Skip to product information
1 of 1

Cisco

Cisco UCSC-885A-M8-M353 | UCS C885A M8 8x MI350X GPU Rack Server, 2.3TB RAM

SKU:UCSC-885A-M8-M353

Request Quote
Sale Sold out
Shipping calculated at checkout.

Stock Status: Enquire

 Trade supply only. Server & GPU quote requests need a company name, ABN and work email address. We do not quote Gmail, Outlook or Hotmail addresses, or supply for home use. 

20-min response

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

Custom Built Servers

Need a different spec, quantity or configuration?

We build configured-to-order refurbished servers from Cisco, Dell, HPE, NVIDIA Lenovo and Supermicro. Tell us what you need and we'll come back with options within 24 hours.

Configure your server now →

Description

The UCS C885A M8 is an 8RU dense-GPU rack server from the Cisco UCS C-Series, built for large deep learning and LLM training, model fine-tuning, large model inferencing, Retrieval-Augmented Generation and HPC workloads in AI clusters. It pairs eight AMD Instinct MI350X OAM GPUs with dual 5th Gen AMD EPYC 9575F 64-core processors running at 3.3 GHz, with a maximum boost of 5 GHz.

This SKU is a fixed configuration with 24x 96GB DDR5 RDIMMs for 2.3TB of memory, 2x 960GB M.2 NVMe boot drives and 2x 1.92TB NVMe SSDs for local data caching. Networking comprises 8x NVIDIA ConnectX-7 1x400G east-west NICs, one per GPU, plus 2x ConnectX-7 2x200G north-south NICs and an OCP 3.0 Intel X710-T2L 2x10GbE RJ45 NIC. Power is supplied by 2x 2.7kW 12V (N+1) and 6x 3kW 54V (N+2) hot-swappable PSUs, with system consumption up to 12.5 kW.

Management is through Cisco Intersight, which requires an Intersight SaaS or Private Virtual Appliance subscription. The server is part of the Cisco AI POD reference architectures alongside Nexus 9364E-SG2 switches, and Cisco optics are selectable for the ConnectX-7 NICs.

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

Features

- Eight AMD MI350X OAM GPUs in one server for large-model training, fine-tuning, inferencing and RAG
- High clock speed 5th Gen AMD EPYC CPUs selected for training workloads
- One 400G NVIDIA ConnectX-7 NIC per GPU for scale-out inter-server, inter-GPU connectivity
- Local NVMe SSDs for data caching
- Hot-swappable redundant power supplies for high availability
- Cisco Intersight management with inventory through the Device Connector
- Part of the Cisco AI POD reference architectures

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

Collapsible content

Technical Specifications

FAQs

Technical Specifications

- Form factor: 8RU rack server
- Processors: 2x 5th Gen AMD EPYC 9575F, 64 core, 3.3 GHz (max boost 5 GHz)
- GPUs: 8x AMD MI350X OAM (UBB)
- GPU memory: up to 1.5 TB
- Memory: 24x 96GB DDR5 RDIMMs up to 6,400 MT/s (2.3TB)
- Boot drives: 2x 960GB M.2 NVMe SSD
- Server drives: 2x 1.92TB NVMe SSD
- East-west NICs: 8x NVIDIA ConnectX-7 (1x400G), PCIe Gen5 x16 HHHL
- North-south NICs: 2x NVIDIA ConnectX-7 (2x200G), PCIe Gen5 x16 FHHL
- OCP LAN: 1x OCP 3.0 PCIe Gen5 x8 Intel X710-T2L 2x10GbE RJ45
- Power supplies: 2x 2.7kW 12V CRPS (N+1) and 6x 3kW 54V MCRPS (N+2), hot-swappable
- System power consumption: up to 12.5 kW
- Cooling: 12x 80105 hot-swappable fans (N+1) plus 4x 6056 SSD cooling fans
- Front I/O: 2x USB 2.0, ID button, power button
- Rear I/O: USB 3.0 A, USB 3.0 C, mDP, USB 2.0 C (debug), RJ45 management
- Management: Cisco Intersight, BMC via DC-SCM, IPMI 2.0, KVM with dedicated LAN
- Dimensions (H x W x D): 353 x 447 x 800 mm
- Weight: 120 kg
- Operating temperature: 5 to 35 C, derated 1 C per 304.8 m up to 3048 m
- Operating humidity: 8% to 90% non-condensing

FAQs

Q: Does this server need a software licence or subscription?
A: No licence is included in the base SKU. A Cisco Intersight subscription (Essentials or Advantage, SaaS or Private Virtual Appliance) is a required selection, and NVIDIA AI Enterprise is optional and sold per GPU.

Q: What power does it need?
A: It uses 2x 2.7kW 12V PSUs (N+1) and 6x 3kW 54V PSUs (N+2), all hot-swappable. System consumption is up to 12.5 kW and input power cords are a required selection.

Q: How does this SKU differ from other UCS C885A M8 configurations?
A: This SKU has 24x 96GB (2.3TB) of RAM, 8x ConnectX-7 1x400G east-west NICs and 2x ConnectX-7 2x200G north-south NICs, with 8x AMD MI350X GPUs.

Q: What NICs and optics does it work with?
A: The north-south slots support NVIDIA ConnectX-7 1x400G, 2x200G and 4x25G NICs. The east-west slots support ConnectX-7 1x400G or AMD Pensando Pollara 400, and Cisco optics are selectable in the configurator.

Q: What is it designed for?
A: It targets large deep learning and LLM training, model fine-tuning, large model inferencing, RAG and HPC workloads in AI clusters.

  • 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