Lenovo ThinkSystem SR670 V2 with Intel GPU Flex 140

The Lenovo ThinkSystem SR670 V2 (7Z22CTO1WW) is a 2U dual-socket rack server engineered for high-density GPU compute workloads, supporting up to ten GPU accelerators in a single chassis. When configured with the Intel GPU Flex 140, the platform combines Intel's Xe-HP microarchitecture with Lenovo's enterprise-grade thermal and power management infrastructure, delivering a tightly integrated solution purpose-built for media, AI inference, and virtual desktop infrastructure at scale.

The Intel GPU Flex 140 is a 75-watt, low-profile discrete GPU built on the Intel Xe-HP architecture, featuring 12 GB of LPDDR5 memory and hardware-accelerated media encode/decode engines supporting AVC, HEVC, AV1, and VP9 codecs. The card targets high-density deployments where multiple accelerator instances per server are required, making it particularly well suited to cloud gaming, media transcoding farms, and AI inference serving. Its low thermal design power profile enables more GPU instances per rack unit compared to higher-wattage alternatives.

Deployed within the SR670 V2 chassis, the Intel GPU Flex 140 benefits from Lenovo's XClarity management stack, support for dual 3rd Gen Intel Xeon Scalable processors, and up to 6 TB of DDR4 system memory. The server supports PCIe Gen 4 interconnects throughout, ensuring sufficient bandwidth headroom for multi-GPU inference pipelines and concurrent media processing streams. The platform is validated for enterprise datacenter environments and is available through Omnixon Global for customers across UAE, GCC, EMEA, and APAC regions.

Ideal for

• High-density AI inference serving for natural language processing and computer vision models requiring multiple low-power GPU instances per rack unit
• Enterprise media transcoding and streaming, leveraging hardware AV1, HEVC, and AVC encode/decode engines for large-scale video processing pipelines
• Cloud gaming and virtual desktop infrastructure (VDI) deployments where Xe-HP graphics virtualization enables multiple concurrent user sessions per GPU
• Edge and core datacenter analytics workloads processing structured and unstructured data with GPU-accelerated compute at controlled power envelopes
• Remote visualization and GPU-accelerated virtual workstations for engineering, design, and scientific computing teams requiring enterprise-managed GPU resources
• Hyperscale media delivery platforms requiring per-rack transcoding density across AVC, HEVC, VP9, and AV1 formats for OTT and broadcast workflows

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.