1 1

Cisco

Cisco UCSX-580P-U | UCS X580p Gen5 PCIe GPU Node, 4x Dual-Slot GPUs

UCSX-580P-U

Stock Status: Enquire

Request Quote
Sale Sold out
Shipping calculated at checkout.
20-min response

Need a quote or bulk pricing? Answered within 20 minutes, Sydney business hours.

Description

The UCS X580p is a second-generation PCIe resource node for the Cisco UCS X-Series Modular System. It hosts up to four full-height, full-length dual-slot PCIe Gen5 GPUs and shares them with up to two compute nodes through UCS X-Fabric. It suits generative AI fine-tuning and inference, LLM inference, VDI, graphics and rendering including NVIDIA Omniverse, scientific computing and professional visualisation.

Each node provides up to four PCIe Gen5 x16 slots and occupies two slots in the UCS X9508 chassis, so a chassis can hold up to four X580p nodes and up to eight GPUs. Supported GPUs are the NVIDIA H200 NVL, L40S, RTX PRO 6000 Blackwell Server Edition and RTX PRO 4500 Blackwell Server Edition. GPU power draw depends on the cards fitted: 600W for the H200 NVL and RTX PRO 6000, 350W for the L40S and 185W for the RTX PRO 4500. The node is powered from the chassis and managed through Cisco Intersight in Intersight Managed Mode, which needs a minimum Essentials licence per server.

The node requires UCS X9516 X-Fabric modules and works with UCS 4th, 5th and 6th Generation Fabric Interconnects. It pairs with UCS X210c M8 and X215c M8 compute nodes, each of which needs a UCSX-V5-PCIME mezzanine to reach the GPUs.

Enquire with XS Network Tech for new or used pricing and availability.

Features

- Supports up to four air-cooled GPUs with NVLink Bridge support, without performance impact or power throttling per compute node
- One X580p can be shared across up to two compute nodes in 1:1 or 1:2 mapping
- Pairs with UCS X210c M8 (Intel Xeon 6) and X215c M8 (AMD EPYC) compute nodes
- Built on the NVIDIA MGX modular reference design, with support planned for next-generation NVIDIA GPUs
- Programmable PCIe fabric through UCS X9516 X-Fabric for flexible GPU provisioning from Intersight
- SmartNIC support for GPU-to-GPU communication over an AI fabric
- Policy-based GPU provisioning and unified monitoring of compute and GPU resources through Cisco Intersight
- Midplane-free X9508 chassis architecture eases airflow and future upgrades

Warranty

All products sold by XS Network Tech include a 12-month warranty on both new and used items. Our in-house technical team thoroughly tests used hardware prior to sale to ensure enterprise-grade reliability.

All technical data should be verified on the manufacturer data sheets.

View full details

specs-tabs

Technical Specifications

FAQs

Technical Specifications

- Part number: UCSX-580P-U
- Node type: Cisco UCS X-Series Gen5 PCIe node
- PCIe slots: Up to four PCIe Gen5 x16 full-height, full-length dual-slot cards
- Supported GPUs: NVIDIA H200 NVL, L40S, RTX PRO 6000 Blackwell Server Edition, RTX PRO 4500 Blackwell Server Edition
- Chassis slots occupied: 2 slots in the UCS X9508 chassis
- Nodes per chassis: Up to 4 X580p nodes (up to 8 GPUs)
- Compatible chassis: Cisco UCS X9508
- Compatible X-Fabric modules: Cisco UCS X9516
- Compatible fabric interconnects: Cisco UCS 4th, 5th and 6th Generation
- Compatible compute nodes: UCS X210c M8, UCS X215c M8
- Management: Cisco Intersight (SaaS), Intersight Managed Mode
- Operating temperature: 10 to 27 C (50 to 81 F)
- Nonoperating temperature: -40 to 65 C (-40 to 149 F)
- Operating humidity: 5% to 93% noncondensing
- Nonoperating humidity: 5% to 93% noncondensing
- Operating altitude: 0 to 3000 m (0 to 10,000 ft), maximum ambient temperature decreases by 1 C per 300 m
- Nonoperating altitude: 12,000 m (40,000 ft)

FAQs

Q: Does the X580p need a licence?
A: The node itself has no separate licence listed. It must be managed through Cisco Intersight in Intersight Managed Mode, which requires a minimum Essentials licence per server.

Q: Which GPUs can be fitted?
A: It supports the NVIDIA H200 NVL (UCSX-GPU-H200-NVL), L40S (UCSX-GPU-L40S), RTX PRO 6000 Blackwell Server Edition (UCSX-GPU-RTXP6000) and RTX PRO 4500 Blackwell Server Edition (UCSX-GPU-RTXP4500). Each slot takes a full-height, full-length dual-slot card.

Q: What chassis and fabric does it need?
A: It installs in the Cisco UCS X9508 chassis, occupies two slots, and requires UCS X9516 X-Fabric modules. It works with UCS 4th, 5th and 6th Generation Fabric Interconnects.

Q: Which compute nodes can use the GPUs?
A: UCS X210c M8 and X215c M8 compute nodes. Each paired compute node needs a UCSX-V5-PCIME mezzanine to reach the GPUs, and one X580p can be shared across up to two compute nodes.

Q: How much power do the GPUs draw?
A: The node is powered from the chassis. Draw depends on the GPUs fitted: 600W for the H200 NVL and RTX PRO 6000, 350W for the L40S and 185W for the RTX PRO 4500. Use the Cisco power calculator for exact configurations.

Related Products

View all Graphics Card | GPU

Recently Viewed

  • Request a Quote

    Looking for competitive pricing? Submit a request, and our team will provide a tailored quote that fits your needs.

  • Contact Us Directly

    Have a question or need immediate assistance? Call us for expert advice and real-time support.

    Call us Now  

  • Contact Us Directly

    Have a question or need immediate assistance? Call us for expert advice and real-time support.

    Contact us