Supercomputers Archives - Gizmochina

Nvidia Unveils Eos Supercomputer to Push the Boundaries of Artificial Intelligence

Anubhav — Fri, 16 Feb 2024 02:44:41 +0000

Nvidia has taken the wraps off Eos, its groundbreaking data center-scale supercomputer, at the Supercomputing 2023 trade show. Dubbed as an “AI factory,” Eos is designed to push the boundaries of artificial intelligence development. Named after the Greek goddess of dawn, this supercomputer symbolizes a new era in AI acceleration.

The supercomputer can achieve 18.4 exaflops of AI performance

Eos is powered by an impressive setup of 576 Nvidia DGX H100 systems, integrated with Quantum-2 InfiniBand networking and specialized software, achieving a staggering 18.4 exaflops of FP8 AI performance. This setup marks an evolution from Nvidia’s previous supercomputing projects, SaturnV and Selene, showcasing an advanced DGX SuperPOD architecture. This design enables rapid scaling of AI data center solutions to meet high-performance demands.

At the heart of Eos are 4,608 H100 GPUs, spread across each DGX H100 system’s eight H100 Tensor Core APUs. This hardware configuration is tailored to manage extensive workloads, such as training large language models, running AI recommenders, conducting large-scale analytics, performing quantum simulations, and more.

Nvidia emphasizes that Eos’s architecture is finely tuned for AI tasks, requiring ultra-low latency and high throughput in massive computing clusters. The supercomputer’s networking capabilities, reaching speeds up to 400GB/s, are critical for handling large datasets necessary for training AI models.

Eos also integrates specialized software to enhance AI development and deployment. Base Command facilitates AI workflow, cluster management, and provides libraries for compute, storage, and network acceleration. AI Enterprise, a cloud-native platform, aims to speed up AI application development, positioning itself as the “operating system” for enterprise-level AI. In recognition of its capabilities, Eos secured the ninth position on the TOP500 list of the world’s fastest supercomputers.

RELATED:

(VIA)

The post Nvidia Unveils Eos Supercomputer to Push the Boundaries of Artificial Intelligence appeared first on Gizmochina.

Beijing Unveils Powerful AI Computing Platform Amid Semiconductor Tensions

Anubhav — Wed, 27 Dec 2023 14:28:52 +0000

Recent reports have stated that Beijing has introduced a state-of-the-art computing platform. This initiative, known as the Beijing AI Public Computing Platform or the Shangzhuang project, comes at a crucial time when global semiconductor tensions, especially with the US, are intensifying.

The project intends to resolve a shortage of computing power in Beijing

Operated by Beijing Energy Holding (BEH), the Shangzhuang project is designed to resolve a critical shortage of computing power in Beijing. This shortage has been particularly challenging for universities, research centers, and small and medium-sized businesses. The new platform is not just a technical advancement but also a strategic step in China’s endeavor to become a leader in AI technology.

BEH’s computing platform boasts an impressive 500 petaflops of power in its initial phase. To put this in perspective, the world’s most powerful supercomputer, the US’s Frontier, has a peak performance of 1,194 petaflops. BEH isn’t stopping there, though. They plan to triple this capacity to 1,500 petaflops in the near future.

Another noteworthy aspect of the Shangzhuang project is its commitment to sustainability. BEH has expressed intentions to eventually power the platform entirely with green energy, aligning with global environmental goals.

The platform’s inauguration was marked by significant collaborations. BEH signed agreements with major Chinese tech firms, including Alibaba Cloud and AI startup Zhipu AI. These partnerships aim to explore various sectors, from green energy to digital economy development.

This initiative is part of a broader strategy by Beijing’s municipal government. Earlier this year, the government announced plans to pool computing resources to bolster AI development. Beijing, already hosting a third of China‘s core AI startups, is rapidly solidifying its status as a hub for AI innovation.

RELATED:

(Via)

The post Beijing Unveils Powerful AI Computing Platform Amid Semiconductor Tensions appeared first on Gizmochina.

China unveils cutting-edge Tianhe Xingyi Supercomputer

Joane — Wed, 06 Dec 2023 18:28:26 +0000

On Wednesday, China proudly unveiled its latest triumph in the realm of supercomputing, the “Tianhe Xingyi Supercomputer.” The announcement, made by the National Supercomputing Center in Guangzhou at a prestigious industry event in Guangdong Province, was reported by state news agency Xinhua.

Unlike its predecessor, the Tianhe Xingyi boasts cutting-edge domestically designed architecture. While the exact computing power details remain shrouded in mystery, Lu Yutong, the director of the National Supercomputing Center, revealed that Tianhe Xingyi surpasses the famed Tianhe-2 across various critical dimensions. These encompass CPU computing power, networking efficiency, storage capacity, and application performance.

The image shown is for illustration purposes only

The Tianhe-2, developed by the National University of Defense Technology, once held the coveted title of the world’s fastest supercomputer for three consecutive years starting in 2013. However, its reign was abruptly disrupted in 2016 when the U.S. government blacklisted the university, limiting access to crucial Intel processors integral to supercomputing capabilities.

Tianhe Xingyi’s unveiling is not an isolated event in China’s supercomputing landscape. The Sunway TaihuLight, another prominent creation by the National Supercomputing Center in Wuxi, secured the seventh position in the June 2023 global rankings, while Tianhe-2 secured the tenth spot.

The lack of specific details regarding Tianhe Xingyi’s computing power has left the computing community eagerly anticipating further revelations. Lu Yutong, shedding light on the supercomputer’s operational expectations, emphasized its pivotal role in meeting the escalating computing demands across diverse fields. These include high-performance computing, artificial intelligence (AI) big model training, and big data analysis.

The purposeful design of Tianhe Xingyi underscores its commitment to addressing evolving challenges in computing, particularly in the realms of AI and big data applications.

Chinese Supercomputers:

China dominates the global supercomputing landscape, with its centers collectively holding a significant performance share of the world’s top 500 supercomputers. In November 2019, China outpaced the United States, boasting 228 supercomputers among the top 500.

The country’s supercomputing journey began in the 1980s, evolving rapidly to claim the top spot in November 2010 with Tianhe-1A. Notably, China’s Sunway TaihuLight and Tianhe-2A globally rank 7 and 10 respectively. In a bid to overcome potential technology embargoes, China has developed its processors, such as the Loongson, signaling a push towards self-sufficiency. However, geopolitical tensions have led to some Chinese supercomputing entities facing restrictions on the global stage.

Related:

(via)

The post China unveils cutting-edge Tianhe Xingyi Supercomputer appeared first on Gizmochina.

The New Vista Supercomputer Excels in AI and Scientific Computing

Anubhav — Mon, 13 Nov 2023 18:20:43 +0000

The Texas Advanced Computing Center (TACC) at the University of Texas, Austin, is unveiling Vista, an AI-focused supercomputer that promises to revolutionize the field of scientific computing. Marking a significant technological leap, Vista departs from the traditional x86-based architecture to Advanced RISC Machines (Arm) architecture, a move that caters to the burgeoning demands of AI and scientific computing.

All thanks to the Supercomputer’s advanced ARM architecture

Central to Vista’s capability is the NVIDIA Grace CPU Superchip, designed specifically for AI applications. This cutting-edge processor will power more than half of Vista’s compute nodes, combining a CPU with an NVIDIA Hopper architecture-based GPU. This integration enables the GPU to access CPU memory seamlessly, facilitating the operation of larger AI models and enhancing computing efficiency.

Vista’s design philosophy extends beyond mere processing power. It incorporates LPDDR5 technology for memory, similar to that found in laptops but optimized for data center needs. This approach not only offers higher bandwidth but also significantly improves power efficiency, a critical aspect in large-scale computing operations.

Vista’s introduction is a response to the evolving landscape of AI and scientific research, where traditional computing solutions are increasingly inadequate. It represents a paradigm shift in how supercomputers are designed, focusing on the specific needs of AI research and applications. The new system promises to provide researchers and scientists with unprecedented computing power, enabling breakthroughs in various fields, from climate modeling to medical research.

As AI continues to push the boundaries of what’s possible in scientific discovery, tools like Vista are key in harnessing this potential. By offering a powerful, efficient, and specialized computing resource, Vista is poised to be a cornerstone in the advancement of responsible and trustworthy AI research, reflecting the growing symbiosis between AI and supercomputing.

RELATED:

(Via)

The post The New Vista Supercomputer Excels in AI and Scientific Computing appeared first on Gizmochina.

Google Claims its 4000-Chip Supercomputer is Faster than NVIDIA’s

Anubhav — Wed, 05 Apr 2023 06:59:39 +0000

Modern supercomputers are some of the most powerful machines in the world, capable of processing vast amounts of data and performing complex calculations at incredible speeds. These machines are typically used for scientific research, data analysis, and other computationally intensive tasks that require massive amounts of computing power. Many modern supercomputers are built using clusters of hundreds or even thousands of individual processors, which are connected together to create a single, unified system. These machines can perform trillions of calculations per second and are used to tackle some of the most challenging problems in fields such as physics, astronomy, climate modeling, and more. With advances in technology continuing to push the boundaries of what’s possible, it’s clear that the age of supercomputing is only just beginning. Tech leaders, especially Google, has been at the forefront of AI for a long time now. Google’s Quantum Computing is cutting edge, and as per the company’s latest statement, its supercomputer is faster and more power-efficient than Nvidia’s machine.

Google has designed a unique chip called the Tensor Processing Unit (TPU) specifically for training AI models. With over 90% of the company’s AI training work relying on these chips, it’s clear that they are an essential part of Google’s success in the field. Google has just released details of their fourth-generation TPU, demonstrating how they used their custom-developed optical switches to connect over 4,000 chips into a single supercomputer. This innovation allows Google’s supercomputer to reconfigure connections between chips in real-time, improving performance and avoiding issues.

The rise of large language models in AI has sparked intense competition among companies building supercomputers. Google’s PaLM model, its largest publicly disclosed language model to date, took over 50 days to train by spreading it across two supercomputers with 4,000 chips. Google’s flexible interconnects help to overcome this challenge by allowing the company to change the topology of their supercomputer interconnects on the fly.

Google’s supercomputer has already been put to use, with startup Midjourney using the system to train their model, which generates images from text inputs. In a recent paper, Google revealed that their supercomputer is 1.7 times faster and 1.9 times more energy-efficient than a system based on Nvidia‘s A100 chip for a system of the same size. Google did not compare their fourth-generation product to Nvidia’s current flagship H100 chip, but there are hints that Google is working on a new TPU to compete with Nvidia’s latest offering.

RELATED:

(Via)

The post Google Claims its 4000-Chip Supercomputer is Faster than NVIDIA’s appeared first on Gizmochina.

Meta builds new AI supercomputer, claims it will be world’s fastest by 2022 end

Sean — Tue, 25 Jan 2022 11:54:05 +0000

Meta has just revealed that it has built a new “AI supercomputer.” The latest innovation from the company is a high speed computer that is designed to train machine learning systems.

The American tech giant claims that its AI Research SuperCluster (or RSC) is already one of the fastest machines of its kind. However, the company believes that it will even grow to become the world’s fastest by the year of this year. In a statement, Mark Zuckerberg, CEO of Meta, said that “Meta has developed what we believe is the world’s fastest AI supercomputer. We’re calling it RSC for AI Research SuperCluster and it’ll be complete later this year.”

Notably, the news showcases the focus of AI research from tech firms like Meta. Similarly, Microsoft and Nvidia have announced their own “AI supercomputers,” which differ from regular supercomputers. The newly revealed RSC will also be used to train a range of systems across Meta’s businesses. This includes functions like content moderation algorithms used to detect hate speech on social media platforms like Facebook and Instagram and would even include augmented reality features for the brand’s future AR based products.

Meta engineers Kevin Lee and Shubho Sengupta further added in a blog post that “RSC will help Meta’s AI researchers build new and better AI models that can learn from trillions of examples; work across hundreds of different languages; seamlessly analyze text, images, and video together; develop new augmented reality tools; and much more.”

Lenovo Becomes the World’s Largest Global Supercomputer Vendor

Jed John Ikoba — Thu, 28 Jun 2018 13:06:14 +0000

Just when we thought Lenovo‘s PC business was struggling, a new report has emerged which shows the company is actually climbing up the ranking of PC manufacturers. Actually, Lenovo has been ranked as the world’s top manufacturer of Supercomputers. A list of the top 500 fastest supercomputers was recently released with Chinese companies taking 206 of the 500 positions. This is a clear indication of how fast the Chinese are getting better at manufacturing Supercomputers.

The list, published Monday, shows the Chinese companies and government pulling away as the most prolific producer of supercomputers. This has been a market dominated by U.S companies in the past but China’s rise to the top 500 began two years ago. The latest listing shows the Chinese have extended that lead even further. Three Chinese companies are ranked among the top five makers of the 500 fastest supercomputers. Amazingly, Lenovo is ranked as number one while Inspur is third, and Sugon is fifth. The second and fifth spots are occupied by two U.S companies – Hewlett Packard Enterprise and Cray.

The manufacture of the most powerful supercomputers is seen as one measure of a nation’s technical prowess despite them being a rarefied niche of technology. In the past, Supercomputers were mostly found in national laboratories and used for government projects like simulating nuclear explosions and modelling weather patterns. But in recent times, more than half the 500 fastest computers are now working for corporations. The Supercomputer market is expected to double globally from 2017 to 2022 to more than US$9.5 billion, according to an estimate from Hyperion Research. For clarity, supercomputers are described as machines that cost more than US$500,000 each.

(source)

The post Lenovo Becomes the World’s Largest Global Supercomputer Vendor appeared first on Gizmochina.