Untether AI Partners with Colfax International to Provide Peak Performance in AI Edge Servers

Untether AI, the leader in at-memory computation and AI acceleration, today announced a partnership with Colfax International in which Untether's tsunAImi^TM accelerator cards, powered by the runAI^TM devices, are now available in the Colfax AIEdgetm servers. Untether AI's tsunAImi accelerator cards help companies running AI workloads to run neural networks faster, cooler, and more cost-effectively using at-memory computing. The cards, along with Untether AI's imAIgine Software Development Kit (SDK), gives AI developers an automated path to running networks at high performance. This frees them from having to perform low-level optimization tasks and instead spend time where it matters to them ? crafting their models.

"Untether AI's innovative approach, utilizing at-memory computation, will enable Colfax to offer unmatched compute density to help customers achieve the lowest TCO for inference workloads. We are excited to begin our journey with Untether AI," said Gautam Shah, CEO of Colfax.

Artificial Intelligence (AI) workloads for inference require increasing amounts of compute resources, far outstripping the gains available to traditional CPU and GPU architectures. Untether AI radically rethought how computation for machine learning is accomplished, by focusing on the needs for inference acceleration and maximizing power efficiency. Using at-memory computation, Untether AI breaks through the barriers of traditional von Neumann architectures, offering industry-leading, cost-effective compute density with power efficiency.

Colfax provides advanced computing solutions for those at the forefront of innovation and discovery and partners with some of the biggest brand names in the industry. The combination of Untether AI's tsunAImi acceleration card and Colfax servers provides the industry's highest compute density for AI inference workloads.

"Over the course of several decades, Colfax has earned a reputation for solving the most complex problems and securing outstanding results for enterprises, national labs, universities, public agencies and start-ups. We are proud to have Colfax as a partner," stated Bob Beachler, VP of Product, Untether AI.

Untether AI tsunAImi card

With the industry's highest compute density and best performance-per-watt, Untether AI tsunAImi cards enable Colfax to custom-configure servers for inference workloads that can reduce their capital and operational expenditures by up to 75%. tsunAImi accelerator cards are powered by four runAI200 devices, providing 2 PetaOperations of compute, more than twice other available card. This compute power translates into more than 80,000 frames per second of ResNet-50 v 1.5 throughput at batch=1, three times the throughput of its nearest competitor. For natural language processing, tsunAImi accelerator cards can process more than 12,000 queries per second (qps) of BERT-base.

The imAIgine Software Development Kit

The Untether AI imAIgine Software Development Kit (SDK) provides an automated path to running high-performance networks, thanks to push-button quantization, optimization, physical allocation, and multi-chip partitioning. The imAIgine SDK also provides an extensive visualization toolkit, cycle-accurate simulator, and an easily integrated runtime API.

For more information on the Colfax AI Edge Servers and availability visit Colfax International.

About Untether AI

Untether AI provides ultra-efficient, high-performance AI chips to enable new frontiers in AI applications. By combining the power efficiency of at-memory computation with the robustness of digital processing, Untether AI has developed a groundbreaking new chip architecture for neural net inference that eliminates the data movement bottleneck that costs energy and performance in traditional architectures. Founded in Toronto in 2018, Untether AI is funded by Radical Ventures and Intel Capital. www.untether.ai.

All reference to Untether AI trademarks are the property of Untether.AI. All other trademarks mentioned herein are the property of their respective owners.