552 Search Results for "NVIDIA ConnectX-7"
-
Vendor:NVIDIA
NVIDIA 900-9X720-007N-SN1 | 4x 400G Eth/IB VPI, PCIe Gen5 x32, RDMA RoCEv2, SR-IOV, GPUDirect, PTP
900-9X720-007N-SN1
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X720-007N-SN1 | 4x 400G Eth/IB VPI, PCIe Gen5 x32, RDMA RoCEv2, SR-IOV, GPUDirect, PTP
Quad-port VPI adapter delivering up to 4x 400GbE or 400Gb/s InfiniBand for AI/HPC fabrics, accelerated storage, and ultra‑low‑latency cloud. PCIe Gen5 x32 provides massive host bandwidth with advanced offloads: RDMA (RoCE v2), GPUDirect RDMA, NVMe‑oF, inline IPsec/TLS, and IEEE 1588 PTP hardware timestamping for precision timing. -
Vendor:NVIDIA
NVIDIA MCX623105AN-VDAT | Mellanox Technologies MCX623105AN-VDAT network card Internal Fiber 200000 Mbit/s
MCX623105AN-VDAT
Stock Status: Enquire
Vendor:NVIDIANVIDIA MCX623105AN-VDAT | Mellanox Technologies MCX623105AN-VDAT network card Internal Fiber 200000 Mbit/s
NVIDIA CONNECTX-6 DX, 200GBE , 1xQSFP56, PCIE 4X16, TALL BRACKET -
Vendor:NVIDIA
NVIDIA 900-9X658-0046-SI0 | ConnectX-6 Dx 100GbE 1x QSFP56 PCIe 4.0 x16 NIC, RoCEv2, SR-IOV, IPsec/TLS offload
900-9X658-0046-SI0
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X658-0046-SI0 | ConnectX-6 Dx 100GbE 1x QSFP56 PCIe 4.0 x16 NIC, RoCEv2, SR-IOV, IPsec/TLS offload
High-performance single-port 100GbE adapter based on ConnectX-6 Dx, designed for low-latency workloads, virtualization, and secure data-in-motion with inline IPsec/TLS offload. QSFP56 cage supports 100G with compatibility for QSFP28 optics and DAC/AOC for 25/50/100G links. Includes internal lock/retainer bracket for secure in-chassis installation. -
Vendor:NVIDIA
NVIDIA 900-5G147-2550-000 | RTX PRO 4500 32GB GDDR7 Graphics Card, 200W, Dual-Slot
900-5G147-2550-000
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-5G147-2550-000 | RTX PRO 4500 32GB GDDR7 Graphics Card, 200W, Dual-Slot
The NVIDIA RTX PRO 4500 Blackwell is a professional workstation graphics card engineered for demanding AI, rendering, and design workflows. Built on the Blackwell architecture with 32GB of GDDR7 ECC memory and 896 GB/s bandwidth, it accelerates complex 3D modeling, generative AI applications, ray tracing, and neural rendering. Featuring 10,496 CUDA cores, 328 fifth-generation Tensor Cores, and 82 fourth-generation RT Cores, this dual-slot card delivers exceptional performance for engineering simulations, data science, content creation, and large-scale visualization projects. -
Vendor:NVIDIA
NVIDIA 900-5G195-2501-000 | RTX Pro 4000 SFF 24GB GDDR7 Graphics Card, 70W, PCIe 5.0
900-5G195-2501-000
In stock
Vendor:NVIDIANVIDIA 900-5G195-2501-000 | RTX Pro 4000 SFF 24GB GDDR7 Graphics Card, 70W, PCIe 5.0
The NVIDIA RTX Pro 4000 SFF Blackwell is a professional low-profile workstation graphics card designed for compact AI workstations and small form factor systems. Powered by Blackwell architecture with 24GB GDDR7 ECC memory, 8,960 CUDA cores, and 280 fifth-generation Tensor Cores, it delivers breakthrough AI performance up to 770 AI TOPS while maintaining a 70W thermal envelope. With PCIe slot-only power and a dual-slot low-profile design, it's ideal for space-constrained professional environments running CAD, 3D rendering, AI inference, and scientific visualization workloads. -
Vendor:NVIDIA
NVIDIA MUA9602H-2SR | UFM 3.0 appliance | 1U | 2x ConnectX-6 HDR dual-port 200Gb/s InfiniBand adapters (4x HDR 200G ports total) | Telemetry/Enterprise software installed
MUA9602H-2SR
Stock Status: Enquire
Vendor:NVIDIANVIDIA MUA9602H-2SR | UFM 3.0 appliance | 1U | 2x ConnectX-6 HDR dual-port 200Gb/s InfiniBand adapters (4x HDR 200G ports total) | Telemetry/Enterprise software installed
Purpose-built UFM 3.0 appliance for managing modern InfiniBand fabrics. This 1U platform ships with UFM Telemetry or UFM Enterprise software installed and two ConnectX-6 HDR 200Gb/s dual-port adapters, delivering four HDR 200G ports for high-throughput fabric visibility and control. Ideal for cluster provisioning, monitoring, and automated operations at scale. -
Vendor:NVIDIA
NVIDIA 900-5G195-2551-000 | RTX PRO 2000 Graphics Card, 16GB GDDR7 ECC, PCIe 5.0 x8
900-5G195-2551-000
In stock
Vendor:NVIDIANVIDIA 900-5G195-2551-000 | RTX PRO 2000 Graphics Card, 16GB GDDR7 ECC, PCIe 5.0 x8
The NVIDIA RTX PRO 2000 Blackwell is a compact professional graphics card built on the advanced Blackwell architecture, delivering exceptional AI and visualization performance in a power-efficient 70W design. Featuring 16GB of ECC GDDR7 memory, 4,352 CUDA cores, 5th Gen Tensor Cores, and 4th Gen Ray Tracing Cores, it accelerates CAD, 3D modeling, AI workflows, and neural rendering for professional workstations. The dual-slot, low-profile form factor with four mini DisplayPort 2.1b outputs makes it ideal for compact workstations requiring multi-display support and enterprise-grade reliability. -
Vendor:NVIDIA
NVIDIA 900-5G147-2570-000 | RTX PRO 4000 Blackwell 24GB GDDR7 Graphics Card
900-5G147-2570-000
In stock
Vendor:NVIDIANVIDIA 900-5G147-2570-000 | RTX PRO 4000 Blackwell 24GB GDDR7 Graphics Card
The NVIDIA RTX PRO 4000 Blackwell is a professional workstation graphics card built on the revolutionary Blackwell architecture. Featuring 24GB of ultra-fast GDDR7 ECC memory with 672GB/s bandwidth and 8,960 CUDA cores, it delivers breakthrough AI performance, neural rendering, and power efficiency in a single-slot design. Ideal for demanding CAD, visualization, AI development, and content creation workflows requiring photorealistic ray tracing and accelerated generative AI capabilities. -
Vendor:NVIDIA
NVIDIA 900-9X6AF-0018-MT2 | ConnectX-6 Single-Port 200Gb HDR IB/Ethernet Adapter
900-9X6AF-0018-MT2
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X6AF-0018-MT2 | ConnectX-6 Single-Port 200Gb HDR IB/Ethernet Adapter
The NVIDIA ConnectX-6 VPI adapter card delivers 200Gb/s InfiniBand HDR and Ethernet connectivity in a single-port QSFP56 form factor with PCIe 4.0 x16 host interface. Engineered for high-performance computing, AI/ML workloads, and data center applications, this adapter provides sub-600ns latency, 215 million messages per second, and supports advanced features including RDMA, RoCEv2, GPUDirect, and PTP hardware timestamping. The Virtual Protocol Interconnect technology enables flexible protocol switching between InfiniBand and Ethernet on a single adapter. -
Vendor:NVIDIA
NVIDIA 900-9X7AH-0078-DTZ | 200GbE 1x QSFP56, PCIe 5.0 x16, HHHL
900-9X7AH-0078-DTZ
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X7AH-0078-DTZ | 200GbE 1x QSFP56, PCIe 5.0 x16, HHHL
Enterprise‑grade 200GbE adapter delivering exceptional throughput and latency performance for modern AI/ML, virtualization, and NVMe‑oF storage deployments. Rich hardware offloads reduce CPU overhead and optimize east‑west traffic at scale. -
Vendor:NVIDIA
NVIDIA MCP7Y70-H002 | Speed: 400G→4x100G | Form: OSFP to 4xQSFP56 | Standard: InfiniBand HDR | Wavelength: n/a (passive copper) | Reach: 2m | Connector: OSFP<>QSFP56
MCP7Y70-H002
Stock Status: Enquire
Vendor:NVIDIANVIDIA MCP7Y70-H002 | Speed: 400G→4x100G | Form: OSFP to 4xQSFP56 | Standard: InfiniBand HDR | Wavelength: n/a (passive copper) | Reach: 2m | Connector: OSFP<>QSFP56
Passive copper splitter DAC to fan out a 400G HDR OSFP host into four 100G QSFP56 endpoints over a 2m reach. Delivers ultra-low latency and simple, power-free connectivity for short-distance HPC and AI clusters. -
Vendor:NVIDIA
NVIDIA 900-9X657-0058-SB0 | 1x200G QSFP56 (Ethernet/InfiniBand), PCIe Gen4 x16
900-9X657-0058-SB0
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X657-0058-SB0 | 1x200G QSFP56 (Ethernet/InfiniBand), PCIe Gen4 x16
High-speed adapter based on ConnectX-6, offering up to 200Gb/s via a QSFP56 port for next‑gen data center fabrics. Supports 200G PAM4 signaling and is suitable for dense virtualization, storage, and HPC deployments. PCIe Gen4 x16 host interface maximizes throughput between server and network. Pair with QSFP56 200G optics, AOCs, or DACs for the required reach. -
Vendor:NVIDIA
NVIDIA 900-5G153-2500-000 | RTX PRO 6000 Blackwell Max-Q Graphics Card, 96GB GDDR7
900-5G153-2500-000
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-5G153-2500-000 | RTX PRO 6000 Blackwell Max-Q Graphics Card, 96GB GDDR7
The NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition delivers breakthrough performance for AI, data science, and professional graphics workflows. Built on the Blackwell architecture with 24,064 CUDA cores and 96GB of ultra-fast GDDR7 ECC memory, this density-optimized GPU provides 1,792 GB/s memory bandwidth and PCIe 5.0 x16 connectivity. Designed for demanding workstation environments requiring advanced AI processing, real-time rendering, and complex simulations in a compact dual-slot form factor with 300W power consumption. -
Vendor:NVIDIA
NVIDIA 900-2G153-0000-000 | RTX PRO 6000 96GB GDDR7 Graphics Card, PCIe 5.0, 600W
900-2G153-0000-000
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-2G153-0000-000 | RTX PRO 6000 96GB GDDR7 Graphics Card, PCIe 5.0, 600W
The NVIDIA RTX PRO 6000 Blackwell Server Edition delivers enterprise-grade AI and graphics acceleration for data center deployments. Built on the Blackwell architecture with 24,064 CUDA cores and 96GB of GDDR7 ECC memory, this passively-cooled GPU excels at large language model inference, agentic AI, engineering simulation, 3D rendering, and scientific computing workloads. The dual-slot PCIe 5.0 form factor integrates into rack-mounted servers with four DisplayPort 2.1 outputs for visualization. -
Vendor:Cisco
Cisco UCSC-885A-M8-M354 | Rack server | 8x MI350X GPUs | 2x ConnectX-7 | 3TB RAM
UCSC-885A-M8-M354
Stock Status: Enquire
Vendor:CiscoCisco UCSC-885A-M8-M354 | Rack server | 8x MI350X GPUs | 2x ConnectX-7 | 3TB RAM
Performance‑tuned 8‑GPU rack server for AI training and inference with MI350X accelerators, ConnectX‑7 networking, and a balanced 3TB memory configuration for cost‑efficient scaling. Ideal for pilot to production AI clusters. -
Vendor:Cisco
Cisco UCSC-885L-M8-M351 | GPU server | 8x MI355X accelerators | 2x ConnectX-7 | 3TB memory | 8x E1.S NVMe
UCSC-885L-M8-M351
Stock Status: Enquire
Vendor:CiscoCisco UCSC-885L-M8-M351 | GPU server | 8x MI355X accelerators | 2x ConnectX-7 | 3TB memory | 8x E1.S NVMe
High-density AI/compute platform preconfigured with 8x MI355X accelerator modules, dual ConnectX-7 adapters for high-throughput fabrics, 3TB system memory, and 8x E1.S NVMe bays for fast local storage. Ideal for training, inference, and HPC workloads where dense GPU, fast networking, and serviceable E1.S media are priorities. -
Vendor:NVIDIA
NVIDIA MCX631102AN-ADAT | Mellanox Technologies MCX631102AN-ADAT network card Internal Fiber 25000 Mbit/s
MCX631102AN-ADAT
Stock Status: Enquire
Vendor:NVIDIANVIDIA MCX631102AN-ADAT | Mellanox Technologies MCX631102AN-ADAT network card Internal Fiber 25000 Mbit/s
NIVIDIA CONNECTX-6 LX EN ADAPTER CARD, 25GBE, DUAL-PORT SFP28, PCIE 4.0 X8, NO CRYPTO, TAL -
Vendor:Cisco
Cisco UCSC-885A-M8-H29 | H200 GPU server | 12x 400GbE ConnectX-7 | 3TB RAM
UCSC-885A-M8-H29
Stock Status: Enquire
Vendor:CiscoCisco UCSC-885A-M8-H29 | H200 GPU server | 12x 400GbE ConnectX-7 | 3TB RAM
GPU-accelerated rack server focused on extreme network throughput, featuring 12x 400GbE ConnectX-7 for blistering east–west and storage traffic alongside H200 GPUs and 3TB system memory. Ideal for network-intensive AI training clusters and distributed HPC. -
Vendor:NVIDIA
NVIDIA 945-13766-0007-000 | Dev board | Arm Cortex‑A78AE | PCIe | GbE | USB | Display | Linux SDK
945-13766-0007-000
Stock Status: Enquire
Vendor:NVIDIANVIDIA 945-13766-0007-000 | Dev board | Arm Cortex‑A78AE | PCIe | GbE | USB | Display | Linux SDK
Embedded/automotive-focused development board built around Arm Cortex‑A78AE CPU cores for evaluating safety‑minded, real‑time applications. Ideal for prototyping edge compute, HMI, and control workloads with rich I/O and Linux SDK support. -
Vendor:NVIDIA
NVIDIA 900-5G144-2500-000 | RTX PRO 6000 Blackwell Graphics Card, 96GB GDDR7, 600W
900-5G144-2500-000
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-5G144-2500-000 | RTX PRO 6000 Blackwell Graphics Card, 96GB GDDR7, 600W
The NVIDIA RTX PRO 6000 Blackwell Workstation Edition delivers unmatched AI and graphics performance for professional workloads including AI development, 3D rendering, simulation, and data science. Built on Blackwell architecture with 24,064 CUDA cores, 752 fifth-generation Tensor Cores, and 96GB of ECC GDDR7 memory, this dual-slot graphics card provides 1792 GB/s memory bandwidth and 600W of processing power. Ideal for engineering simulation, content creation, scientific computing, and large-scale AI model inference in single-GPU workstation configurations. -
Vendor:Cisco
Cisco UCSC-885A-M8-M353 | Rack server | 8x MI350X GPUs | 2x ConnectX-7 | 23TB RAM
UCSC-885A-M8-M353
Stock Status: Enquire
Vendor:CiscoCisco UCSC-885A-M8-M353 | Rack server | 8x MI350X GPUs | 2x ConnectX-7 | 23TB RAM
8‑GPU AI compute node tailored for advanced training and large‑context inference, integrating MI350X accelerators with high‑speed ConnectX‑7 networking and a 23TB memory build. Ideal as a scale-out building block for modern AI clusters. -
Vendor:NVIDIA
NVIDIA MTM016920 | PCIe bracket | full-height | perforated | 1x zQSFP cutout | 2x LED windows | fits ConnectX-6/5/4 single-port adapters
MTM016920
Stock Status: Enquire
Vendor:NVIDIANVIDIA MTM016920 | PCIe bracket | full-height | perforated | 1x zQSFP cutout | 2x LED windows | fits ConnectX-6/5/4 single-port adapters
Full-height perforated PCIe bracket for single-port zQSFP adapters. Provides one cage opening with two LED sight windows and perforations to aid airflow while maintaining structural rigidity. Tailored for select ConnectX NICs using a single zQSFP port. -
Vendor:NVIDIA
NVIDIA MTM016918 | PCIe bracket | full-height | 2x zQSFP cutouts | 4x LED windows | solid | fits ConnectX-6/5/4 dual-port adapters
MTM016918
Stock Status: Enquire
Vendor:NVIDIANVIDIA MTM016918 | PCIe bracket | full-height | 2x zQSFP cutouts | 4x LED windows | solid | fits ConnectX-6/5/4 dual-port adapters
Field-replaceable full-height PCIe bracket for dual-port zQSFP adapters. Features precise dual-cage cutouts and four LED sight windows to align with onboard link/activity indicators. Designed for select ConnectX-6/5/4 NICs requiring a solid (non-perforated) tall bracket. -
Vendor:NVIDIA
NVIDIA MUA9652H-2SF | UFM 4.0 appliance | 2U | 2x ConnectX-6 HDR 200Gb/s InfiniBand adapters | Cyber-AI/Enterprise/Telemetry
MUA9652H-2SF
Stock Status: Enquire
Vendor:NVIDIANVIDIA MUA9652H-2SF | UFM 4.0 appliance | 2U | 2x ConnectX-6 HDR 200Gb/s InfiniBand adapters | Cyber-AI/Enterprise/Telemetry
Next-generation UFM 4.0 appliance for comprehensive InfiniBand fabric operations, supporting Cyber-AI, Enterprise, and Telemetry feature sets. The 2U platform integrates two ConnectX-6 HDR 200Gb/s adapters to deliver high-bandwidth visibility, anomaly detection, and automated fabric control for large-scale AI and HPC data centers.
