• Gold-colored matte nvidia h100 80gb gpu with black ventilation slats cisco ucsc-gpu-h100-80 pcie gen5 card
1 1

Cisco

Cisco UCSC-GPU-H100-80 | NVIDIA H100 80GB PCIe GPU, 350W, 2-Slot FHFL

UCSC-GPU-H100-80

Stock Status: Enquire

Request Quote
Sale Sold out
Shipping calculated at checkout.

 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 We do not quote Gmail, Outlook or Hotmail addresses, or supply for home use. 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 NVIDIA H100 Tensor Core PCIe GPU is a GPU accelerator card for Cisco UCS C-Series rack servers. It is built for AI training and inference, large language models, HPC and data analytics in enterprise data centres.

This SKU carries 80GB of GPU memory with 2TB/s of memory bandwidth and runs at a 350W maximum thermal design power. It is a dual-slot, full-height, full-length air-cooled card on a PCIe Gen5 x16 interface rated at 128GB/s. The GPU can be partitioned into up to 7 Multi-Instance GPUs at 10GB each, and an NVLink bridge connection delivers 600GB/s between cards. Peak throughput includes 51 TFLOPS of FP32 and 3,026 TFLOPS of FP8 Tensor Core performance with sparsity.

It is listed for the Cisco UCS C240 M7 (SFF) as a PCIe GPU card, with a maximum of 2 per node installed in the Gen5 Riser 1C. Check the Cisco UCS Hardware Compatibility List for other supported UCS C-Series and X-Series servers.

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

Features

- Fourth-generation Tensor Cores and Transformer Engine with FP8 precision for AI training and inference on large language models
- Multi-Instance GPU (MIG) partitioning into up to 7 isolated instances
- PCIe Gen5 host interface for high-bandwidth connection to the server
- NVLink bridge support for connecting GPUs
- Dual-slot air-cooled design for standard rack servers
- NVIDIA AI Enterprise add-on listed for the H100 PCIe

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

Datasheet

Technical Specifications

- GPU memory: 80GB
- GPU memory bandwidth: 2TB/s
- Max thermal design power (TDP): 350W for this SKU (300-350W configurable)
- Host interface: PCIe Gen5 x16, 128GB/s
- Form factor: PCIe dual-slot, full-height full-length (FHFL), air-cooled
- FP64: 26 TFLOPS
- FP64 Tensor Core: 51 TFLOPS
- FP32: 51 TFLOPS
- TF32 Tensor Core: 756 TFLOPS (with sparsity)
- BFLOAT16 Tensor Core: 1,513 TFLOPS (with sparsity)
- FP16 Tensor Core: 1,513 TFLOPS (with sparsity)
- FP8 Tensor Core: 3,026 TFLOPS (with sparsity)
- INT8 Tensor Core: 3,026 TOPS (with sparsity)
- Decoders: 7 NVDEC, 7 JPEG
- Multi-Instance GPUs: Up to 7 MIGs at 10GB each
- NVLink bridge: 600GB/s
- Server options: Partner and NVIDIA-Certified Systems with 1-8 GPUs

FAQs

Q: Which Cisco servers is this GPU listed for?
A: It is listed for the Cisco UCS C240 M7 (SFF) as a PCIe GPU card, with a maximum of 2 per node installed in Riser 1C (Gen5 riser). It is not supported in Riser 2A or 2C. Check the Cisco UCS Hardware Compatibility List for other servers.

Q: How much power does this GPU need?
A: The H100 PCIe has a configurable TDP of 300-350W, and this SKU is the 350W configuration. It needs a GPU power cable from the server, so confirm the cabling for your server model.

Q: Does it need a software licence?
A: The NVIDIA datasheet lists NVIDIA AI Enterprise as an add-on for the H100 PCIe. Confirm the exact licence SKU required for your deployment with Cisco.

Q: Can I connect multiple H100 cards together?
A: Yes. The card supports an NVLink bridge at 600GB/s for connecting GPUs. The UCS C240 M7 supports up to 2 of these cards per node.

Q: Is this card air-cooled or liquid-cooled?
A: It is an air-cooled, dual-slot, full-height full-length PCIe card, designed for standard rack servers with suitable airflow.

Datasheet

This product has a detailed PDF datasheet available. Open PDF datasheet in a new tab .

Your browser can’t display the embedded PDF. Click here to open it in a new tab.

Related Products

View all Graphics Card | GPU

  • 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