Energy-efficient Ai Tech At Sc23

By Author: Arina Smith
Total Articles: 185
Comment this article

Untether AI® is a leader in energy-centric AI acceleration technology, and they will be participating and presenting at the Super Computing Conference scheduled for November 13-17, 2023. The company's focus is on addressing the data movement bottleneck that affects traditional CPUs and GPUs, leading to increased energy consumption and reduced performance. Their solution involves an at-memory compute architecture that allows for high-performance, low-latency neural network inference acceleration while maintaining accuracy.

Key highlights of Untether AI's participation at the conference include:

1. At-Memory Compute Architecture: Untether AI's innovative at-memory compute architecture aims to solve the data movement bottleneck issue. By placing compute elements directly next to the memory array, they reduce power consumption and increase throughput through parallelism and improved memory bandwidth.

2. AI Video Analysis Reference Design: At booth #2250, Untether AI will showcase an accelerated AI video analysis reference design. This full-stack solution runs a multi-stream instance segmentation neural network, ...
... demonstrating the practical application of their technology.

3. Presentation by CEO Arun Iyengar: Arun Iyengar, the CEO of Untether AI, will deliver a presentation on "Overcoming the Cost of Data Movement in AI Inference Accelerators" at SC23 on November 15 at 11:00 a.m. This presentation is expected to shed light on the challenges of data movement in AI and how Untether AI's approach addresses these issues.

4. Efficiency for AI Workloads: Traditional processor architectures face challenges in meeting the increasing demands of AI workloads. Untether AI's energy-centric acceleration approach offers improved efficiency, making it suitable for various AI tasks, including vision, natural language processing, and recommendation engines.

In summary, Untether AI's participation at the Super Computing Conference revolves around their groundbreaking energy-centric AI acceleration technology, which aims to revolutionize AI inference acceleration by mitigating the data movement bottleneck and enhancing energy efficiency. This approach has the potential to significantly impact various AI applications, making them more energy-efficient and high-performing.

https://www.techdogs.com/tech-news/business-wire/energy-centric-ai-acceleration-technology-to-be-showcased-at-super-computing-2023