Supermicro presents new Intel-based X14 server with maximum performance for AI, HPC and critical external workloads – Hartware

(Auszug aus der Pressemitteilung)

Supermicro, Inc. (NASDAQ: SMCI), a leading provider of IT solutions for AI/ML, HPC, Cloud, Storage, and 5G/Edge, today announced a new, all-new X14 server platform for next-generation technology, uhm the performance for resource-intensive workloads and applications to maximize. At the end of the Supermicros efficient X14 server, due to hit the market in June 2024, new system-significant upgrades are being made across the board. There are not yet 256 Performance Cores (P-Cores) in a single button, storage support for MRDIMMs with its 8800MT/s and compatibility with next-generation SXM, OAM and PCIe GPUs. This combination can provide the AI ​​and the important data distribution and the time and cost wall for a wide range of AI training, high performance assessments and complex data analyses. Additional knowledge can consist of Supermicros Early Ship program or for early testing with Supermicro JumpStart a fresh start with completes, all functional systems are safe.

“With these new platforms, which are still fresh and have new functionalities, we can further expand the data center building blocks,” said Charles Liang, president and CEO of Supermicro. “Supermicro is the best to offer these high-speed solutions in rack mass stack, and with the ramifications of direct-to-chip solutions, complete rack integration services and an excellent production cap of 5,000 racks per month, will work at least 1,350 fluid-cooled racks. With a good delivery time, the delivery can be optimally optimized, which no longer offsets the Time-to-Delivery and reduces the TCO.”

Click here for more information.

This new X14 system is equipped with a new modern architecture, with new 10U and Multi-Node form tower, supporting the latest next-generation GPUs and higher CPU features, with 12 specific CPU capabilities and new MRDIMMs, the display of DDR5-6400 DIMMS is one of the 37% of the best performance. It may be that more functions are needed to bring the work processes closer together and integrate more systems with direct integration into the server architecture.

The new generation in the Supermicro X14 family is increasingly focused on the new system, by finding a new complete architecture in three different workload-specific categories:

• GPU optimized platforms, which can be used in a clean way and can provide extensive thermal capabilization, are the main GPUs that can use them. The system architectures can be used by Grund for large English AI training, LLMs, generative AI, 3D media and virtualization solutions.
• Higher density multi-knobs, a SuperBlade® and the new FlexTwin™, the Direct-to-Chip fluid cooling nut, one of the performance cores in a Standard Rack in the comparison of the German system generation that is German.
• The market-standard Hyper Rackmounts combine single- or dual-socket architectures with flexible I/A and Speicher configurations in versatile form factors, with an internal and external processing function and reduced work costs.

Supermicro’s optimized X14 systems are equipped with Intel® Xeon® 6900-series P-core processors over time and will be compatible with Intel Xeon 6900-series E-core processors in Q1 2025. This built-in feature is light for the latest optimized system for pairing Power pro Core or Power pro Watt.

“The new Intel Xeon 6900-series processors with P-cores are no longer the standard, end-to-end processors, with more cores and greater storage bandwidth and I/O, delivering new levels of performance for the AI ​​and resource-intensive workloads out there,” said Ryan Tabrah, VP and GM of Xeon 6 at Intel. “Our deepened partnership with Supermicro delivers an industry-first system that combines the industry’s leading edge requirements with modern AI and high-performance computing.”

Configured with Intel Xeon 6900-series P-core processors, Supermicro’s new FP16 systems are installed on integrated Intel® AMX devices, further expanding the performance of AI workloads. This system offers 12 CPU specifications with support for DDR5-6400 and MRDIMMs running at 8800MT/s, CXL 2.0, and a number of industry-standard high-density EDSFF E1.S and E3.S NVMe running solutions.

Supermicros fluid cooling solutions

There is an enhanced X14 product portfolio by the Supermicros functionality for rack scale integration and flow solutions. With a global branch, an extensive integration and testing facility in rack scale format and a customized palette of management software solutions, tested, validated and lovely Supermicro complete solutions in your larger inner half-weeks.

Furthermore, Supermicro offers a complete, even built-in whistle function, cooling plates for CPUs, GPUs and storage, Cooling Distribution Units, Cooling Distribution Manifolds, Shut-offs, Connections and Cooling Towers. The flow cooling can cause problems with the integrated rack integration, a further increase in system efficiency, the powerful throttle will reduce and the total operating costs (TCO) that the total environmental costs (TCE) of research centers reduce.

The recommended performance optimizations of the Supermicro X14 system are:

GPU optimized – The most powerful Supermicro X14 system is equipped with a number of AI Training, Large Language Models (LLMs), Generative AI and HPC, and eight next-generation SXM5 and SXM6 GPUs. These systems are intended for air or airflow devices.

PCIe GPU – Benefit from maximum GPU flexibility, with 10 dual-width PCIe 5.0 cards in a thermally optimized 5U chassis. This server is ideal for media, collaborative design, simulation, cloud gaming, and virtualization workloads.

Intel® Gaudi® 3 KI-Beschleuniger – Supermicro factory uses the expansion of the branches with AI servers, based on the Intel Gaudi 3 Beschleuniger, on the Intel Xeon 6 processors hosted. The system offers efficiency and cost for training AI models in large sizes and for AI inferencing. The system features eight Intel Gaudi 3 versions of an OAM universal base board, integrates OSFP ports for cost-efficient scale-out networking and an offensive platform, which provides for the pampering of a community base, which can configure software stacks and incurs no software license costs.

SuperBlade® – Supermicro’s SuperBlade X14 6U is high capacity, dense optimization, energy efficiency, and maximum rack density with 100 servers and 200 GPUs per Rack. Optimize for AI, HPC, and other resource-intensive workloads, choose the best air cooling or direct-to-chip cooling, maximum efficiency, and the new PUE value with the best TCO there is, so who are Konnektivität with its four integrated Ethernet switches with 100G uplinks and front I/O offer a choice of flexible networking options on 400G InfiniBand or 400G Ethernet-professional-efficient-competitive.

FlexTwin™ – The new Supermicro X14 FlexTwin architecture delivers maximum computing power and density in a multi-node configuration with 24,576 performance cores in a 48U rack. Optimized for HPC and other resource-intensive workloads, enabling Direct-to-Chip fluid cooling, maximizing efficiency and reducing CPU throttling so you can leverage HPC Low Latency Front and Rear E/Os, supporting a range of flexible networking options up to 400G per node.

Hyper – X14 Hyper is Supermicro’s flagship rackmount platform, offering the highest performance for running AI, HPC, and business applications. Configuration with one or two other sockets supports dual-expansion PCIe GPUs for maximum workload performance. There is an air-cooled model as well as a direct-chip fluid cooling model, one of the top-bin CPUs that increase processing efficiency and cost-effectiveness for the evaluation of the evaluation and its efficiency.