Your Browsing History
- Free 30-day Returns
$148.50
Ships from China.
Meet Your Seller
40Gb PCIe Network Card, ConnectX-3 VPI, FDR InfiniBand / 40GbE
- ONE PORT, TWO PROTOCOLS: VPI dual-mode design runs 56Gb/s FDR InfiniBand or 40GbE Ethernet on the same QSFP+ port switch modes in firmware to match your fabric (MCX353A-FCBT).; ULTRA-LOW LATENCY: PCIe 3.0 x8 interface with as low as 1µs port-to-port latency, built for HPC clusters, distributed storage and latency-sensitive workloads.; NATIVE RDMA: Hardware InfiniBand RDMA and RoCE acceleration move data between nodes without CPU copies MPI jobs and Ceph/GPFS clusters finish faster.; OFFLOADS THAT SAVE CPU: Hardware I/O virtualization, QoS, congestion control and GPU communication offloading keep the host processor focused on compute.; CLUSTER-READY KIT: PXE/UEFI boot, energy-efficient design, Linux/Windows/VMware driver support; pair with QSFP+ DAC/AOC or 40G SR4 transceivers (sold separately).
+
+
Overview
Specs
Reviews
Any questions? Ask CelesteSM will help you find out quickly.
Try me now
ONE PORT, TWO
PROTOCOLS: VPI dual-mode design runs 56Gb/s FDR InfiniBand or 40GbE Ethernet on the same QSFP+ port switch modes in firmware to match your fabric (MCX353A-FCBT).;
ULTRA-LOW LATENCY: PCIe 3.0 x8 interface with as low as 1µs port-to-port latency, built for HPC clusters, distributed storage and latency-sensitive workloads.;
NATIVE RDMA: Hardware InfiniBand RDMA and RoCE acceleration move data between nodes without CPU copies MPI jobs and Ceph/GPFS clusters finish faster.;
OFFLOADS THAT SAVE CPU: Hardware I/O virtualization, QoS, congestion control and GPU communication offloading keep the host processor focused on compute.;
CLUSTER-READY KIT: PXE/UEFI boot, energy-efficient design, Linux/Windows/VMware driver support; pair with QSFP+ DAC/AOC or 40G SR4 transceivers (sold separately).
PROTOCOLS: VPI dual-mode design runs 56Gb/s FDR InfiniBand or 40GbE Ethernet on the same QSFP+ port switch modes in firmware to match your fabric (MCX353A-FCBT).;
ULTRA-LOW LATENCY: PCIe 3.0 x8 interface with as low as 1µs port-to-port latency, built for HPC clusters, distributed storage and latency-sensitive workloads.;
NATIVE RDMA: Hardware InfiniBand RDMA and RoCE acceleration move data between nodes without CPU copies MPI jobs and Ceph/GPFS clusters finish faster.;
OFFLOADS THAT SAVE CPU: Hardware I/O virtualization, QoS, congestion control and GPU communication offloading keep the host processor focused on compute.;
CLUSTER-READY KIT: PXE/UEFI boot, energy-efficient design, Linux/Windows/VMware driver support; pair with QSFP+ DAC/AOC or 40G SR4 transceivers (sold separately).
Warranty & Returns
Warranty, Returns, And Additional Information
Warranty
- Please contact the Seller directly for warranty information. Warranty information may also be found on the Manufacturer's website.
- CONTACT
Return Policies
- Return for refund within: 30 days
- Return for replacement within: 30 days
- This item is covered by VoltSync Return Policy
Manufacturer Contact Info
- For a directory of all our manufacturers, please click below.
- Manufacturers Directory
LOADING...















