New HPE ProLiant Compute XD685 helps eight AMD Intuition™ MI325X accelerators and two AMD EPYCTM CPUs to ship optimum efficiency and suppleness to effectively construct and practice giant language fashions
Hewlett Packard Enterprise (NYSE: HPE) introduced the HPE ProLiant Compute XD685 for complicated AI mannequin coaching duties, powered by 5th Gen AMD EPYC™ processors and AMD Intuition™ MI325X accelerators. The brand new HPE system is optimized to shortly deploy high-performing, safe and power environment friendly AI clusters to be used in giant language mannequin coaching, pure language processing and multi-modal coaching.
The race is on to unlock the promise of AI and its potential to dramatically advance outcomes in workforce productiveness, healthcare, local weather sciences and rather more. To seize this potential, AI service suppliers, governments and huge mannequin builders require versatile, high-performance options that may be delivered to market shortly.
“Coaching giant language fashions, and doing so effectively, requires strong-scaling, huge parallel computing capabilities, and distinctive providers that solely HPE’s excessive efficiency computing options ship,” mentioned Trish Damkroger, senior vp and common supervisor, HPC & AI Infrastructure Options at HPE. “As a frontrunner in delivering the world’s strongest and energy-efficient methods, along with AMD, we’re extending this innovation to the brand new HPE ProLiant Compute XD685 to assist a rising market of AI mannequin builders and speed up scientific and engineering breakthroughs throughout industries.”
“Via the highly effective mixture of the newest AMD EPYC processors and Intuition accelerators, HPE ProLiant Compute XD685 prospects can speed up time to market and expertise true management in AI efficiency and effectivity,” mentioned Forrest Norrod, govt vp and common supervisor, Knowledge Heart Options Group, AMD. “Our collaboration with HPE continues to ship versatile, extremely performant options that assist maximize AI efforts for elevated demand in giant language mannequin coaching and advances aggressive {industry} innovation.”
Velocity to answer and suppleness improve aggressive edge for AI
The HPE ProLiant Compute XD685 delivers velocity to answer for big AI mannequin coaching and tuning tasks utilizing a brand new HPE-designed modular chassis. The modular, compact 5U chassis shortens time to marketplace for options and offers the flexibleness to accommodate a variety of GPUs, CPUs, parts, software program and cooling strategies. Moreover, the HPE ProLiant Compute XD685 brings collectively HPE’s a long time of direct liquid cooling experience, together with safety innovation with HPE iLO, making a high-performance answer designed to be sustainable and safe.
- Run the most important AI fashions with fewer GPUs: Powered by AMD CDNA™ 3 structure, AMD Intuition™ MI325X, accelerators supply main AI efficiency and effectivity for coaching and inference duties. The MI325X accelerators characteristic management HBM3E reminiscence capability with 6 TB/s of reminiscence bandwidth optimizes efficiency and lowers TCO.
- Enhance effectivity with elective direct liquid cooling: The direct liquid cooling choice helps organizations meet escalating energy necessities, advance sustainability objectives and decrease operational prices—all by leveraging HPE’s a long time of experience and market management in deploying the world’s largest liquid-cooled IT environments.
- Speed up deployment and operations: HPE Companies presents the total vary of customizable and versatile providers to put in and deploy giant AI clusters wherever on this planet with stability and operational excellence. Knowledgeable providers construct, combine, validate, absolutely take a look at and customise options within the manufacturing unit, facilitating faster onsite deployment.
- Shield in opposition to rising safety threats: Securely handle HPE ProLiant Compute XD685 methods with HPE iLO, delivering seamless administration and built-in, industry-leading safety innovation that begins within the provide chain and is rooted within the silicon.
- Optimize rack density: The HPE-designed 5U server chassis makes use of a compact 8-nodes-per-rack association to maximise rack density for 8-way GPU methods. The HPE ProLiant Compute XD685 helps eight AMD Intuition™ MI325X or AMD Intuition™ MI300X accelerators, two of the newest AMD EPYC™ 9005 Sequence processors and air cooling or direct liquid cooling.
- Simplify manageability of enormous AI clusters: HPE Efficiency Cluster Supervisor is a completely built-in system administration software program that will get complicated methods up-and-running shortly with automated setup from naked steel and retains clusters wholesome and operational with detailed telemetry, GPU stress assessments, and extra.
Availability
The HPE ProLiant Compute XD685 is offered to order at the moment via HPE and can be usually out there within the first quarter of 2025.
Join the free insideAI Information e-newsletter.
Be part of us on Twitter: https://twitter.com/InsideBigData1
Be part of us on LinkedIn: https://www.linkedin.com/firm/insideainews/
Be part of us on Fb: https://www.fb.com/insideAINEWSNOW