES 2020.’ SOYNET’s technology is being used by several companies, government agencies and manufacturers in South Korea. The company is pacing to step into the global market and therefore exploring opportunities for collaboration and investments.
Also Read,
- RoboTry makes robot building for kids fun & affordable with digital twin technology
- The app of a Korean startup Petnow can help lost dogs reunite with their families using AI biometric recognition
- Beyond Honeycomb’s AI-Driven Robot can reproduce world-class chef meals for everyone
- Korean startup’s design platform Interior Teacher can make your home-styling dreams come true
