Skip to product information
1 of 1

Cisco

Cisco UCSC-GPU-L40S | UCS C-Series NVIDIA L40S 48GB GPU, 350W, PCIe Gen4

SKU:UCSC-GPU-L40S

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 We do not quote Gmail, Outlook or Hotmail addresses, or supply for home use. 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 UCSC-GPU-L40S is an NVIDIA L40S data-centre GPU accelerator for Cisco UCS C-Series rack servers. Built on the NVIDIA Ada Lovelace architecture, it is aimed at generative AI inference and training, large language model workloads, 3D graphics and rendering, video and media processing, virtual workstations, VDI, and HPC or data analytics in enterprise data centres.

The card is a passively cooled, dual-slot, full-height full-length PCIe Gen4 x16 board with 48GB of ECC GDDR6 and 864GB/s of memory bandwidth. It carries 18,176 CUDA cores, 568 fourth-generation Tensor cores and 142 third-generation RT cores, and draws a maximum of 350W through a 16-pin power connector. Passive cooling relies on server chassis airflow. No licence is needed for the hardware itself, while NVIDIA vGPU or virtual workstation use requires separate NVIDIA vGPU software licences.

This is the C-Series part, distinct from the X-Series UCSX-GPU-L40S used in the UCS X440p PCIe node. For C-Series servers such as the UCS C245 M8, up to three cards are supported per server, with a minimum Cisco IMC and BIOS release of 4.3(1). Check the individual server specification sheet for riser and slot support.

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

Features

- Up to 5X higher inference performance than the previous-generation NVIDIA A40 for generative AI
- Third-generation RT Cores and 48GB GDDR6 for real-time ray tracing and graphics
- Fourth-generation Tensor Cores with FP8 Transformer Engine support
- vGPU software support for RTX Virtual Workstation, Virtual PC and Virtual Compute Server
- 3x NVENC and 3x NVDEC engines including AV1 encode and decode
- Secure boot with hardware root of trust
- NEBS Level 3 ready

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 architecture: NVIDIA Ada Lovelace
- GPU memory: 48GB GDDR6 with ECC
- Memory bandwidth: 864GB/s
- Interface: PCIe Gen4 x16, 64GB/s bidirectional
- CUDA cores: 18,176
- Fourth-generation Tensor cores: 568
- Third-generation RT cores: 142
- RT core performance: 212 TFLOPS
- FP32: 91.6 TFLOPS
- TF32 Tensor Core: 183 | 366 TFLOPS
- BFLOAT16 Tensor Core: 362.05 | 733 TFLOPS
- FP16 Tensor Core: 362.05 | 733 TFLOPS
- FP8 Tensor Core: 733 | 1,466 TFLOPS
- Peak INT8 Tensor: 733 | 1,466 TOPS
- Peak INT4 Tensor: 733 | 1,466 TOPS
- Form factor: 4.4" (H) x 10.5" (L), dual slot, full-height full-length
- Display ports: 4x DisplayPort 1.4a
- Maximum power consumption: 350W
- Power connector: 16-pin
- Cooling: Passive
- NVLink: Not supported
- MIG: Not supported
- With sparsity

FAQs

Q: Does this GPU need a licence?
A: No licence is needed for the hardware itself. If you run NVIDIA vGPU or virtual workstation workloads, separate NVIDIA vGPU software licences are required.

Q: What power does the card require?
A: It draws up to 350W through a 16-pin power connector. The server must supply the GPU power cable and have enough PSU capacity for the card.

Q: How is it cooled?
A: The card is passively cooled, so it depends on airflow from the server chassis fans. It must be installed in a server designed to support passive dual-slot GPUs.

Q: Which servers does it work with?
A: It is the C-Series part for UCS rack servers. Cisco lists support in the UCS C245 M8, up to 3 per server with a minimum Cisco IMC and BIOS release of 4.3(1), and positions L40S for the UCS C845A M8.

Q: How does it differ from UCSX-GPU-L40S?
A: UCSC-GPU-L40S is for UCS C-Series rack servers. UCSX-GPU-L40S is a different part number for the UCS X440p PCIe node in the X-Series.

  • 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