Descartes Labs Achieves #41 in TOP500 With Cloud-Based Supercomputing Demonstration Powered by AWS, Signaling New Era for Geospatial Data Analysis at Scale

Descartes Labs uses Amazon Website Expert services, Inc. (AWS) to run a 9.95 petaflops Large-Effectiveness LINPACK (HPL) benchmark, putting it #41 in the June 2021 Best500 position. The business enhances on its preceding AWS-based mostly 2019 Prime500 submission (1.926 petaflops, #136) by 417{46dd52bca0123ad67b2d1222819e83fd0a56e45ca5068239f05f0c514f1e20f9} in HPL efficiency and 95 ranking places in only two a long time.

Push Launch



current: Jun 28, 2021 10:31 MDT


Descartes Labs, Inc. now introduced a new cloud supercomputing achievement with a Top rated500 operate on AWS of 9.95 petaflops making use of virtualized Amazon Elastic Compute Cloud (Amazon EC2) cases. The Top rated500 corporation makes use of the LINPACK Benchmark—a examination that consists of fixing a dense method of linear equations—to rank the overall performance of the 500 most strong commercially readily available computer programs. Descartes Labs enhanced its prior cloud report of 1.926 petaflops established in its 2019 Best500 submission (#136 rating) with a 417{46dd52bca0123ad67b2d1222819e83fd0a56e45ca5068239f05f0c514f1e20f9} HPL overall performance improvement in excess of two decades, and improved effectiveness by a variable of 10 relative to the overall performance gains achieved by classic, on-premises supercomputers in the similar two-calendar year period of time. For 2021, the general computation took around nine hours and done 3 x 1020 complete floating stage operations across 172,692 Intel Xeon Scalable Processor cores.

The quantity of world wide sensor facts processed by Descartes Labs requires cutting-edge cloud technologies capable of operating High Performance Computing (HPC) and Device Studying (ML) workflows at scale. This lets the Descartes Labs engineering groups to compute, examine, and retailer around authentic-time information in get to provide mission-significant organization analytics at a world wide scale to buyers in the agriculture, mining, protection and intelligence, and consumer packaged merchandise (CPG) industries. This capability is getting more and more critical for applications that rely on a combination of geospatial analytics on a world wide level, this kind of as environmental monitoring, extreme temperature monitoring and forecasting, multimodal data fusion, and weather improve assessment. An example of this mission is the 6 petabytes of Sentinel-1 radar details lately processed by Descartes Labs to enable track deforestation at continental scale, realize where wildfire possibility is optimum, and monitor regenerative agriculture techniques.

“Obtaining the extensive computational potential supplied by AWS, linked to great amounts of storage which is obtainable at pretty large speeds, lets you do matters that you just could not do prior to in phrases of facts investigation and knowing the fact of our world or the universe as a full,” explained Mike Warren, Descartes Labs Cofounder and CTO. “Our cloud utilization nowadays spans the selection of evaluation of earth observation datasets at the petascale, specially relocating beyond just optical imagery into far more esoteric kinds of earth observation such as radar, InSAR, and AIS details.”

In collaboration with the AWS HPC team, Descartes was in a position to spin up the means desired for its Top500 run making use of basic-purpose, publicly out there Amazon EC2 instances across C5/M5/R5 households. Thanks to the scale and elasticity of the AWS Cloud, Descartes Labs was equipped to scale up infrastructure for this take a look at without having influencing its creation workflows. Also, due to the elasticity of Amazon EC2’s compute resources, Descartes was also ready to exam out distinct cluster measurements with numerous HPL operates spanning 1 petaflops to 8.5 petaflops, major up to the remaining 9.95 petaflops final result.

“The reality that Descartes Labs was in a position to use Amazon EC2 to develop a 9.95 petaflops supercomputer rating #41 on the Top rated500 list is a truly impressive accomplishment,” said Dave Brown, Vice President, Amazon EC2, AWS. “AWS is proud to assistance Descartes Labs in their mission to create planetary scale geospatial insights for their clients. We hope the desire for instant data and insights to increase into the future, and this is a potent demonstration of the worth that AWS can bring to customers and partners on any scale.”

With petabytes of earth observation and floor sensor details transferring at the speed of collection, it really is now the information that is driving the computational requirements, not the other way about. The architecture of a traditional supercomputer is frequently restricted in terms of cluster size and enter/output (I/O). Obtaining versatile computational ability, linked to remarkable quantities of storage in the cloud, all available at pretty superior speeds, lets Descartes Labs have out science that the entire world could not complete before.

The LINPACK benchmark used in this supercomputing demonstration will help to approximate how quickly a computer will carry out when resolving “tightly coupled” authentic-entire world troubles, these as weather conditions forecasting. By making units that can operate the LINPACK benchmark at superior functionality, Descartes Labs can use the very same units to design the climate rapidly, properly, and more than multiple places simultaneously to enhance agricultural yield forecasts.

These types of reducing-edge science has been instrumental in Descartes Labs’ collaboration with clients like CJ OliveNetworks, an group that a short while ago migrated all of CJ Group Companies’ IT units to AWS. The enterprise leveraged Descartes Labs engineering to have out a buying innovation project to forecast sugar input costs for CJ CheilJedang, which imports 600,000 tons of raw sugar for each 12 months. CJ has reached prosperous enterprise innovation from this undertaking by satellite imaging know-how, local weather, and vegetation details evaluation.

States CEO of OliveNetworks and Group Chief Digital Officer In hyok Cha, “I desire to congratulate equally Descartes Labs and AWS on achieving these a significant scale computation to help the infrastructure do the job behind planetary sustainability initiatives, and also want to go on to collaborate with the two businesses to establish additional revolutionary earth observation use scenarios and company programs.”

The changeover from on-premises to cloud computing mirrors the transition from just examining an organization’s inside information to incorporating massive volumes of geospatial and sensor data from the world past. Planetary scale data requires planetary scale compute. Appropriately, the dimensions of this “outside the house the firewall” info requires the variety of cloud infrastructure Descartes Labs has cautiously formulated and utilized for this most current Prime500 operate. The company appears to be ahead to serving to others do the exact same.

“The cloud is where by all the newest technologies is showing up first. You happen to be looking at all the most up-to-date processors from Intel and AMD in the cloud prior to you have entry to them in your on-prem supercomputers,” reported Warren. “The financial investment in the networking technology to aid these hyperscale apps in the cloud is turning out to be fairly helpful for common tightly coupled supercomputing purposes as well. Now, the large push is into exascale in HPC. But at the identical time, equipment understanding and deep neural networks are creating big progress in other types of challenges. I feel all those worlds may well arrive with each other, or they could drift additional apart. And that is one of the large inquiries for the subsequent ten years.”

Visit descarteslabs.com or learn more about our 2019 Best500 accomplishment.

About Descartes Labs

Descartes Labs is a geospatial intelligence organization that performs scientific analysis of geospatial, distant sensing, and assorted complementary knowledge sets to help sustainable sourcing best tactics, commodity value forecasting, and efficient mineral exploration for primary CPG, Agriculture, and Mining businesses. Our SaaS platform automates the examination of geospatial imagery for our users, enabling planetary scale assessment by synthetic intelligence and machine understanding. The firm also supports a diverse set of federal governing administration endeavours to curate, examine, and deliver exclusive actionable insights from geospatial details. For additional details, visit www.descarteslabs.com or abide by us on Twitter or LinkedIn.

Push Make contact with
Alex Diamond
Descartes Labs
[email protected]
303-877-2721

Resource: Descartes Labs