552 Search Results for "NVIDIA ConnectX-7"
-
Vendor:Cisco
Cisco UCSC-885A-M8-H13 | AI server | H100 GPUs (high-density) | 2x ConnectX-7 400G | 15TB RAM
UCSC-885A-M8-H13
Stock Status: Enquire
Vendor:CiscoCisco UCSC-885A-M8-H13 | AI server | H100 GPUs (high-density) | 2x ConnectX-7 400G | 15TB RAM
Enterprise AI compute server configured with NVIDIA H100 accelerators and dual ConnectX-7 400G-class adapters for ultra-fast fabric connectivity. With 15TB installed memory and UCS integration, it’s engineered for model training, fine-tuning, and high-throughput inference in production environments. -
Vendor:Cisco
Cisco UCSC-885A-M8-H20 | AI server | H200 GPUs | 2x ConnectX-7 400G | 15TB RAM
UCSC-885A-M8-H20
Stock Status: Enquire
Vendor:CiscoCisco UCSC-885A-M8-H20 | AI server | H200 GPUs | 2x ConnectX-7 400G | 15TB RAM
Next-generation AI server featuring NVIDIA H200 accelerators paired with dual ConnectX-7 400G-class adapters for extreme bandwidth and low latency across AI fabrics. With 15TB of installed memory, it’s tuned for large-scale training, fine-tuning, and high-performance inference, and integrates seamlessly with UCS management. -
Vendor:Cisco
Cisco UCSC-885A-M8-H23 | Rack server | 8x H200 GPUs | Networking: ConnectX-7 | Memory: 2.3TB
UCSC-885A-M8-H23
Stock Status: Enquire
Vendor:CiscoCisco UCSC-885A-M8-H23 | Rack server | 8x H200 GPUs | Networking: ConnectX-7 | Memory: 2.3TB
AI-optimized rack server with 8x NVIDIA H200 GPUs, ConnectX-7 networking, and 2.3TB of installed system memory. Built for multi-GPU training, inference at scale, and accelerated data pipelines in enterprise data centers. -
Vendor:Cisco
Cisco UCSC-885A-M8-H25 | Rack server | 8x H200 GPUs | Networking: ConnectX-7 | Memory: 3TB
UCSC-885A-M8-H25
Stock Status: Enquire
Vendor:CiscoCisco UCSC-885A-M8-H25 | Rack server | 8x H200 GPUs | Networking: ConnectX-7 | Memory: 3TB
GPU-dense rack server tailored for AI/ML with 8x NVIDIA H200 GPUs, ConnectX-7 networking, and 3TB of installed system memory. Ideal for training at scale, accelerated analytics, and complex simulations. -
Vendor:NVIDIA
NVIDIA 980-9IA0J-00N002 | 400Gb/s OSFP AOC | InfiniBand NDR | 850nm MMF | OSFP–OSFP | 2m | Flat-top heatsink
980-9IA0J-00N002
Stock Status: Enquire
Vendor:NVIDIANVIDIA 980-9IA0J-00N002 | 400Gb/s OSFP AOC | InfiniBand NDR | 850nm MMF | OSFP–OSFP | 2m | Flat-top heatsink
High-performance 400G active optical cable for NDR InfiniBand links. OSFP-to-OSFP design with 850 nm multimode optics delivers low latency and stable throughput for HPC, AI, and fabric backbones. Flat-top heatsink variant fits tight-clearance OSFP cages. Ideal for short intra-rack and adjacent-rack connections and validated for NVIDIA Quantum-2 switches and ConnectX-7 adapters with OSFP ports. -
Vendor:NVIDIA
NVIDIA 980-9IA0R-00N01A | 400Gb/s OSFP AOC | InfiniBand NDR | 850nm MMF | OSFP–OSFP | 1.5m | Flat-top heatsink
980-9IA0R-00N01A
Stock Status: Enquire
Vendor:NVIDIANVIDIA 980-9IA0R-00N01A | 400Gb/s OSFP AOC | InfiniBand NDR | 850nm MMF | OSFP–OSFP | 1.5m | Flat-top heatsink
Balanced-length 400G active optical cable for NDR InfiniBand. OSFP-to-OSFP with 850 nm MMF ensures high signal integrity and low latency for top-of-rack or adjacent-rack links. The flat-top heatsink supports high-density, low-clearance OSFP cages and is validated for NVIDIA Quantum-2 and ConnectX-7 platforms. -
Vendor:Cisco
Cisco UCSC-885A-M8-H11 | AI training server | 8x H100 GPUs | 8x ConnectX-7 | BlueField-3 DPU | 23TB memory
UCSC-885A-M8-H11
Stock Status: Enquire
Vendor:CiscoCisco UCSC-885A-M8-H11 | AI training server | 8x H100 GPUs | 8x ConnectX-7 | BlueField-3 DPU | 23TB memory
High-performance rack server configuration built for large-scale AI training and inference. Equipped with 8x NVIDIA H100 GPUs, 8x ConnectX-7 network adapters for high-throughput fabric connectivity, a BlueField-3 DPU for accelerated networking and security offload, and 23TB system memory for massive in-memory datasets. Ideal for LLM training, HPC, and advanced analytics workloads. -
Vendor:NVIDIA
NVIDIA 900-9X6AG-0076-ST0 | ConnectX-6 Dx Dual-Port 100GbE Network Adapter, PCIe 4.0 x16
900-9X6AG-0076-ST0
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X6AG-0076-ST0 | ConnectX-6 Dx Dual-Port 100GbE Network Adapter, PCIe 4.0 x16
The NVIDIA ConnectX-6 Dx is a high-performance dual-port 100GbE SmartNIC designed for cloud, data center, and enterprise workloads. Featuring two QSFP56 ports with support for 1/10/25/40/50/100 Gb/s speeds, PCIe 4.0 x16 host interface, and advanced offload capabilities including RDMA over Converged Ethernet (RoCEv2), SR-IOV virtualization, and hardware-accelerated IPsec and TLS encryption. This adapter delivers exceptional throughput and low latency for demanding applications including storage, HPC, AI/ML, and virtualized environments. -
Vendor:NVIDIA
NVIDIA 900-9X662-0073-ST0 | ConnectX-6 Lx Dual-Port 25GbE Network Adapter, PCIe 4.0 x8
900-9X662-0073-ST0
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X662-0073-ST0 | ConnectX-6 Lx Dual-Port 25GbE Network Adapter, PCIe 4.0 x8
The NVIDIA ConnectX-6 Lx EN dual-port 25GbE network adapter delivers high-performance Ethernet connectivity for cloud, enterprise, and telco data centers. Featuring PCIe Gen 4.0 x8 host interface, dual SFP28 ports supporting 1/10/25GbE speeds, and advanced offload capabilities including RoCEv2, SR-IOV, and VXLAN/GENEVE, this SmartNIC accelerates virtualized workloads while reducing CPU overhead. With secure boot functionality and support for up to 75 Mpps message rates, it provides enterprise-grade performance and security for modern network infrastructures. -
Vendor:NVIDIA
NVIDIA DGX Spark Founders Edition ANZ 4TB | GB10 Grace Blackwell | 128GB Unified Memory | 1 PFLOP FP4 | 940-54242-0002-000
940-54242-0002-000
Stock Status: Enquire
Vendor:NVIDIANVIDIA DGX Spark Founders Edition ANZ 4TB | GB10 Grace Blackwell | 128GB Unified Memory | 1 PFLOP FP4 | 940-54242-0002-000
NVIDIA DGX Spark Founders Edition (ANZ) 4TB is a compact personal AI computer built on the NVIDIA GB10 Grace Blackwell Superchip, designed to build and run AI locally with the NVIDIA AI software stack preinstalled. AI performance: up to 1 PFLOP at FP4 Memory: 128GB LPDDR5x coherent unified system memory Storage: 4TB NVMe M.2 with self-encryption Networking: 10GbE RJ45 plus ConnectX-7 NIC (200Gbps), Wi-Fi 7, Bluetooth 5.4 LE I/O: 4 x USB-C and 1 x HDMI 2.1a DGX Spark supports local workflows such as prototyping, fine-tuning, and inference, including working with models up to 200B parameters. Two DGX Spark systems can be connected to work with even larger models (up to 405B parameters). Part number: 940-54242-0002-000 (Founders Edition, ANZ, 4TB). -
Vendor:NVIDIA
NVIDIA 900-9X658-0086-SB0 | ConnectX-6 Dx 100GbE 1x QSFP56 PCIe 4.0 x16 NIC, RoCEv2, SR-IOV, IPsec/TLS offload
900-9X658-0086-SB0
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X658-0086-SB0 | ConnectX-6 Dx 100GbE 1x QSFP56 PCIe 4.0 x16 NIC, RoCEv2, SR-IOV, IPsec/TLS offload
Single-port 100GbE QSFP56 adapter built on ConnectX-6 Dx for secure, low-latency networking in cloud, HPC, and storage fabrics. Backward compatible with QSFP28 optics and DAC/AOC to support 25/50/100G. Hardware offloads include RoCEv2 and inline IPsec/TLS to free host CPU cycles. -
Vendor:NVIDIA
NVIDIA 900-9X658-0056-SB1 | ConnectX-6 Dx 100GbE 1x QSFP56 PCIe 4.0 x16 NIC, RoCEv2, SR-IOV, IPsec/TLS offload
900-9X658-0056-SB1
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X658-0056-SB1 | ConnectX-6 Dx 100GbE 1x QSFP56 PCIe 4.0 x16 NIC, RoCEv2, SR-IOV, IPsec/TLS offload
Single-port 100GbE PCIe 4.0 x16 adapter delivering low latency, high throughput, and robust virtualization. QSFP56 cage supports 100G and is backward compatible with QSFP28 modules and DAC/AOC for 25/50/100G operation. Designed for cloud, HPC, and secure data center networking with inline IPsec/TLS offload. -
Vendor:NVIDIA
NVIDIA 900-9X658-0076-MB0 | ConnectX-6 Dx 100GbE 1x QSFP56 PCIe 4.0 x16 NIC, RoCEv2, SR-IOV, IPsec/TLS offload
900-9X658-0076-MB0
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X658-0076-MB0 | ConnectX-6 Dx 100GbE 1x QSFP56 PCIe 4.0 x16 NIC, RoCEv2, SR-IOV, IPsec/TLS offload
Enterprise-grade 100GbE adapter providing deterministic latency, scalable virtualization, and secure offloads. The single QSFP56 port supports 100G and interoperability with QSFP28 modules and DAC/AOC for 25/50/100G links. Ideal for NVMe-oF (RDMA), cloud, and HPC deployments. -
Vendor:NVIDIA
NVIDIA 900-9X662-0053-ST1 | ConnectX-6 Lx Dual-Port 25GbE NIC, PCIe 4.0 x8
900-9X662-0053-ST1
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X662-0053-ST1 | ConnectX-6 Lx Dual-Port 25GbE NIC, PCIe 4.0 x8
The NVIDIA ConnectX-6 Lx EN adapter delivers dual-port 25GbE or single-port 50GbE connectivity with PCIe Gen 4.0 x8 host interface for high-performance cloud, enterprise, and telco data centers. This SmartNIC features advanced hardware acceleration for RoCEv2, SR-IOV virtualization, OVS offloads, and ASAP2 packet processing, optimizing network efficiency while reducing CPU overhead. Designed for scalable 25GbE deployments, it combines cutting-edge performance with low total cost of ownership. -
Vendor:NVIDIA
NVIDIA 900-9X6AG-0018-ST0 | ConnectX-6 Dx 200G QSFP56 Network Adapter, PCIe 4.0
900-9X6AG-0018-ST0
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X6AG-0018-ST0 | ConnectX-6 Dx 200G QSFP56 Network Adapter, PCIe 4.0
The NVIDIA ConnectX-6 Dx is an advanced 200 Gigabit Ethernet SmartNIC designed for mission-critical cloud and data center workloads. Featuring a single-port QSFP56 interface with PCIe 4.0 x16 connectivity, this adapter delivers enterprise-grade RDMA over Converged Ethernet capabilities, hardware-accelerated virtualization, and inline security offloads. Purpose-built for demanding applications including AI/ML, storage acceleration, SDN/NFV, and high-performance computing environments where ultra-low latency and maximum throughput are essential. -
Vendor:NVIDIA
NVIDIA 980-9I949-00N005 | 800G to 2x400G Active Copper Splitter, OSFP to 2xOSFP, 5m
980-9I949-00N005
Stock Status: Enquire
Vendor:NVIDIANVIDIA 980-9I949-00N005 | 800G to 2x400G Active Copper Splitter, OSFP to 2xOSFP, 5m
The NVIDIA 980-9I949-00N005 is an NDR InfiniBand active copper splitter cable designed for high-performance data center connectivity. This twin-port cable splits a single 800Gb/s OSFP connection into two 400Gb/s OSFP connections over 5 meters, ideal for linking Quantum-2 or Spectrum-4 switches to ConnectX-7 network adapters. The active copper design delivers low latency and power-efficient connectivity for AI, HPC, and data-intensive workloads. -
Vendor:NVIDIA
NVIDIA 900-9X662-0083-ST0 | ConnectX-6 Lx Dual-Port 25GbE Network Adapter, PCIe 4.0 x8
900-9X662-0083-ST0
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X662-0083-ST0 | ConnectX-6 Lx Dual-Port 25GbE Network Adapter, PCIe 4.0 x8
The NVIDIA ConnectX-6 Lx is a high-performance dual-port 25GbE network adapter designed for enterprise data centers, cloud infrastructure, and virtualized environments. Featuring PCIe Gen 4.0 x8 connectivity and advanced offload capabilities including RoCE v2, SR-IOV, and NVIDIA ASAP2 acceleration, it delivers exceptional throughput with a message rate of 75 Mpps. This adapter includes crypto acceleration with IPsec inline encryption and secure boot functionality for enhanced data center security. -
Vendor:NVIDIA
NVIDIA 900-9X6AG-0056-ST1 | ConnectX-6 Dx Dual-Port 100G Network Adapter, PCIe 4.0
900-9X6AG-0056-ST1
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X6AG-0056-ST1 | ConnectX-6 Dx Dual-Port 100G Network Adapter, PCIe 4.0
The NVIDIA ConnectX-6 Dx is a high-performance dual-port 100GbE SmartNIC designed to accelerate mission-critical data center workloads including virtualization, security, SDN/NFV, machine learning, and storage. Featuring QSFP56 connectivity and PCIe 4.0 x16 interface, it delivers up to 200Gb/s total bandwidth with advanced RDMA, SR-IOV, and inline IPsec/TLS offload capabilities. This adapter brings hardware-accelerated security and programmable pipeline technology to modern cloud and enterprise infrastructures. -
Vendor:Cisco
Cisco UCSC-885A-M8-H26 | Rack server | 8x H200 GPUs | IO: 2x B3220 + CX7 networking | Memory: 23TB
UCSC-885A-M8-H26
Stock Status: Enquire
Vendor:CiscoCisco UCSC-885A-M8-H26 | Rack server | 8x H200 GPUs | IO: 2x B3220 + CX7 networking | Memory: 23TB
Enterprise AI rack server with 8x NVIDIA H200 GPUs, a blend of BlueField-class DPUs (2x B3220) and ConnectX-7 networking, and 23TB of installed system memory. Built for distributed training, secure accelerated networking, and large model pipelines. -
Vendor:NVIDIA
NVIDIA 900-9X6AG-0086-ST0 | ConnectX-6 Dx Dual-Port 100GbE SmartNIC, PCIe 4.0 x16
900-9X6AG-0086-ST0
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X6AG-0086-ST0 | ConnectX-6 Dx Dual-Port 100GbE SmartNIC, PCIe 4.0 x16
The NVIDIA ConnectX-6 Dx is an advanced SmartNIC designed to accelerate mission-critical data center applications including security, virtualization, SDN/NFV, big data, machine learning, and storage. Featuring dual-port 100GbE QSFP56 connectivity with PCIe 4.0 x16 host interface, it delivers hardware-accelerated SR-IOV and VirtIO offloads via ASAP2 technology. Built-in inline IPsec and TLS encryption engines provide robust data-in-motion security with unprecedented performance and scalability for cloud and enterprise deployments. -
Vendor:NVIDIA
NVIDIA 900-9X6AG-0016-ST0 | 100GbE, QSFP28, PCIe Gen4 x16, FHHL, RoCEv2, PTP/SyncE, inline IPsec/TLS
900-9X6AG-0016-ST0
Stock Status: Enquire
Vendor:NVIDIANVIDIA 900-9X6AG-0016-ST0 | 100GbE, QSFP28, PCIe Gen4 x16, FHHL, RoCEv2, PTP/SyncE, inline IPsec/TLS
Enterprise Ethernet adapter based on the ConnectX‑6 Dx architecture delivering up to 100 GbE with extensive hardware offloads. Designed for accelerated clouds, NVMe‑oF/TCP or RoCE storage, microservices, and virtualized environments that require line‑rate encryption and precise time synchronization. -
Vendor:NVIDIA
NVIDIA MCX623436AN-CDAB | Mellanox Technologies MCX623436AN-CDAB network card Internal Fiber 100000 Mbit/s
MCX623436AN-CDAB
Stock Status: Enquire
Vendor:NVIDIANVIDIA MCX623436AN-CDAB | Mellanox Technologies MCX623436AN-CDAB network card Internal Fiber 100000 Mbit/s
NVIDIA CONNECTX-6 DX, 100GBE, OCP3.0, 2xQSFP56, PCIE 4X16, THUMBSCREW BRACKET -
Vendor:Cisco
Cisco UCSC-885A-M8-M355 | Rack server | 8x MI350X GPUs | 2x ConnectX-7 | 3TB RAM
UCSC-885A-M8-M355
Stock Status: Enquire
Vendor:CiscoCisco UCSC-885A-M8-M355 | Rack server | 8x MI350X GPUs | 2x ConnectX-7 | 3TB RAM
Enterprise AI compute node featuring 8x MI350X accelerators, ConnectX‑7 networking, and 3TB memory—ideal for training pipelines, fine‑tuning, and high‑throughput inference. Designed for dense rack deployments and scalable fabrics. -
Vendor:NVIDIA
NVIDIA MCX623430MS-CDAB | NVIDIA ConnectX-6 Dx EN adapter card, 100GbE OCP3.0, With Host management , Dua l-port DSFP, Multi Host or Socket Direct,PCIe 4.0 x16, Secure Boot
MCX623430MS-CDAB
Stock Status: Enquire
Vendor:NVIDIANVIDIA MCX623430MS-CDAB | NVIDIA ConnectX-6 Dx EN adapter card, 100GbE OCP3.0, With Host management , Dua l-port DSFP, Multi Host or Socket Direct,PCIe 4.0 x16, Secure Boot
NVIDIA CONNECTX-6 DX, 100GBE, 2xDSFP, PCIE 4X16, THUMBSCREW (PULL TAB) BRACKET
