NVIDIA ConnectX-4 VPI MCX453A-FCAT - Network adapter - PCIe 3.0 x8 - 56 Gigabit QSFP28 x 1
- World-class cluster, network, and storage performance
- Smart interconnect for x86, Power, ARM, and GPU-based compute and storage platforms
- High performance in virtualized overlay networks (VXLAN and NVGRE)
- Efficient I/O consolidation, lowering data center costs and complexity
- Virtualization acceleration
- Power efficiency
- Scalability to tens-of-thousands of nodes
ConnectX-4 provides exceptional high performance for demanding data centers, public and private clouds, Web 2.0 and Big Data applications, as well as High-Performance Computing (HPC) and storage systems, enabling today's corporations to meet the demands of the data explosion.
-
Coherent Accelerator Processor Interface (CAPI)
ConnectX-4 enabled CAPI provides better performance for Power and OpenPower based platforms. Such platforms benefit from better interaction between the Power CPU and the ConnectX-4 adapter, lower latency, higher efficiency of storage access, and better Return on Investment (ROI), as more applications and more virtual machines run on the platform.
-
I/O virtualization
ConnectX-4 SR-IOV technology provides dedicated adapter resources and ensured isolation and protection for virtual machines (VMs) within the server. I/O virtualization with ConnectX-4 gives data center administrators better server utilization while reducing cost, power, and cable complexity, allowing more virtual machines and more tenants on the same hardware.
-
Overlay networks
In order to better scale their networks, data center operators often create overlay networks that carry traffic from individual virtual machines over logical tunnels in encapsulated formats such as NVGRE and VXLAN. While this solves network scalability issues, it hides the TCP packet from the hardware offloading engines, placing higher loads on the host CPU. ConnectX-4 effectively addresses this by providing advanced NVGRE and VXLAN hardware offloading engines that encapsulate and de-capsulate the overlay protocol headers, enabling the traditional offloads to be performed on the encapsulated traffic. With ConnectX-4, data center operators can achieve native performance in the network architecture.
-
HPC environments
ConnectX-4 delivers high bandwidth, low latency, and high computation efficiency for the High-Performance Computing clusters. Collective communication is a communication pattern in HPC in which all members of a group of processes participate and share data. CORE-Direct (Collective Offload Resource Engine) provides advanced capabilities for implementing MPI and SHMEM collective operations. It enhances collective communication scalability and minimizes the CPU overhead for such operations, while providing asynchronous and high-performance collective communication capabilities. It also enhances application scalability by reducing the exposure of the collective communication to the effects of system noise (the bad effect of system activity on running jobs). ConnectX-4 enhances the CORE-Direct capabilities by removing the restriction on the data length for which data reductions are supported.
-
RDMA and RoCE
ConnectX-4, utilizing IBTA RDMA (Remote Data Memory Access) and RoCE (RDMA over Converged Ethernet) technology, delivers low-latency and high-performance over InfiniBand and Ethernet networks. Leveraging data center bridging (DCB) capabilities as well as ConnectX-4 advanced congestion control hardware mechanisms, RoCE provides efficient low-latency RDMA services over Layer 2 and Layer 3 networks.
-
Mellanox PeerDirect
PeerDirect communication provides high efficiency RDMA access by eliminating unnecessary internal data copies between components on the PCIe bus (for example, from GPU to CPU), and therefore significantly reduces application run time. ConnectX-4 advanced acceleration technology enables higher cluster efficiency and scalability to tens of thousands of nodes.
-
Storage acceleration
Storage applications will see improved performance with the higher bandwidth EDR delivers. Moreover, standard block and file access protocols can leverage RoCE and InfiniBand RDMA for high-performance storage access. A consolidated compute and storage network achieves significant cost-performance advantages over multi-fabric networks.
-
Distributed RAID
ConnectX-4 delivers advanced Erasure Coding offloading capability, enabling distributed RAID (Redundant Array of Inexpensive Disks), a data storage technology that combines multiple disk drive components into a logical unit for the purposes of data redundancy and performance improvement. The ConnectX-4 family's Reed-Solomon capability introduces redundant block calculations, which, together with RDMA, achieves high performance and reliable storage access.
-
Signature handover
ConnectX-4 supports hardware checking of T10 Data Integrity Field / Protection Information (T10-DIF/PI), reducing the CPU overhead and accelerating delivery of data to the application. Signature handover is handled by the adapter on ingress and/or egress packets, reducing the load on the CPU at the Initiator and/or Target machines.
