Arena Physica will train its next electromagnetism AI model on Oak Ridge's AMD supercomputer
Startup Arena Physica will train the next versions of Heaviside, its AI model for electromagnetism, on the Lux AI supercomputer at Oak Ridge National Laboratory, which runs on AMD hardware, the company said in a press release.
Heaviside predicts how electromagnetic fields behave in a physical design, work that normally needs slow, expensive simulation. The first version, released in September, has 350 million parameters and was trained on 250,000 3D simulations. Arena Physica says the field matters more as chips move to stacked 3D packaging and data links get faster.
The next versions will cover radiation patterns, more complex shapes and wider frequency ranges, for uses such as 5G and 6G antennas, satellite links, car radar and chip packaging. "Lux brings physics, AI, and compute together, enabling us to push the frontier of physics foundation models and train at a far larger scale than was possible before," said chief research officer Michael Frei. The release does not say how much compute time is involved.
We hope you enjoyed this article
Consider subscribing to one of our newsletters like Silicon Brief, Industrial AI Weekly or Daily AI Brief.
Also, consider following us on social media:
More from Data Centers
Nov 12 Innodisk adds data-centre SSDs and server memory for AI workloads Nov 12 Altman tells Samsung executives AI infrastructure demand may swing in the near term Nov 12 Indiantown, Florida, moves toward a one-year data center moratorium Nov 12 Developers propose an AI 'inference point' facility in Swansea, Massachusetts Nov 12 PJM's market monitor says data-centre demand has added $29.4 billion to capacity costsIndustrial AI Weekly
The latest advancements in smart manufacturing, predictive maintenance, AI-driven quality control, supply chain & more.
Whitepaper
Tensordyne Napier: What If One Rack Could Do the Work of Nine?
Tensordyne
This Tensordyne whitepaper presents Napier, an inference-focused AI processor and rack-scale system based on the company’s TDN Math logarithmic number system. It examines infrastructure requirements for large mixture-of-experts and agentic models, compares major inference architecture approaches, and details the TDN AIP processor, TDN72 pod, TDN Link fabric, and Napier Ultra configuration. The paper reports simulation-based performance, cost, and accuracy-validation results, including Tensordyne’s projected comparison of one Napier rack with a nine-rack Nvidia Rubin plus Groq deployment; the chip is reported as taped out and in fabrication.
Read moreYou may also like
ARIA to Test Lumai Optical AI Inference Servers
Luminal raises $25 million to run AI inference on chips beyond Nvidia's
Generative Bionics Plans European Robot Training Network With AMD
OpenAI Tests Astra for Hardware and Vehicle Design
CERN and Mistral AI Plan Foundation Model for Collider Physics
Daily AI Brief: the AI news that matters, in your inbox.