-
Vendor:Cisco
Cisco CAI-GPU-MI210 | Instinct MI210 GPU Accelerator, 64GB HBM2e, 300W, PCIe 4.0
CAI-GPU-MI210
Stock Status: Enquire
Vendor:CiscoCisco CAI-GPU-MI210 | Instinct MI210 GPU Accelerator, 64GB HBM2e, 300W, PCIe 4.0
The AMD Instinct MI210 GPU accelerator delivers exascale-class double-precision compute performance for HPC and AI workloads in data center and rack-mount servers. Featuring 64GB HBM2e memory with 1.6 TB/s bandwidth, 104 compute units, and 45.3 TFLOPS FP64 matrix performance, the MI210 fits standard PCIe Gen4 x16 slots with a 300W TDP. Built on CDNA 2 architecture with three Infinity Fabric links for multi-GPU scalability, it supports ROCm open-source software stack and is optimized for computational fluid dynamics, molecular dynamics, climate modeling, and machine learning training. -
Vendor:Cisco
Cisco CAI-GPU-RKIT-NV | C845A GPU NVLink bridge removal tool kit — NVIDIA variant for dual‑GPU service
CAI-GPU-RKIT-NV
Stock Status: Enquire
Vendor:CiscoCisco CAI-GPU-RKIT-NV | C845A GPU NVLink bridge removal tool kit — NVIDIA variant for dual‑GPU service
Precision tool set designed for safe removal and reinstallation of NVLink bridges on NVIDIA GPU assemblies in C845A servers. Protects connectors and PCBs while ensuring consistent, guided extraction and placement. Ideal for upgrades, replacements, and RMA processes in data center environments. -
Vendor:Cisco
Cisco CAI-GPU-RKIT-AMD | C845A GPU NVLink bridge removal tool kit — AMD variant for dual‑GPU service
CAI-GPU-RKIT-AMD
Stock Status: Enquire
Vendor:CiscoCisco CAI-GPU-RKIT-AMD | C845A GPU NVLink bridge removal tool kit — AMD variant for dual‑GPU service
Purpose-built tool set for safe removal and reinstallation of NVLink bridges on AMD GPU assemblies in C845A servers. Reduces risk of fin damage, PCB stress, and connector misalignment during maintenance. Ideal for field service, upgrades, and RMA workflows where precision and repeatability matter. -
Vendor:Cisco
Cisco CAI-GPU-RTXP6000 | RTX Pro 6000 Graphics Card, 96GB GDDR7, 600W, PCIe 5.0
CAI-GPU-RTXP6000
Stock Status: Enquire
Vendor:CiscoCisco CAI-GPU-RTXP6000 | RTX Pro 6000 Graphics Card, 96GB GDDR7, 600W, PCIe 5.0
The NVIDIA RTX Pro 6000 Blackwell Workstation Edition delivers professional-grade AI and visual computing for workstations and servers. Built on the Blackwell architecture with 96GB GDDR7 ECC memory, 24,064 CUDA cores, and 600W TDP, this dual-slot GPU accelerates AI inference, large language models, 3D rendering, simulation, and data science workloads. Ideal for engineering, content creation, scientific computing, and enterprise AI applications requiring massive memory capacity and proven professional reliability. -
Vendor:Cisco
Cisco CAI-NVL2-H100 | NVLink 2-Way Bridge for H100 NVL GPU Interconnect
CAI-NVL2-H100
Stock Status: Enquire
Vendor:CiscoCisco CAI-NVL2-H100 | NVLink 2-Way Bridge for H100 NVL GPU Interconnect
A high-speed NVLink bridge enabling direct GPU-to-GPU communication between two NVIDIA H100 NVL PCIe cards, delivering 600 GB/s bidirectional bandwidth—10x faster than PCIe Gen4. Designed specifically for AI training, large language model inference, and HPC workloads requiring memory-intensive multi-GPU configurations, this bridge maximizes application performance when deploying H100 NVL pairs in standard PCIe servers. -
Vendor:Cisco
Cisco CAI-INF4-MI210 | MI210 4-Way Infinity Fabric Bridge, 300GB/s Per Link
CAI-INF4-MI210
Stock Status: Enquire
Vendor:CiscoCisco CAI-INF4-MI210 | MI210 4-Way Infinity Fabric Bridge, 300GB/s Per Link
The Cisco CAI-INF4-MI210 is a 4-way Infinity Fabric Link bridge card designed to interconnect four AMD Instinct MI210 GPU accelerators in a high-bandwidth mesh topology. Each Infinity Fabric link delivers 300GB/s of peer-to-peer bandwidth, enabling up to 1.2TB/s total theoretical GPU-to-GPU bandwidth within a server. This bridge card is essential for HPC and AI workloads requiring direct, low-latency communication between multiple MI210 accelerators in dual or quad GPU configurations. -
Vendor:Cisco
Cisco CAI-INF2-MI210 | 2-Way Infinity Fabric Bridge for MI210 GPU, 300 GB/s P2P
CAI-INF2-MI210
Stock Status: Enquire
Vendor:CiscoCisco CAI-INF2-MI210 | 2-Way Infinity Fabric Bridge for MI210 GPU, 300 GB/s P2P
The Cisco CAI-INF2-MI210 is a 2-way Infinity Fabric Link bridge designed specifically for AMD Instinct MI210 GPUs deployed in UCS C845A M8 AI servers and compatible systems. This bridge enables direct GPU-to-GPU peer-to-peer communication at up to 300 GB/s bandwidth, dramatically reducing latency for multi-GPU HPC and AI training workloads. Ideal for organizations scaling MI210 compute capacity in dual-GPU configurations. -
Vendor:Cisco
Cisco CAI-NVL2-H200 | NVLink 2-Way Bridge for H200 NVL GPU, 900GB/s
CAI-NVL2-H200
Stock Status: Enquire
Vendor:CiscoCisco CAI-NVL2-H200 | NVLink 2-Way Bridge for H200 NVL GPU, 900GB/s
The CAI-NVL2-H200 is a 2-way NVLink bridge designed for direct GPU-to-GPU interconnection between NVIDIA H200 NVL GPUs in enterprise server environments. Utilizing 4th generation NVLink technology, it delivers 900GB/s bidirectional bandwidth to scale multi-GPU compute for AI inference, large language model training, and high-performance computing workloads. This bridge enables low-latency, high-throughput communication between paired H200 NVL PCIe GPUs in air-cooled rack configurations. -
Vendor:Cisco
Cisco CAI-P-N3140H | 1x400GbE QSFP112 | PCIe Gen5 x16 | RoCEv2/RDMA, SR-IOV | DOCA/GPUDirect RDMA | NVMe-oF acceleration
CAI-P-N3140H
Stock Status: Enquire
Vendor:CiscoCisco CAI-P-N3140H | 1x400GbE QSFP112 | PCIe Gen5 x16 | RoCEv2/RDMA, SR-IOV | DOCA/GPUDirect RDMA | NVMe-oF acceleration
Ultra-high-performance 400GbE SuperNIC designed for AI, HPC, and cloud-scale fabrics. Delivers line-rate RDMA (RoCEv2), advanced congestion control, and GPU-accelerated networking via GPUDirect and DOCA, enabling low-latency east–west traffic and storage acceleration. -
Vendor:Cisco
Cisco CAI-NVL4-H200 | 4-Way NVLink Bridge for H200 NVL, 900GB/s Interconnect
CAI-NVL4-H200
Stock Status: Enquire
Vendor:CiscoCisco CAI-NVL4-H200 | 4-Way NVLink Bridge for H200 NVL, 900GB/s Interconnect
The CAI-NVL4-H200 is an NVIDIA 4-way NVLink Bridge designed for H200 NVL GPUs, enabling high-bandwidth, low-latency GPU-to-GPU interconnect at 900GB/s bidirectional throughput. This bridge connects up to four H200 NVL accelerators in a single server, providing unified memory pooling and bypassing PCIe Gen5 bottlenecks for optimal multi-GPU performance in AI inference, large language model training, and HPC workloads. Essential for enterprise data centers deploying high-density GPU compute clusters. -
Vendor:Cisco
Cisco CAI-P-N7D200GFO | 2x200GbE QSFP112 | PCIe Gen5 x16 | RoCEv2, SR-IOV, GPUDirect | PTP 1588v2 | NVMe-oF/TCP offload
CAI-P-N7D200GFO
Stock Status: Enquire
Vendor:CiscoCisco CAI-P-N7D200GFO | 2x200GbE QSFP112 | PCIe Gen5 x16 | RoCEv2, SR-IOV, GPUDirect | PTP 1588v2 | NVMe-oF/TCP offload
Dual-port 200GbE PCIe 5.0 adapter with QSFP112 cages for ultra‑low latency networking in AI/ML, HPC, and high‑throughput storage fabrics. Supports advanced offloads including RoCEv2, GPUDirect RDMA, SR‑IOV, and NVMe‑oF/TCP for maximum efficiency and reduced CPU overhead. Backward compatible with lower link rates using supported QSFP-class optics and DAC/AOC cables. -
Vendor:Cisco
Cisco CAI-P-N7S400GFO | 1x400GbE QSFP112 | PCIe Gen5 x16 | RoCEv2, SR-IOV, GPUDirect | PTP 1588v2 | NVMe-oF/TCP offload
CAI-P-N7S400GFO
Stock Status: Enquire
Vendor:CiscoCisco CAI-P-N7S400GFO | 1x400GbE QSFP112 | PCIe Gen5 x16 | RoCEv2, SR-IOV, GPUDirect | PTP 1588v2 | NVMe-oF/TCP offload
Single-port 400GbE PCIe 5.0 network adapter built on QSFP112 for maximum throughput in data center, AI/ML, and HPC deployments. Delivers advanced acceleration with RoCEv2, GPUDirect RDMA, SR‑IOV, and NVMe‑oF/TCP to minimize latency and CPU load. Supports lower-rate operation and breakout modes where enabled by firmware and optics. -
Vendor:Cisco
Cisco HCI-ADGPU-240M6 | GPU air duct | fits C240 M6 2U SFF NVMe | optimized GPU cooling | tool-less fitment
HCI-ADGPU-240M6
Stock Status: Enquire
Vendor:CiscoCisco HCI-ADGPU-240M6 | GPU air duct | fits C240 M6 2U SFF NVMe | optimized GPU cooling | tool-less fitment
Directed airflow shroud for C240 M6 2U SFF NVMe configurations with GPUs, channeling intake air across accelerators to maintain thermal stability. Intended for specific DW FL configuration use where applicable. -
Vendor:Cisco
Cisco HCI-GPU-A100-80M6 | A100 80GB GPU Accelerator, 300W PCIe 4.0, Passive
HCI-GPU-A100-80M6
Stock Status: Enquire
Vendor:CiscoCisco HCI-GPU-A100-80M6 | A100 80GB GPU Accelerator, 300W PCIe 4.0, Passive
The Cisco HCI-GPU-A100-80M6 is an NVIDIA A100 Tensor Core GPU accelerator with 80GB HBM2e memory, designed for demanding AI training, deep learning inference, and high-performance computing workloads in data center environments. Built on NVIDIA's Ampere architecture, it delivers 312 TFLOPS FP16 performance and 2,039 GB/s memory bandwidth, with Multi-Instance GPU capability supporting up to seven isolated GPU instances. The passively cooled, dual-slot PCIe 4.0 x16 form factor integrates into standard Cisco UCS rack servers for scalable compute acceleration. -
Vendor:Cisco
Cisco HCI-GPU-A100-80 | A100 80GB PCIe GPU, 300W Passive, MIG-Enabled, HBM2e
HCI-GPU-A100-80
Stock Status: Enquire
Vendor:CiscoCisco HCI-GPU-A100-80 | A100 80GB PCIe GPU, 300W Passive, MIG-Enabled, HBM2e
The Cisco HCI-GPU-A100-80 is an enterprise-grade compute accelerator designed for AI training, deep learning inference, and high-performance computing workloads in data center environments. Built on NVIDIA Ampere architecture with 80GB HBM2e memory and 6,912 CUDA cores, it delivers 19.5 TFLOPS FP32 performance and up to 312 TFLOPS Tensor Core acceleration. Multi-Instance GPU technology enables partitioning into up to seven isolated instances for maximized resource utilization across mixed workloads. -
Vendor:Cisco
Cisco HCI-GPU-A10-M6 | NVIDIA A10 GPU, 24GB GDDR6, PCIe Gen4, 150W Passive
HCI-GPU-A10-M6
Stock Status: Enquire
Vendor:CiscoCisco HCI-GPU-A10-M6 | NVIDIA A10 GPU, 24GB GDDR6, PCIe Gen4, 150W Passive
The Cisco HCI-GPU-A10-M6 is a data center GPU module featuring the NVIDIA A10 Tensor Core GPU with 24GB GDDR6 memory and 150W passive cooling. Built on NVIDIA Ampere architecture, it delivers powerful AI inference, rendering, and virtual workstation performance in a single-slot PCIe Gen4 form factor. Ideal for enterprise servers running virtualization, graphics-intensive VDI, and compute workloads in space-constrained data center environments.
Why Choose XS Network Tech?
We are a trusted partner for businesses of all sizes, including government departments, data centres, and service providers. Our expert team in sales, logistics, and customer service ensures that your needs are met with speed and precision, from pre-sales consultation to order fulfilment. With a commitment to fast delivery, sustainability, and a wide inventory, XS Network Tech is your go-to source for IT hardware. Trust us for quality, fast responses, and a hassle-free experience every time.