Skip to product information
1 of 1

Cisco

Cisco UCSX-GPU-A100-80 | UCS X-Series A100 80GB GPU, Passive 300W, Riser A

SKU:UCSX-GPU-A100-80

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 We do not quote Gmail, Outlook or Hotmail addresses, or supply for home use. 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 X-Series NVIDIA A100 80GB is a PCIe Gen4 Tensor Core GPU accelerator built for the Cisco UCS X440p PCIe Node. It suits AI and machine learning training and inference, data analytics and HPC GPU offload for UCS X-Series compute nodes in a UCS X9508 chassis.

The card carries 80GB of HBM2e memory with 1,935 GB/s of bandwidth and a 300W maximum TDP. It is passively cooled and relies on chassis airflow. Peak throughput includes 9.7 TFLOPS FP64, 19.5 TFLOPS FP64 Tensor Core and 19.5 TFLOPS FP32, with up to 624 TFLOPS FP16 and BFLOAT16 Tensor Core performance using sparsity. Multi-Instance GPU partitions the card into up to 7 instances of 10GB each. This SKU is the dual-slot, full-height full-length Riser A (Gen4) version for the standard X440p node, not the -D variant.

Up to 2 cards are supported per X440p PCIe Node, which pairs with UCS X210c M6/M7 and X410c M7 compute nodes through UCS X9416 X-Fabric modules. An NVLink bridge can link 2 GPUs at 600GB/s.

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

Features

- Third-generation Tensor Cores for AI training, inference and HPC
- Multi-Instance GPU partitions the card into up to seven hardware-isolated instances
- NVLink bridge support for 2 GPUs at 600GB/s
- Structural sparsity for up to 2X higher performance on sparse models
- 80GB HBM2e memory for large models and datasets
- Up to 16 GPUs per UCS X9508 chassis using X440p PCIe Nodes with X-Fabric
- Managed through Cisco Intersight and UCS Manager

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

- GPU: NVIDIA A100 80GB PCIe (Ampere) Tensor Core
- GPU memory: 80GB HBM2e
- Memory bandwidth: 1,935 GB/s
- Max TDP: 300W
- Cooling: Passive
- Host interface: PCIe Gen4, 64 GB/s
- Form factor: Dual-slot, full-height full-length (FHFL)
- FP64: 9.7 TFLOPS
- FP64 Tensor Core: 19.5 TFLOPS
- FP32: 19.5 TFLOPS
- TF32 Tensor Core: 156 TFLOPS, 312 TFLOPS with sparsity
- BFLOAT16 and FP16 Tensor Core: 312 TFLOPS, 624 TFLOPS with sparsity
- INT8 Tensor Core: 624 TOPS, 1248 TOPS with sparsity
- Multi-Instance GPU: up to 7 instances at 10GB each
- Interconnect: NVLink bridge for 2 GPUs at 600GB/s
- Slot type in X440p: Riser A (Riser 1A and 2A, Gen4)
- Maximum quantity per X440p PCIe Node: 2
- X440p operating temperature: 10 to 35 C
- X440p operating humidity: 5% to 93% non-condensing
- X440p operating altitude: 0 to 3000 m

FAQs

Q: Does the GPU need a licence?
A: No licence is needed for the card itself. The host server is managed through Cisco Intersight in Intersight Managed Mode, which needs at least an Essentials licence per server. NVIDIA vGPU or AI Enterprise software, if used, is licensed separately.

Q: Which Cisco node does this GPU fit?
A: It fits the UCS X440p PCIe Node (UCSX-440P) using Riser A (UCSX-RIS-A-440P), with a maximum of 2 per node. The related UCSX-GPUA100-80-D is the version for the UCSX-440P-D.

Q: What compute nodes and fabric does it work with?
A: The X440p pairs with UCS X210c M6/M7 and X410c M7 compute nodes in a UCS X9508 chassis. It also needs UCS X9416 X-Fabric modules and UCS 6454, 64108 or 6536 Fabric Interconnects.

Q: How is the card cooled and powered?
A: It is passively cooled with a 300W maximum TDP, so it depends on chassis airflow. The X440p riser provides the power cabling for supported GPUs, and the spare cable PID is UCS-P100CBL-240M5=.

Q: How does this SKU differ from the -D variant?
A: UCSX-GPU-A100-80 is the Riser A (Gen4) dual-slot card for the standard X440p node. The -D variant is intended for the UCSX-440P-D used with M7 servers.

  • 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