Synopsis
While many organisations can’t afford to sustain an AI infrastructure, artificial intelligence is becoming a common skill used by companies to remain competitive. A number of IT services are available to clients in Korea and abroad through NHN Cloud Corp, a division of NHN, the top Cloud Service Provider (CSP) in Korea. AIaaS, or artificial intelligence as a service, is an addition to NHN Cloud’s current cloud services infrastructure. To do this, NHN Cloud required a brand-new supercomputer that could handle cloud-based AI for a variety of clients, including government organisations, small and medium-sized businesses, and large corporations. For AIaaS, NHN Cloud’s new 88.5 PetaFLOPS supercomputer was constructed using a 4th generation Intel Xeon processor with GPUs.
Challenge
A wide range of software applications, such as those for gaming, webcomics, music, advertising, collaboration and communication, fintech, and more, are available from NHN Cloud. Global clientele of NHN Cloud includes small and medium-sized companies, government agencies, and big companies.
AI is being used by both large and small businesses as a critical tool for both technical and strategic business operations. Because of the usefulness of AI, businesses must use the technology to remain competitive in their marketplaces, even though building AI infrastructure needs knowledge that may not fall within the purview of many businesses. The ability to provide AI training and inference as a service was made possible for NHN Cloud by this.
The addition of AIaaS is a logical step for NHN Cloud to broaden its offerings to clients globally, given its technical proficiency in managing IT operations. Building a new cloud-focused data centre was NHN Cloud’s way of supporting an AIaaS ambition. To help the regional economy, the corporation is also growing its regional data centres. But the business required new computing resources designed for SaaS, PaaS, IaaS, and AIaaS in order to offer hyperscale AIaaS for industry and government.
NHN Cloud’s AI as a Service offers a broad spectrum of machine learning and deep learning capabilities, driven by the 4th generation Intel Xeon CPU.
Solution
Customers could immediately benefit from the insights that artificial intelligence may deliver because to the route that NHN Cloud was able to offer. Numerous deep learning and machine learning functionalities have to be supported by an AIaaS platform. The infrastructure required to accommodate a variety of client projects, from small and medium-sized firms to government and huge corporate initiatives, to serve the company’s target market. A solution like that might provide more businesses and organisations with scalable AI resources.
The system has to lead in price/performance, efficiency, and performance because to the high cost of implementing and maintaining AIaaS. The business made the decision to build its new AI/HPC cloud infrastructure using 4th generation Intel Xeon processors with AI accelerator engines in order to meet these numerous demands be kept in their new data centre. Along with new power-saving features to improve price/performance in the data centre, these most recent generation Intel data centre CPUs enable the newest I/O and memory, such as PCIe5 and DDR5.
NHN Cloud provides IaaS, PaaS, and SaaS to customers worldwide.
Furthermore, the 4th generation Intel Xeon processor architecture’s integrated Intel Accelerator Engines allow NHN Cloud to provide customers with processors that have the highest number of integrated accelerators of any system on board. Intel Advanced Vector Extensions 512 (Intel AVX-512), Intel Deep Learning Boost (Intel DL Boost), enhanced deep learning performance using lower-precision (INT8) without sacrificing accuracy, and other accelerated capabilities are just a few of the tasks that the 4th generation Intel Xeon processors will help speed up processing of. Additionally, Intel Accelerator Engines expedite cryptography activities, which are crucial to many company operations due to security.
The later part of 2023 will see the deployment of the new 88.5 PetaFLOPS system.
The end result
Expecting to fulfil the high needs of its customers, NHN Cloud bases the infrastructure of the National AI Data Centre on Intel 4th Gen Xeon processors and GPUs. With respect to performance, the new supercomputer is expected to outperform the current one, which is built on an older CPU model lacking GPUs, by up to 2.5 times.
The intricacy of AI and HPC applications benefits greatly from compatibility with H100 GPU accelerators, which also offers significant performance and scale. In contrast to other comparable cloud service providers in Korea and abroad, NHN Cloud has established a cloud service that is highly competitive thanks to the introduction of the newest Intel Xeon processor technology.
Synopsis of the solution
Advanced computing capabilities built for AI workloads were necessary for NHN Cloud to provide AIaaS to consumers. For its 88.5 pF supercomputer, the Korean CSP opted for GPUs and 4th generation Intel Xeon CPUs. NHN Cloud promises that GPUs and new Intel CPUs with Intel Accelerator Engines will perform 2.5 times better than the previous version. End of 2023, customers will have access to new system.
Ingredients for the Solution
- Offering Artificial Intelligence as a Service (AIaaS) as part of an expansion of services
- Intel AVX-512 for quicker vector computation
- Increased accuracy in deep learning (DL) using Intel DL Boost while maintaining improved precision.
