Fans

TES-IT supplies ConnectX-6 Dx adapters (e.g., MCX623106AC-CDAT) that deliver 200Gb/s throughput per card with RDMA, RoCE v2, and NVMe-oF support. These are ideal for distributed training workloads and large-scale AI clusters.

Yes, many Mellanox NICs are VPI (Virtual Protocol Interconnect) adapters like the MCX653106A-HDAT. They can switch between 200Gb/s InfiniBand HDR and 200GbE with a simple configuration command. TES-IT can pre-configure these cards before shipping.

Absolutely. TES-IT refurbished NICs undergo a 12-step certification including port testing, firmware updates, burn-in, and protocol validation. All units ship with a warranty, ensuring enterprise-grade reliability.

TES-IT ConnectX-6 Dx NICs (e.g., MCX623435AC-ADAB) and ConnectX-5 cards (e.g., MCX556A-ECAT) fully support RoCE v2 and NVMe-oF, enabling low-latency, high-throughput data transfer for storage and AI workloads.

Yes. Mellanox adapters are widely used in financial trading and telecom networks for their deterministic ultra-low latency. Models like ConnectX-6 Lx provide hardware offloads and crypto acceleration, ideal for sensitive workloads.

TES-IT refurbished NICs deliver identical performance to new OEM hardware. For example, the MCX623106AC-CDAT achieves the same 200Gb/s line rate with the same firmware features, but at up to 70% lower cost.

TES-IT provides PCIe Gen3, Gen4, and Gen5 NICs. ConnectX-6 supports PCIe Gen4 x16 for 200Gb/s bandwidth, while ConnectX-7 (latest) supports PCIe Gen5 for 400Gb/s fabrics.

Yes. TES-IT can ship NICs with Ethernet or InfiniBand mode pre-set, with latest firmware and drivers applied. This ensures plug-and-play deployment in HPC labs, universities, or enterprise IT environments.

es. ConnectX-6 NICs support SR-IOV, GPUDirect, and container acceleration. This makes them ideal for Kubernetes-based cloud environments, AI/ML research, and multi-tenant HPC clusters.

TES-IT maintains 10,000+ NICs in stock across EU, US, and APAC. Orders are shipped same-day, with 24–72h delivery to most major data center hubs worldwide.

Mellanox VPI adapters like the MCX653106A-HDAT support both InfiniBand HDR (200Gb/s) and Ethernet 200GbE. TES-IT can pre-configure these NICs so universities can run hybrid clusters without additional complexity.

Yes. ConnectX-5, ConnectX-6, and ConnectX-7 NICs all support GPUDirect RDMA, which allows direct communication between GPUs and NICs, bypassing the CPU. This dramatically reduces latency in AI/ML training and HPC workloads.

Yes. ConnectX-6 Dx and newer cards include secure boot, TLS/IPsec offload, and hardware crypto acceleration. This is particularly valuable for financial services, telecoms, and enterprise IT environments requiring encryption at line rate.

Absolutely. By extending the lifecycle of NICs through refurbishment, TES-IT helps universities and enterprises reduce CO₂ emissions by up to 85% compared to buying new. All refurbished NICs are fully tested, certified, and warrantied.

ConnectX-5 supports 100Gb/s Ethernet/EDR InfiniBand, while ConnectX-6 adds 200Gb/s throughput, PCIe Gen4 support, hardware crypto, and advanced virtualization features. TES-IT supplies both, depending on workload needs and budget.

Yes. Mellanox ConnectX adapters support SR-IOV, DPDK, and container acceleration, making them compatible with VMware ESXi, Proxmox VE, OpenStack, and Kubernetes clusters. TES-IT provides NICs pre-tested for virtualization environments.

Yes. While InfiniBand HDR/NDR is common, Mellanox NICs also support RoCE v2 for Ethernet-based HPC fabrics. This gives HPC labs flexibility to choose between Ethernet or InfiniBand, depending on topology and budget.

TES-IT refurbished NICs perform identically to OEM and have been deployed in hyperscale environments where uptime is critical. ConnectX-6 NICs support millions of connections, RoCE v2, NVMe-oF, and are tested to ensure cloud-grade reliability.

Yes. TES-IT delivers complete bundles including NICs, Spectrum/Quantum switches, and DAC/AOC cabling. This ensures compatibility, reduces integration errors, and accelerates deployment in HPC clusters, enterprises, and cloud environments.