HPE Unveils Liquid-Cooled Supercomputing Options and New AI Techniques for Enterprises and Service Suppliers

New portfolio of compute, networking, storage and software program advances leadership-class supercomputers and purpose-built AI coaching options

• New merchandise provide alternative of air cooling or HPE’s industry-first 100% fanless direct liquid cooling system structure

• Supercomputing and large-scale AI servers characteristic next-generation compute and accelerators from AMD, Intel and NVIDIA, which give clients with choices to greatest go well with efficiency wants and price necessities

• HPE expands portfolio of servers optimized for high-end AI coaching and tuning workloads

In the present day, Hewlett Packard Enterprise (HPE)broadcasts its new excessive efficiency computing (HPC) and synthetic intelligence (AI) infrastructure portfolio that features leadership-class HPE Cray Supercomputing EX options and two techniques optimized for giant language mannequin (LLM) coaching, pure language processing (NLP) and multi-modal mannequin coaching. The brand new supercomputing options are designed to assist world clients fast-track scientific analysis and invention.

Additionally Learn: Kloudfuse Launches Kloudfuse 3.0: One Platform, Limitless Potentialities

“Service suppliers and nations investing in sovereign AI initiatives are more and more turning to high-performance computing because the crucial spine enabling large-scale AI coaching that accelerates discovery and innovation”

“Service suppliers and nations investing in sovereign AI initiatives are more and more turning to high-performance computing because the crucial spine enabling large-scale AI coaching that accelerates discovery and innovation,” mentioned Trish Damkroger, senior vice chairman and basic supervisor, HPC & AI Infrastructure Options at HPE. “Our clients flip to us to fast-track their AI system deployment to appreciate worth sooner and extra effectively by leveraging our world-leading HPC options and a long time of expertise in delivering, deploying and servicing fully-integrated techniques.”

Finish-to-end portfolio of industry-leading HPC options: HPE Cray Supercomputing EX

HPE leads the best way offering a few of the quickest and most power environment friendly supercomputers on this planet. Primarily based on HPE Cray Supercomputing EX techniques, HPE’s net-new choices for its whole leadership-class HPC portfolio are designed for analysis establishments entrusted with fixing the world’s greatest issues and authorities entities creating sovereign AI initiatives. The portfolio relies on the {industry}’s first 100% fanless direct liquid cooling system structure and spans each layer of HPE’s supercomputing options together with compute nodes, networking and storage, that are supplemented by a brand new software program providing.

• HPE Cray Supercomputing EX4252 Gen 2 Compute Blade – Able to delivering as much as 98,304 cores in a single cupboard, the HPE Cray Supercomputing EX4252 Gen 2 Compute Blade delivers probably the most highly effective one-rack unit system accessible for supercomputing. That includes eight 5^th Gen AMD EPYC™ processors, this compute blade provides the good thing about CPU density, permitting clients to appreciate higher-performing compute throughout the similar house. HPE Cray Supercomputing EX4252 Gen 2 Compute Blade might be accessible Spring 2025.
• HPE Cray Supercomputing EX154n Accelerator Blade – To drastically scale back the time it takes to finish a supercomputing workload, the HPE Cray Supercomputing EX154n Accelerator Blade can accommodate as much as 224 NVIDIA Blackwell GPUs in a single cupboard. That includes the NVIDIA GB200 Grace Blackwell NVL4 Superchip, every accelerator blade holds 4 NVIDIA NVLink™-connected Blackwell GPUs unified with two NVIDIA Grace CPUs over NVIDIA NVLink-C2C. Normal availability for HPE Cray Supercomputing EX154n Accelerator Blade is predicted by the tip of 2025.
• HPE Slingshot interconnect 400 – The following era of HPE’s exascale-capable interconnect portfolio provides community interface controllers (NICs), cables and switches at 400 gigabit-per-second speeds. HPE Slingshot interconnect 400 delivers twice the road pace over the earlier eraⁱ whereas providing options like automated congestion administration and adaptive routing for ultra-low tail latency, permitting clients to run giant workloads with considerably much less community infrastructure. This model of HPE Slingshot might be accessible for clusters primarily based on HPE Cray Supercomputing EX techniques starting Fall 2025.
• HPE Cray Supercomputing Storage Techniques E2000 – This high-performance storage system designed for large-scale supercomputers greater than doubles the enter/output (I/O) efficiency in comparison with the earlier eraⁱⁱ. HPE Cray Supercomputing Storage Techniques E2000 relies on the open supply Lustre file system and permits higher utilization of each CPU and GPU-based compute nodes by lowering idle time throughout I/O operations. The HPC storage system will turn into usually accessible on HPE Cray Supercomputing EX techniques in early 2025.
• HPE Cray Supercomputing Person Providers Software program – HPE is introducing a brand new software program providing that improves the person expertise of working compute-intensive workloads. Obtainable now, HPE Cray Supercomputing Person Providers Software program consists of options that assist clients optimize system effectivity, regulate energy consumption, and flexibly run numerous workloads on supercomputing infrastructure.

Additionally Learn: AiThority Interview Clarence Rozario, the World Head of Zoho Analytics Enterprise

New HPE ProLiant Compute XD server household optimizes for AI mannequin coaching and tuning

HPE continues the rollout of a brand new class of servers that permits clients to streamline deployment of enormous, highly-performant AI clusters. Designed for service suppliers and huge enterprises coaching their very own AI fashions, HPE ProLiant Compute XD servers leverage the corporate’s experience in putting in and deploying giant AI techniques. Non-obligatory HPE Providers can be found to assist constructing, customization, integration, validation, and full testing of the answer inside HPE’s state-of-the-art manufacturing facility to expedite on-site deployment.

Solely accessible on HPE ProLiant Compute servers, HPE Built-in Lights-Out (iLO) administration expertise permits choose licensed personnel with out-of-band distant management entry to servers, thereby enhancing safety over customary in-band community entry.

• HPE ProLiant Compute XD680 server – Optimized with price-for-performance in thoughts, the air-cooled HPE ProLiant Compute XD680 server is designed to deal with demanding AI coaching, tuning and inferencing workloads. An HPE-designed chassis homes eight Intel Gaudi 3 AI accelerators in a single compact node. HPE ProLiant Compute XD680 server with Intel Gaudi 3 might be accessible in December 2024.
• HPE ProLiant Compute XD685 server – For patrons prioritizing efficiency, aggressive benefit and power effectivity, a brand new model of the HPE ProLiant Compute XD685 server will turn into accessible with NVIDIA GPUs to speed up coaching for giant, advanced AI fashions. The server is powered by eight NVIDIA H200 SXM Tensor Core GPUs or NVIDIA Blackwell GPUs in a 5 rack-unit chassis and leverages HPE’s multi-decades experience in liquid cooling to effectively cool GPUs, CPUs and switches. The NVIDIA HGX H200 8-GPU model of HPE ProLiant Compute XD685 server will turn into accessible in early 2025 and HPE might be time-to-market with NVIDIA Blackwell GPUs.

A model of HPE ProLiant Compute XD685 server that includes eight AMD Intuition™ MI325X accelerators and two AMD EPYC™ CPUs was beforehand introduced in October. HPE ProLiant Compute XD servers are a part of HPE’s complete AI choices that embrace HPE Non-public Cloud AI and HPE ProLiant Compute DL servers.

Because the wants of shoppers evolve, HPE continues to push the boundaries of innovation by assembly market calls for and demonstrating why HPE is wanted to assist the rising variety of conventional supercomputing clients which can be utilizing AI fashions to boost scientific discovery. Discover HPE options on show at SC24 (sales space #2219) in Atlanta, November 17-22 for extra data on the next-generation of HPC and large-scale AI techniques.

[To share your insights with us as part of editorial or sponsored content, please write to psen@itechseries.com]