Video Intelligence Platform, VisualCortex, has partnered with Firmus, builders of Supercloud, the hyper energy-efficient computational cloud to reduce computer vision emissions by 50 percent. The partnership will make large-scale video analytics environmentally sustainable and commercially achievable.
Lower carbon footprint while model training
Supercloud, powered by Firmus’ immersion-cooled data center, enables VisualCortex and its clients to perform compute-exhaustive machine learning workloads while future-proofing its carbon footprint. The 50 percent reduction in CO2 emissions is achieved via Firmus’ fluid-based cooling system (a biodegradable, non-conductive liquid), which is 1000 times more efficient at removing heat from data centers compared to traditional air-based methods of heat exchange.
The partnership delivers three core outcomes:
- A dedicated VisualCortex R&D environment on Supercloud, enabling the vendor to train more computer vision models, in less time, while producing fewer CO2 emissions.
- A referral agreement, enabling VisualCortex to offer Supercloud’s hosting services to its current and future clients at discounted rates.
- A second referral agreement allowing Firmus to recommend VisualCortex’s Video Intelligence Platform to partners and clients at discounted rates.
Patrick Elliott, CEO and co-founder of VisualCortex said;
« At VisualCortex, our mission is to make video data actionable throughout the enterprise. Our Video Intelligence platform makes it technologically and financially possible to pervasively scale computer vision programs. At the same time, we fully acknowledge that computer vision requires significant GPU processing power. Supercloud, powered by Firmus, provides an ideal solution to address this challenge for VisualCortex and our clients. »
A recent paper from Cornell University highlighted that training artificial intelligence models in traditional data center environments can emit over 284,000 kilograms of carbon dioxide over its lifetime. For context, that’s almost five times greater than that produced by an average American car over its lifetime. Supercloud’s immersion-cooled approach will enable VisualCortex to reduce emissions produced from model training by around 50 percent.
Supercloud delivers an alternative cloud platform, ideal for facilitating the high-performance and resource-intensive machine learning workloads that underpin visual computing applications.