Skip to product information
1 of 1

Cisco

Cisco HCIX-GPU-A40 | NVIDIA A40 GPU, 48GB GDDR6 ECC, 10752 CUDA Cores, 300W

SKU:HCIX-GPU-A40

Stock Status: Enquire

Request Quote
Sale Sold out
Shipping calculated at checkout.

Description

The Cisco HCIX-GPU-A40 is a professional-grade GPU accelerator built on NVIDIA Ampere architecture, designed for AI workloads, deep learning, data science, and visual computing in enterprise data centers. With 48GB of GDDR6 ECC memory, 10752 CUDA cores, and third-generation Tensor Cores, it delivers exceptional performance for compute-intensive applications. The dual-slot passive cooled design with 300W TDP fits Cisco UCS and HyperFlex platforms, offering PCIe 4.0 connectivity and optional NVLink scaling to 96GB memory.

Features

NVIDIA Ampere architecture with 10,752 CUDA cores for parallel processing
- 48GB GDDR6 ECC memory for data integrity and handling large datasets
- Third-generation Tensor Cores with support for TF32, BF16, FP16, INT8, and INT4 precision
- Second-generation RT Cores for hardware-accelerated ray tracing
- PCIe 4.0 x16 interface with backward compatibility to PCIe 3.0
- NVLink support to connect two GPUs for 96GB combined memory and increased bandwidth
- SR-IOV and vGPU support for virtualization and multi-tenant workloads
- NVIDIA RTX Virtual Workstation (vWS) and Virtual Compute Server (vCS) software compatibility
- Three DisplayPort 1.4a outputs supporting up to 4x 5K or 2x 8K displays
- 300W TDP with passive cooling design optimized for data center efficiency
- Hardware-accelerated video encoding and decoding (H.264, HEVC)
- GPUDirect RDMA and GPUDirect Storage for fast data transfers
- NVIDIA CUDA, cuDNN, TensorRT, and NGC container support
- Dual-slot FHFL form factor compatible with standard server chassis
- Secure boot and attestation capabilities for data center security
- Fine-grained structured sparsity for up to 2X AI inference throughput

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 Ampere (GA102-895)
- CUDA Cores: 10,752
- Tensor Cores: Third-generation with support for TF32, BF16, FP16, INT8, and INT4
- RT Cores: Second-generation for real-time ray tracing
- Memory: 48GB GDDR6 with ECC
- Memory Interface: 384-bit
- Memory Bandwidth: 696 GB/s
- GPU Clock: 1305 MHz base, 1740 MHz boost
- Interface: PCIe 4.0 x16 (backward compatible with PCIe 3.0)
- Form Factor: Dual-slot Full-Height Full-Length (FHFL)
- Power Consumption: 300W TDP
- Power Connector: 8-pin EPS
- Cooling: Passive heatsink (requires server airflow)
- Display Outputs: 3x DisplayPort 1.4a
- NVLink Support: Optional NVLink bridge to connect two A40 GPUs for 96GB total memory
- Max Display Resolution: Up to 4x 5K at 60Hz or 2x 8K at 60Hz with DSC
- Single Precision (FP32): 37.4 TFLOPS
- Tensor Performance (TF32): 149.7 TFLOPS
- Double Precision (FP64): 0.5 TFLOPS
- Virtualization: SR-IOV, vGPU, vPC, and NVIDIA RTX Virtual Workstation (vWS) support
- Compatibility: Cisco UCS C-Series, HyperFlex, and X-Series modular systems with x16 PCIe slots

FAQs

Q: What Cisco platforms support the HCIX-GPU-A40?
A: The HCIX-GPU-A40 is designed for Cisco HyperFlex and UCS platforms, including UCS C-Series rack servers and UCS X-Series modular systems with PCIe expansion nodes. It requires a PCIe 4.0 x16 slot and adequate cooling airflow. Up to three A40 GPUs can be installed in supported servers depending on the configuration.

Q: Does this GPU require active cooling?
A: No, the A40 uses a passive heatsink and relies on server-grade airflow for cooling. It is designed for data center rack environments with sufficient chassis airflow and cannot be used in desktop systems without active cooling solutions.

Q: Can I connect multiple A40 GPUs together?
A: Yes, two A40 GPUs can be connected using an NVLink bridge to scale total GPU memory from 48GB to 96GB and increase GPU-to-GPU bandwidth. NVLink support requires application-level compatibility and appropriate NVLink hardware accessories.

Q: What workloads is the A40 optimized for?
A: The A40 is optimized for AI training and inference, deep learning, data science, visual computing, real-time ray tracing, 3D rendering, virtual workstations, and compute-intensive professional applications. It features ECC memory for mission-critical reliability and supports virtualization for multi-user deployments.

Q: Does the A40 support video outputs for workstation use?
A: Yes, the A40 includes three DisplayPort 1.4a outputs and can drive up to four 5K displays at 60Hz or two 8K displays at 60Hz with Display Stream Compression. It supports HDR color and is suitable for both compute and visual workstation deployments.

Recently Viewed

  • 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