Skip to product information
1 of 1

Cisco

Cisco UCSC-GPU-A30-D | NVIDIA A30 Tensor Core GPU, 24GB HBM2, 180W, Passive

SKU:UCSC-GPU-A30-D

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 We do not quote Gmail, Outlook or Hotmail addresses, or supply for home use. 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

This NVIDIA A30 Tensor Core GPU accelerator is the Cisco-orderable card for UCS C-Series rack servers. It is a double-wide, passively cooled PCIe Gen4 card aimed at AI inference and training, HPC and mainstream data-centre acceleration. Multi-Instance GPU partitioning allows several workloads to share a single card.

The card carries 24GB of HBM2 memory and has a 180W maximum power draw. Peak performance is rated at 5.2 TFLOPS FP64 and 10.3 TFLOPS FP64 Tensor Core. Third-generation Tensor Cores and structural sparsity support accelerate AI networks, and two A30 cards can be joined with an optional NVLink bridge. Because it is passively cooled, it relies on the host server airflow and is powered through the GPU power cable connected to the PCIe riser.

It is supported in Cisco UCS C-Series servers including the C240 M6, C245 M6 and C240 M7, and needs an x16 slot. A C245 M6 supports up to 3 cards. Cisco IMC and BIOS must be at or above the minimum version for the host server, and the BIOS setting Memory Mapped I/O Above 4 GB must be enabled.

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

Features

- Multi-Instance GPU (MIG) partitions one A30 into several isolated GPU instances
- 24GB of high-bandwidth HBM2 memory
- Third-generation Tensor Cores for AI training and inference
- Structural sparsity support to accelerate AI networks
- Optional NVLink bridge links two A30 PCIe GPUs
- Up to 10.3 TFLOPS of HPC performance

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 A30 Tensor Core
- Memory: 24GB HBM2
- Maximum power: 180W
- Cooling: Passive
- Interface: PCIe Gen4, x16 slot
- Form factor: Double-wide
- Peak FP64: 5.2 TFLOPS
- Peak FP64 Tensor Core: 10.3 TFLOPS
- Multi-Instance GPU (MIG): Supported
- NVLink: Optional bridge between two A30 GPUs
- Supported GPUs per C245 M6 server: 3
- Minimum Cisco IMC/BIOS on C245 M6: 4.2(2f)
- Minimum Cisco IMC/BIOS on C240 M7: 4.1(3)
- Host BIOS requirement: Memory Mapped I/O Above 4 GB enabled

FAQs

Q: Does the A30 need a licence or subscription?
A: No licence is required for the card itself. The host server must run Cisco IMC and UCSM firmware at or above the minimum version for the server model.

Q: How is the card powered?
A: It is a 180W passive card powered through the GPU power cable, which connects the card's black connector to the white GPU POWER connector on the PCIe riser.

Q: Which servers does it work with?
A: It is supported in Cisco UCS C-Series rack servers including the C240 M6, C245 M6 (IMC 4.2(2f) or later) and C240 M7 (IMC 4.1(3) or later). It needs an x16 slot, and a C245 M6 supports up to 3 cards.

Q: Can I mix it with other GPU models in the same server?
A: No. Do not mix different GPU brands or models in one server. Cisco GPUs carry a unique SBIOS ID that CIMC and UCSM rely on.

Q: Does the host need any BIOS changes?
A: Yes. The BIOS setting Memory Mapped I/O Above 4 GB must be set to Enabled.

Q: Can two cards be linked together?
A: Yes. Two A30 GPUs can be connected with an optional NVLink bridge.

  • 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