552 Search Results for "NVIDIA ConnectX-7"
-
Vendor:NVIDIA
NVIDIA 945-13730-0057-000 | ARM Cortex-A78AE development board; safety-ready CPU cores; embedded prototyping platform
945-13730-0057-000
In stock
Vendor:NVIDIANVIDIA 945-13730-0057-000 | ARM Cortex-A78AE development board; safety-ready CPU cores; embedded prototyping platform
Development board built around Arm Cortex-A78AE CPU core architecture, suitable for functional safety–focused embedded evaluation and prototyping. -
Vendor:NVIDIA
NVIDIA 980-9I71G-00J000 | QSFP+ to SFP+ adapter (QSA), enables 10Gb SFP+ in QSFP+ port
980-9I71G-00J000
Stock Status: Enquire
Vendor:NVIDIANVIDIA 980-9I71G-00J000 | QSFP+ to SFP+ adapter (QSA), enables 10Gb SFP+ in QSFP+ port
Compact QSA adapter that converts a QSFP+ cage to accept standard SFP+ modules, enabling 10Gb connectivity on devices equipped with QSFP+ ports. Ideal for mixed-speed migrations and sparing flexibility without replacing hardware. -
Vendor:NVIDIA
NVIDIA 900-2G179-0020-101 | A2 Tensor Core 16GB GPU, PCIe 4.0 x8, 40-60W TDP
900-2G179-0020-101
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-2G179-0020-101 | A2 Tensor Core 16GB GPU, PCIe 4.0 x8, 40-60W TDP
The NVIDIA A2 Tensor Core GPU delivers entry-level AI inference acceleration with exceptional power efficiency for edge deployments and compact server environments. Built on the Ampere architecture with 16GB GDDR6 memory and 1280 CUDA cores, it provides 4.5 TFLOPS FP32 performance and up to 72 TOPS INT4 tensor performance in a single-slot, low-profile form factor. Ideal for intelligent video analytics, machine learning inference, and virtual desktop infrastructure workloads where space and power constraints are critical. -
Vendor:NVIDIA
NVIDIA MGA100-HS2 | IB↔Ethernet gateway, 8x HDR IB 200G + 8x 200GbE, 2U, 8x ConnectX-6 dual‑port, 2x AC PSUs
MGA100-HS2
Stock Status: Enquire
Vendor:NVIDIANVIDIA MGA100-HS2 | IB↔Ethernet gateway, 8x HDR IB 200G + 8x 200GbE, 2U, 8x ConnectX-6 dual‑port, 2x AC PSUs
Skyway is a high‑throughput InfiniBand‑to‑Ethernet gateway appliance for bridging HDR (200Gb/s) IB fabrics to 200GbE networks. This 2U system integrates eight dual‑port ConnectX‑6 adapters to deliver eight 200G HDR IB ports and eight 200GbE ports with low‑latency translation for HPC and AI workloads. Dual hot‑swap AC PSUs provide power redundancy, and QSFP56 ports support 200G optics or DAC/AOC cabling for flexible deployments. -
Vendor:NVIDIA
NVIDIA 920-9N201-00F7-0C0 | MSN3700-CS2FC 32x 100GbE QSFP28, license: Linux (Cumulus)
920-9N201-00F7-0C0
Stock Status: Enquire
Vendor:NVIDIANVIDIA 920-9N201-00F7-0C0 | MSN3700-CS2FC 32x 100GbE QSFP28, license: Linux (Cumulus)
High‑density 1U 100GbE leaf/spine switch delivering wire‑rate performance on 32 QSFP28 ports. Ideal for spine or leaf roles in EVPN‑VXLAN fabrics and HPC/AI clusters requiring predictable, ultra‑low latency and RoCEv2‑friendly QoS. Runs a Linux‑based NOS for feature‑rich L2/L3, automation, and advanced telemetry. -
Vendor:NVIDIA
NVIDIA MSN2010-CB2FC | 18x 10/25G SFP28 | non-PoE | 4x 100G QSFP28 uplinks | license: N/A
MSN2010-CB2FC
Stock Status: Enquire
Vendor:NVIDIANVIDIA MSN2010-CB2FC | 18x 10/25G SFP28 | non-PoE | 4x 100G QSFP28 uplinks | license: N/A
Ultra‑low‑latency, fixed 1U switch for top‑of‑rack and compact aggregation, with 18x 10/25G SFP28 and 4x 100G QSFP28 uplinks. Based on Spectrum silicon for line‑rate L2/L3, VXLAN/EVPN (with compatible NOS), and open networking via ONIE, supporting Cumulus Linux or MLNX‑OS/Onyx. Ideal for modern leaf/ToR deployments and spine connectivity. -
Vendor:NVIDIA
NVIDIA 900-54935-2500-000 | NVLink bridge, 3-slot spacing, RTX 5000 compatible, NVLink 2.0, up to 50 Gb/s aggregate
900-54935-2500-000
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-54935-2500-000 | NVLink bridge, 3-slot spacing, RTX 5000 compatible, NVLink 2.0, up to 50 Gb/s aggregate
High-bandwidth NVLink 2.0 bridge engineered to connect two RTX 5000 workstation GPUs with 3-slot spacing, enabling larger combined frame buffers and higher inter-GPU bandwidth for demanding visualization and compute workloads. Confirm your motherboard provides 3-slot center-to-center spacing between the two PCIe x16 slots housing the GPUs. -
Vendor:NVIDIA
NVIDIA 900-53651-2500-000 | NVLink Bridge 2-Slot for Ampere GPUs
900-53651-2500-000
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-53651-2500-000 | NVLink Bridge 2-Slot for Ampere GPUs
The NVIDIA NVLink Bridge 2-Slot enables high-speed GPU-to-GPU interconnect for dual NVIDIA RTX A6000, A5500, A5000, and A4500 graphics cards. This bridge connects two adjacent GPUs with 2-slot spacing, delivering significantly faster data transfer than traditional PCIe solutions. Ideal for workstations requiring expanded memory capacity and enhanced performance for demanding visual computing, AI, and rendering workloads. -
Vendor:Cisco
Cisco C885A-M8-CC-SLD20 | UCS C885A M8 compute sled | AMD EPYC 9575F | 96GB DDR5-6400 ECC | ConnectX-7 (CX7) networking
C885A-M8-CC-SLD20
Stock Status: Enquire
Vendor:CiscoCisco C885A-M8-CC-SLD20 | UCS C885A M8 compute sled | AMD EPYC 9575F | 96GB DDR5-6400 ECC | ConnectX-7 (CX7) networking
Modular compute sled for UCS C885A M8 platforms featuring an AMD EPYC 9575F CPU, 96GB of DDR5-6400 ECC memory, and ConnectX-7 networking readiness for high-performance fabrics. Ideal for latency-sensitive virtualization and data-intensive workloads. -
Vendor:NVIDIA
NVIDIA MCP7Y00-N01A | IB NDR 800G OSFP-to-2xOSFP passive DAC splitter, 2x400G breakout, 1.5m reach
MCP7Y00-N01A
Stock Status: Enquire
Vendor:NVIDIANVIDIA MCP7Y00-N01A | IB NDR 800G OSFP-to-2xOSFP passive DAC splitter, 2x400G breakout, 1.5m reach
Passive direct-attach copper Y-splitter enabling one 800G OSFP twin-port to break out into two 400G OSFP links. Optimized for very short-reach InfiniBand NDR connections with minimal latency and power versus optics. Ideal for top-of-rack to server or intra-rack switch links. -
Vendor:NVIDIA
NVIDIA 900-2G147-0000-000 | RTX PRO 4500 Blackwell Server GPU, 32GB GDDR7, 165W
900-2G147-0000-000
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-2G147-0000-000 | RTX PRO 4500 Blackwell Server GPU, 32GB GDDR7, 165W
The NVIDIA RTX PRO 4500 Blackwell Server Edition is a power-efficient single-slot PCIe Gen5 GPU built on the revolutionary Blackwell architecture for enterprise AI inference, data science, and visual computing workloads. Featuring 32GB of ECC GDDR7 memory, 10,496 CUDA cores, and fifth-generation Tensor Cores in a compact 165W thermal envelope, it delivers breakthrough performance for data center, edge, and cloud deployments. Designed as the successor to the NVIDIA L4, this passively cooled accelerator excels at AI video understanding, vector database acceleration, and multi-workload enterprise applications. -
Vendor:Cisco
Cisco UCSC-885A-M8-M3X6 | Rack server | MI300X GPU | CX7: 8 + 4 ports | 23TB RAM
UCSC-885A-M8-M3X6
Stock Status: Enquire
Vendor:CiscoCisco UCSC-885A-M8-M3X6 | Rack server | MI300X GPU | CX7: 8 + 4 ports | 23TB RAM
Scalable rack-mount compute platform optimized for MI300X-accelerated AI, HPC, and analytics. This configuration delivers 8 + 4 ConnectX-7 ports to support high-bandwidth cluster fabrics and 23TB of system memory for very large models and datasets. Designed for data center reliability with flexible storage and expansion (as configured) and enterprise remote management. -
Vendor:Cisco
Cisco UCSC-885A-M8-M3X4 | Rack server | MI300X GPU | 10x CX7 | 3TB RAM
UCSC-885A-M8-M3X4
Stock Status: Enquire
Vendor:CiscoCisco UCSC-885A-M8-M3X4 | Rack server | MI300X GPU | 10x CX7 | 3TB RAM
High-density rack-mount compute platform designed for MI300X-accelerated AI, HPC, and data analytics. This build features ConnectX-7 networking with 10 ports to enable robust east–west cluster bandwidth and 3TB of system memory for memory-intensive workloads. It offers flexible storage and expansion options (as configured) with enterprise remote management for streamlined operations. -
Vendor:NVIDIA
NVIDIA 920-9B020-00FH-0D0 | UFM 4.0 Appliance | 2U | 2x HDR 200Gb/s IB (ConnectX‑6)
920-9B020-00FH-0D0
Stock Status: Enquire
Vendor:NVIDIANVIDIA 920-9B020-00FH-0D0 | UFM 4.0 Appliance | 2U | 2x HDR 200Gb/s IB (ConnectX‑6)
Turnkey fabric-management and cyber‑analytics platform for InfiniBand environments. Ships as a 2U appliance with dual HDR 200Gb/s adapters, enabling rapid rollout of UFM Enterprise and Cyber‑AI for real‑time telemetry, anomaly detection, and proactive remediation across large fabrics. -
Vendor:Cisco
Cisco UCSX-GPU-L4 | NVIDIA L4 24GB Graphics Card, 70W, PCIe Gen4 x16
UCSX-GPU-L4
Stock Status: Enquire
Vendor:CiscoCisco UCSX-GPU-L4 | NVIDIA L4 24GB Graphics Card, 70W, PCIe Gen4 x16
The Cisco UCSX-GPU-L4 features the NVIDIA L4 Tensor Core GPU with 24GB GDDR6 memory, built on the Ada Lovelace architecture for efficient AI inference, video processing, and virtualization workloads. This single-slot, half-height half-length GPU delivers 485 TOPS INT8 performance at just 70W power consumption, making it ideal for space-constrained UCS X-Series compute nodes. Designed for enterprise data centers and edge deployments requiring accelerated AI, VDI, and graphics in a compact, energy-efficient form factor. -
Vendor:NVIDIA
NVIDIA 900-13701-0050-000 | Jetson AGX Orin Module, 64GB, 275 TOPS, 12-Core CPU
900-13701-0050-000
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-13701-0050-000 | Jetson AGX Orin Module, 64GB, 275 TOPS, 12-Core CPU
The NVIDIA Jetson AGX Orin 64GB Module is a high-performance system-on-module delivering up to 275 TOPS of AI performance for robotics, autonomous machines, and edge AI applications. Featuring a 2048-core Ampere GPU, 12-core Arm Cortex-A78AE CPU, and 64GB LPDDR5 memory, this compact 100x87mm module supports multiple concurrent AI pipelines with power configurable from 15W to 60W. Ideal for demanding workloads including natural language processing, 3D perception, multi-sensor fusion, and vision AI. -
Vendor:Cisco
Cisco UCSX-GPU-T4-MEZZ | UCS X-Series T4 GPU Module, 16GB GDDR6, 75W Mezzanine
UCSX-GPU-T4-MEZZ
Stock Status: Enquire
Vendor:CiscoCisco UCSX-GPU-T4-MEZZ | UCS X-Series T4 GPU Module, 16GB GDDR6, 75W Mezzanine
The Cisco UCSX-GPU-T4-MEZZ is a mezzanine GPU module designed for Cisco UCS X-Series compute nodes, featuring the NVIDIA Tesla T4 with 16GB GDDR6 memory and 2560 CUDA cores. Built on NVIDIA Turing architecture with 320 Tensor Cores, it delivers 8.1 TFLOPS FP32 and 130 TOPS INT8 performance for AI inference, machine learning, and video transcoding workloads. With 75W passive cooling and PCIe Gen3 x8 connectivity, it integrates directly into UCS X210c M6 servers for scalable GPU acceleration in enterprise data centers. -
Vendor:NVIDIA
NVIDIA 920-9N110-00F1-0C0 | 48x 1G RJ45 access + 4x 100G QSFP28 uplinks | Non‑PoE | 1U | Cumulus Linux
920-9N110-00F1-0C0
Stock Status: Enquire
Vendor:NVIDIANVIDIA 920-9N110-00F1-0C0 | 48x 1G RJ45 access + 4x 100G QSFP28 uplinks | Non‑PoE | 1U | Cumulus Linux
Open Ethernet 1G/100G switch for campus/edge aggregation and server access. Provides 48x 10/100/1000BASE‑T access ports with 4x 100GbE QSFP28 uplinks, powered by NVIDIA Spectrum switching and an x86 control plane, shipping with Cumulus Linux for modern L2/L3/EVPN fabrics. Dual hot‑swap PSUs for resilience. -
Vendor:NVIDIA
NVIDIA 980-9I510-00NS00 | 800G OSFP twin-port | 2x NDR (IB 400G/port) | MMF 850nm | 2x MPO-12 APC | 50 m
980-9I510-00NS00
Stock Status: Enquire
Vendor:NVIDIANVIDIA 980-9I510-00NS00 | 800G OSFP twin-port | 2x NDR (IB 400G/port) | MMF 850nm | 2x MPO-12 APC | 50 m
Twin‑port OSFP optical transceiver delivering two independent 400G InfiniBand NDR short‑reach links over OM4 MMF via 2x MPO‑12 APC. Finned thermal design supports dense, high‑power switch and adapter environments; up to 50 m reach per port for leaf–spine and HPC clusters. -
Vendor:NVIDIA
NVIDIA 980-9I08Q-00W020 | QSFP-DD 400G AOC | 8x50G PAM4 | 850nm MMF | 20m reach | QSFP-DD to QSFP-DD
980-9I08Q-00W020
Stock Status: Enquire
Vendor:NVIDIANVIDIA 980-9I08Q-00W020 | QSFP-DD 400G AOC | 8x50G PAM4 | 850nm MMF | 20m reach | QSFP-DD to QSFP-DD
20‑meter 400G active optical cable for extended short‑reach interconnects across rows or between racks. The QSFP‑DD AOC integrates optics and multimode fiber to provide stable 8x50G PAM4 signaling at 850nm, delivering low power, low EMI, and streamlined installation compared with discrete optics plus patch cords. -
Vendor:NVIDIA
NVIDIA 920-9N110-00R1-0C0 | 48x 1G RJ45 access + 4x 100G QSFP28 uplinks | Non‑PoE | 1U | Cumulus Linux (Open Ethernet)
920-9N110-00R1-0C0
Stock Status: Enquire
Vendor:NVIDIANVIDIA 920-9N110-00R1-0C0 | 48x 1G RJ45 access + 4x 100G QSFP28 uplinks | Non‑PoE | 1U | Cumulus Linux (Open Ethernet)
Spectrum‑based 1GBase‑T/100GbE open Ethernet switch with Cumulus Linux, ideal for 1G access or campus aggregation with 100G uplinks to the core. Designed for automated, EVPN‑ready leaf deployments and seamless integration into modern data center or enterprise networks. -
Vendor:NVIDIA
NVIDIA 980-9I510-F4NS00 | 800G OSFP twin-port | 2x 400GbE (SR) | MMF 850nm | 2x MPO-12 APC | 50 m
980-9I510-F4NS00
Stock Status: Enquire
Vendor:NVIDIANVIDIA 980-9I510-F4NS00 | 800G OSFP twin-port | 2x 400GbE (SR) | MMF 850nm | 2x MPO-12 APC | 50 m
Twin‑port OSFP Ethernet transceiver providing two independent 400GbE short‑reach links over OM4 MMF via 2x MPO‑12 APC. Ideal for 800G aggregate per module in dense switches and AI/ML clusters, with finned thermal design and up to 50 m reach per port. -
Vendor:NVIDIA
NVIDIA 900-9X6AF-0016-ST2 | 1x QSFP28 100G InfiniBand EDR | PCIe 4.0 x8 | Low-profile | RDMA, GPUDirect, Adaptive Routing
900-9X6AF-0016-ST2
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X6AF-0016-ST2 | 1x QSFP28 100G InfiniBand EDR | PCIe 4.0 x8 | Low-profile | RDMA, GPUDirect, Adaptive Routing
High-throughput InfiniBand adapter for HPC and AI clusters, providing 100G EDR connectivity with ultra-low latency and advanced congestion control. QSFP28 interface supports short- and long-reach cabling options. PCIe 4.0 x8 host bandwidth and GPUDirect RDMA enable efficient, direct data paths for accelerated compute workflows. Built for dense deployments with a low-profile form factor. -
Vendor:Cisco
Cisco N7K-F348XP-25 | 48x 1/10GbE SFP+ line card | 1-slot | Fabric-2 compatible
N7K-F348XP-25
Stock Status: Enquire
Vendor:CiscoCisco N7K-F348XP-25 | 48x 1/10GbE SFP+ line card | 1-slot | Fabric-2 compatible
High-density 10GbE SFP+ line card with 48 ports for Nexus 7000 Series deployments. Provides line-rate Layer 2/Layer 3 10GbE with support for 1GbE operation using compatible SFP optics. Ideal for top-of-rack aggregation and data center access; occupies a single slot and integrates with Fabric-2 for high backplane bandwidth.
