Search
  • Christian LaRosa

These are our cloud Supercomputers, Gaudi and Cortex

Updated: Jul 25

Supporting a new breed of design and applied artificial intelligence.


PHOENIX, AZ, USA - JULY 18, 2020 - Rosotics, Inc. (www.rosotics.com), a Phoenix-based industrial robotics corporation, details today our new cloud-based supercomputers constructed in order to power a new landscape of design and manufacturing for clients, In the early 20th century, a Spanish architect by the name of Antoni Gaudí i Cornet sought a new language of design, a language modeled after nature. He believed nature was the best teacher, that it forms the best possible solutions you can hope to find, and we share that belief too. We named our design workhorse, the Gaudi Supercomputer in his honor. Our other cloud supercomputer, used for complex deep learning applications and operational AI at scale, is named Cortex, as a reference to the human brain's Cerebral Cortex, the segment of the brain responsible for higher thought and consciousness.



Each Supercomputer is a network of computational nodes constructed physically and accessed through the cloud, from any device. Using a long-time institutional relationship with Arizona State University, a relationship that has granted us a great deal of resources to deploy on the AWS platform, we have committed resources to the pursuit of greater computational power under our own wing. Our engineers have architected two networks of parallel AWS instances that each work as a whole as common entities. The Gaudi Supercomputer is built on Rosotics dedicated hardware at a Boardman, Oregon data center, within the United States. The Cortex Supercomputer is also built on Rosotics dedicated hardware at an Umatilla, Oregon data center, within the United States. Each individual node of the Gaudi Supercomputer comprises a number of EC2 z1d.metal instances, each instance comprising 48 vCPUs at an all-core clock of 4 GHz, the fastest computational speed available in any of Amazon's data centers. Each instance contains 384 GB of RAM and NVME SSDs, which when combined with the other instances and in turn nodes that together form Gaudi, provide the horsepower to crunch design workloads that would be otherwise impossible to compute on a workstation or in-house server. Gaudi provides razor-sharp computing accuracy and groundbreaking reductions in the computation time of generative design workloads - allowing us to deploy these workloads at previously unrealized scale. Each individual node of the Cortex supercomputer comprises a number of EC2 p3dn.24xlarge instances, each instance comprising 96 vCPUs, 768 GB of RAM, and 8X Tesla V100 GPUs (for a total of 256 GB GPU memory). The Tesla V100 GPU is the most powerful data center GPU ever devised and currently available for service. Each node instance is architected within Cortex to house a group of Committee of Machines (CoMs), a type of modular neural network that is able to produce significantly more refined results than a Convolutional Neural Network in applied artificial intelligence. The combination of each instance and in turn node into the greater Cortex Supercomputer allows Rosotics to remotely perform extremely complex, rapid, accurate, and proprietary deep learning algorithms on supplied visual information. Rosotics applies Cortex for the autonomous operability of its mobilized manufacturing workforce.


For more information on Rosotics, please visit www.rosotics.com.


About Rosotics, Inc.

Born in the valley of the sun, Rosotics is a supercharged design house and robotic manufacturing corporation that works to harness next-generation technological advances to conceive and create objects faster and more effectively than ever thought possible before. Unlike engineering organizations who design using dated methods and produce using obsolete technology, Rosotics has both deployed and developed breakthrough technologies which leverage principles of bionics, and are far more capable than any that have come before them.


//


Press Contact

Austin Thurman

Project Manager

austin@rosotics.com

Visit our Press Page

Resources

Tech

Company

Rosotics 2 Trans 2.png

© 2020 ROSOTICS

Supporter Badge rectangle.png