High-Performance Computing as a Service Innovative Products to be Launch Soon

Several companies are trying to develop products to boost up high performance computing as a service. Companies are making much use of supercomputers to solve complex problems. They are adopting certain strategies such as acquisition, collaboration, innovation, and technological advancements. Companies utilizing supercomputers expect products that can increase the efficiency of the system and take maximum advantage of the system to solve complex problems. NVIDIA has developed an Ampere Data Centre Graphics Processing Unit (GPU) to make supercomputers faster than before. Lenovo collaborated with Karlsruhe Institute of Technology (KIT) to create a supercomputer that can solve many complex problems in less time with maximum accuracy. According to the report published by Allied Market Research, the high- performance computing as a service market is expected to reach $17.00 billion by 2026. The companies have been riding on the current trends.

Several market players operating in the high performance computing as a service sector are developing new products and innovating the existing products to grow demand in the market. Alchip Technologies Limited launched a Chip on Wafer on Substrate (CoWoS), which specifically targets high performance computing (HPC) ASIC devices. This device is necessary for successful deployment of modern HPC. Some information about CoWos is-

• CoWoS packaging was developed by TSMC.
• It is a 2.5D wafer level multichip packaging technology.
• It incorporates multiple sides by side dye on a silicon interposer.
• Chips are bonded through micro-bumps on a silicon interposer.
• The improvement of performance takes place due to the configuration.

Alchip's CoWoS design service includes power supply planning, architecture planning, SoC floor painting, interposer dye placement, package ball assignments, and IP/IO placement collaboration. According to the company's officials, the product offers greater performance in a smaller physical form.

Companies are shaking hands with educational institutions to deliver a great product for improving high performance computing as a service. The current example of this is the deal between Lenovo and the Karlsruhe Institute of Technology (KIT). KIT promised to deliver the supercomputer system. The key points of this deal are-

• The supercomputer is named- HoreKa.
• It will be handed over to the scientific communities in 2021.
• The contract signed is of EUR 15 million.
• In 2021, the Karlsruhe supercomputer (HoreKa) will be used to understand complex systems in several research areas.

The supercomputer will provide advantage in the high performance computing market over the competitors.
It will lead to the development of innovative products and processes in key economic areas.

The "Hochleistungrechner Karlsruhe", HoreKa for short, is expected to be one of the 10 most powerful computers in Europe. Researchers can analyze much more details than before in large systems. The computer systems installed in the supercomputer are manufactured by Lenovo, whereas, general contractor pro-com Datensysteme GmbH is in charge of system integration, delivery, project coordination, and customer support.

To gain a greater share in the high performance computing as a service market than the competitors available in the market, several companies are acquiring other companies to improve their product quality and acquire market share. One of the examples is NVIDIA. Recently, the company announced that they are planning to acquire Cumulus Networks. Cumulus Networks is a pioneer of hyperscale networking. This strategy will-

• Boost up NVIDIA's networking software capabilities.
• Take the company to a new era of the software-defined data center.
• Offer an opportunity to innovate and optimize in the entire networking stack for NVIDIA.

Cumulus supports more than 100 hardware platforms. It will give NVIDIA a huge benefit as the product quality will improve and they can enter a whole new segment in high performance computing as a service market.

Talking about NVIDIA, recently the company launched an Ampere Data Centre Graphics Processing Unit (GPU). The company announced the arrival of its new product i.e. NVIDIA A100. This is based on the first GPU based on the NVIDIA Ampere architecture.

• It is the largest leap by the company in performance within its eight generations of GPU.
• It can unify the AI training and interface.
• It can boost up performance by up to 20 times over its previous version.
• It is built for data analytics, cloud graphics, and scientific computing.

The new computing technologies built into A100 make computers carry out every job. It interconnects technology with multiple GPUs to operate as one giant GPU to do big tasks.

After the launch, many companies and research institutions such as Microsoft, Max Planck Computing and Data Facility, and Karlsruhe Institute of Technology are powering their supercomputers with A100. This will lead companies to technological advancements and offer a leading edge over the competitors in the high performance computing as a service market.

References:

https://insidehpc.com/2020/05/video-nvidia-launches-ampere-data-center-gpu/

https://www.hpcwire.com/off-the-wire/kit-to-deploy-horeka-supercomputer-supplied-by-lenovo/

Related Post