ClimaCell Delivers Innovative Weather Prediction

Intel® Xeon® Scalable processors and Google Cloud C2 nodes provide performance, flexibility, and AI-readiness for ClimaCell.

At a Glance:

  • ClimaCell offers people and organizations the best weather insights through big data collection and exclusive analysis and modeling approaches.

  • ClimaCell selected Google Cloud’s Compute Engine C2 instances, powered by 2nd Generation Intel® Xeon® processors, with AI acceleration built-in. The solution offers the scalability, performance, and cost-effectiveness needed to process and interpret the broad mix of data.

BUILT IN - ARTICLE INTRO SECOND COMPONENT

Executive Summary
ClimaCell disrupts today’s weather prediction industry with all-new tools and an Application Programming Interface (API) for high precision micro-forecasting. That information helps ClimaCell’s clients make more informed operational decisions to increase efficiency, reduce costs, and minimize employee safety risks. The global API platform runs on Google Cloud Compute Optimized instances (C2) featur­ing 2nd Generation Intel Xeon Scalable processors to deliver a high performance computing (HPC) solution. The combination provides ClimaCell’s unique business model a fast and scalable solution to accommodate massive datasets and artificial intelligence (AI) assisted modeling.

Challenge
In the past, accurate weather prediction proved a daunting task. ClimaCell’s gov­ernment and enterprise clients needed better ways to understand current and future weather patterns. That actionable information helps clients mitigate busi­ness and safety consequences caused by severe weather conditions. ClimaCell’s micro-forecasts gather weather data from government sources, Internet of Things (IoT) devices, connected cars, wireless signals, drones, and other proprietary data. Making sense of the enormous data volumes, and forming accurate predictive models using it, requires HPC systems that can accommodate heavy workloads.

Solution
By incorporating AI into their detailed analysis process, ClimaCell can maximize its predictive services to produce a holistic and highly accurate model of impending weather conditions. After rigorous proof-of-concept testing, ClimaCell selected Google Cloud’s Compute Engine C2 instances, powered by 2nd Generation Intel Xeon processors with AI acceleration built-in. The infrastructure solution offers the scalability, performance, and cost-effectiveness needed to process and interpret the broad mix of data.

Driven by 2nd Generation Intel Scalable Processors, CPU-only C2 machine types offer up to 3.8 GHz of sustained all-core turbo boost. The instances also provide transparency into the architecture of the underlying server platforms, letting users fine-tune the performance. Compared with N1 high-CPU machine types, C2 machine types offer much more computing power, run on a newer platform, and are generally more robust for compute-intensive workloads. The C2 instances also scale globally as needed.

Customers can make reservations to ensure their applications have the capacity required for scale. Together, these features simplify integration and help users derive more value from their hosted environment.

In addition, ClimaCell used Intel® compilers and Intel® MPI with their code to help boost weather forecasting application performance on C2 instances. The Intel® software helps maxi­mize the code’s performance by utilizing the latest hardware instructions (such as Intel® Advanced Vector Extensions 512 (Intel® AVX-512)), optimally use all the cores and memory footprint, and help the code scale without degradation among multiple C2 nodes.

Results
To find the instance that would provide the best micro-forecast performance, ClimaCell performed internal bench­marking of the proof-of-concept solution, comparing a Google Cloud C2 instance with previous-generation N1 clus­ters. With the C2 instances, ClimaCell achieved 40% better price/performance.1, 2 than the N1 instances. The Google Cloud Compute Engine C2 nodes also address ClimaCell’s unique needs for elasticity. The cloud solution can spin up additional nodes whenever workloads require it.

Advanced Cloud-Based Weather Modeling Enables Breakthroughs for Multiple Industries
With real-time insights, ClimaCell’s enterprise customers derive value from the service in multiple ways:

  • A major airline taps ClimaCell data to increase passenger safety and minimize operational costs. With ClimaCell solutions, the airline can determine in advance when extreme weather conditions pose a safety risk or impact flight times. In the winter, micro-predictions can also inform staff if their planes require de-icing before takeoff. By providing insights like these, the airline claims that ClimaCell’s weather prediction technology saves the company $50,000 per month per airport hub.
  • A large investment trust owns power transmission infrastructure across India. In 2020, a cyclone posed an unprecedented safety threat to the region. With the assistance of ClimaCell technology, the company found it was able to predict the path of the cyclone and determine which of its hundreds of locations were at risk. ClimaCell solutions identified the two vulnerable sites, allowing their client to prioritize their attention on the safety and service at those specific locations.
  • A professional football team in the United States depends on ClimaCell to mitigate operational challenges. With insights about the approaching weather, managers can determine the best times to water or paint the football field. They can also plan for snow, or halt operations when lightning storms approach. The team asserts that the combination of benefits reduces the cost of stadium upkeep and reduces risk to staff.
  • ClimaCell.org solutions offer weather prediction capability to the 5 billion people living in regions where traditional weather forecasting technologies do not reach. In parts of Africa, for example, ClimaCell solutions help farmers determine the ideal time to plant crops. With foreknowledge of coming weather, farmers can lessen the cost of field maintenance and maximize crop yields.
  • A class-A railroad uses ClimaCell technology to predict weather patterns that could impact train travel. In one case, the railroad found the weather model predicted 90 mile-per-hour winds along their route of travel. By delaying the trip, the train avoided the possibility of a dangerous and costly derailment.

Key Takeaways

  • ClimaCell’s micro-forecasts derive from a mix of proprietary and publicly available weather data that is combined and processed using HPC-driven AI, simulation, and modeling workloads.
  • With the underlying support of 2nd Generation Intel Xeon Scalable processors and Intel® software, Google Cloud C2 nodes provide a fast, highly elastic, scalable, and AI-ready solution.
  • Google Cloud’s Intel®-based HPC solution provides ideal price-performance for ClimaCell’s unique business service.

Table 1: Google Cloud’s Compute Optimized (C2) HPC Instances3, 4, 5, 6

C2 machine types have the following restrictions:

  • You cannot use regional persistent disks with compute-optimized machine types.
  • C2 machine types are subject to different disk limits than general-purpose and memory-optimized machine types.
  • C2 machine types are only available on select CPU platforms.
  • Caution: C2 machine types do not support GPUs.

Intel® Xeon® Scalable processors are architected to deliver performance leadership across a broad range of demanding workloads. 2nd Generation Intel® Xeon® Scal­able processors are the only data center CPU with AI acceleration built in. The C2 instance with the 2nd Gen Intel Xeon Scalable processor scales up to 60 vCPUs with exceptional memory bandwidth per vCPU.

Spotlight on ClimaCell
ClimaCell offers people and organizations the best weather insights through big data collection and exclusive analysis and modeling approaches. By providing its clients with highly accurate micro-forecasts, ClimaCell solutions offer invaluable business insights. ClimaCell’s patented MicroWeather technology engine gathers Weather of Things data from wireless signals, connected cars, drones, and other devices. By combining that data with proprietary AI-driven models, insights from ClimaCell solutions improve safety, operational efficiency, and profitability.

Spotlight on Google Cloud
Google Cloud solutions help customers solve their biggest business challenges through outstanding technology and a toolset engineered for ease of use. As a leader in cloud services, Google Cloud provides customers a holistic solution for data security, intelligent analytics, and hybrid or multi-cloud solutions. Today, customers in over 150 countries trust Google Cloud to modernize their computing environment to bring ideas, insights, and people together.

Learn More

Explore Related Products and Solutions

Intel® Xeon® Scalable Processors

Drive actionable insight, count on hardware-based security, and deploy dynamic service delivery with Intel® Xeon® Scalable processors.

Learn more

Avisos y exenciones de responsabilidad

Las características y ventajas de las tecnologías Intel® dependen de la configuración del sistema y es posible que necesiten hardware y software habilitados, o la activación del servicio. El desempeño varía según la configuración del sistema. Ningún sistema informático puede proporcionar una seguridad absoluta. Consulte con el vendedor o el fabricante del sistema, o visite https://www.intel.com para obtener más información. // Es posible que el software y las cargas de trabajo utilizados en las pruebas de desempeño solo se hayan optimizado para desempeño en los microprocesadores Intel. Las pruebas de desempeño, como SYSmark y MobileMark, se miden utilizando sistemas de computación, componentes, software, operaciones y funciones específicos. Cualquier cambio en alguno de esos factores podría generar un cambio en los resultados. Debe consultar otra información y pruebas de desempeño que lo ayuden a evaluar plenamente las compras consideradas, incluido el desempeño de ese producto al combinarse con otros. Para obtener más información, visite https://www.intel.com/benchmarks. Los resultados de desempeño se basan en pruebas realizadas en la fecha indicada en la configuración y es posible que no reflejen todas las actualizaciones de seguridad que están a disposición del público. Consulte la divulgación de configuración para obtener más información. Ningún producto o componente puede proporcionar una seguridad absoluta. Los escenarios de reducción de costo publicados se utilizan como ejemplo de cómo un producto equipado con Intel® dado, en circunstancias y configuraciones específicas, puede afectar los costos futuros y proporcionar recortes en cuanto a costo. Las circunstancias pueden variar. Intel no garantiza costos o reducciones de costos, // Intel no ejerce control ni inspección algunos sobre los datos de análisis de desempeño o los sitios web de terceros a los que se hace referencia en este documento. Debe visitar el sitio web referido y confirmar si los datos a los que se hacen referencia son precisos. // En algunos casos de prueba, los resultados estimados o simulados con modelado o simulación de arquitectura o análisis internos de Intel con fines informativos. Cualquier diferencia en el hardware, software o configuración del sistema puede afectar el desempeño real.

Información sobre productos y desempeño

1“The new Compute-Optimized VMs offer a greater than 40% perfor­mance improvement compared to current GCP VMs.” Bart Sano, VP of Platforms, Google Cloud. Learn more here.
2From Google Cloud’s CPU platforms page: Learn more here. C2 instance all-core turbo CPU frequency is 3.8 GHz, with the lowest first-generation N1 machine type base frequency being 2.0 GHz. Note, all-core turbo not available for first generation N1 instances. Improve­ment from 2.0GHz to 3.8GHz=40%. Also see https://cloud.google.com/blog/products/compute/expanding-virtual-machine-types-to-drive-performance-and-efficiency
4A vCPU is implemented as a single hardware Hyper-thread on one of the available CPU platforms.
5Persistent disk usage is charged separately from machine type pricing.
6Network egress bandwidth is up to the specified limit. Actual performance depends on factors such as network congestion or protocol overhead.