Immersion cooling redefined for future ready datacentres Immersed Computing^®

The AIC24 solution with its award-winning technology, Immersed Computing®, meets the demand for high density compute on a system and server level for Crédit Agricole through the installation of multiple Asperitas modules in two datacentres. The system makes use of the natural convection driven fluid circulation within.

The Asperitas Immersed Computing^® solutions did meet the requirements to deliver a 300kW facility on less than 35m² with ten systems. In order to do this a unique server configuration had to be developed by Asperitas for a GPU-based cluster. This Immersed Computing^® optimised configuration is able to run 12 GPUs in a 1U server; totaling up to 288 GPUs per system. For the system to run all year round a simple dry-cooler system was deployed. Optionally, the high temperature cooling water is ready to be fed into a district heating system.

“An enhanced performance, no peak heat slowdowns and reliability are the three absolute essentials for Peasoup cloud infrastructure services, and this was fully recognised and mitigated by Asperitas. Also, by adding our cloud services to the coolDC solutions portfolio and choosing to work with Asperitas and Boston, we can deliver an cost-efficient service with superb hardware performance and reliability.”

The Department for Applied Bioinformatics in conjunction with the LOEWE Centre for Translational Biodiversity Genomics (LOEWE-TBG) was looking for a solution, which is highly energy efficient, sustainable, reliable and maintainable. It should support our existing IT infrastructure by fulfilling ambitious HPC requirements for comparative genomics research across hundreds and even thousands of species. The newly acquired solution forms the central computing infrastructure of the recently founded LOEWE-TBG.

This project has been delivered in partnership with Boston.

The solution facilitates an energy efficient, scalable and future ready High Performance Computing cluster. Easily integrates into datacentre facilities and requires minimal overheads while also complementing infrastructure monitoring platforms, allowing for real-time management by the service providers.