Intel Data Center GPU Flex Series 170 Tray GDCFLX170T

The Intel Data Center GPU Flex Series 170 (Tray, part number GDCFLX170T) is a purpose-built enterprise graphics processing unit engineered on Intel's Xe-HPC microarchitecture, delivering high-throughput parallel compute for demanding data center environments. Built with 4096 Xe cores and equipped with 16 GB of HBM2e memory across a 256-bit memory bus, the Flex 170 offers substantial memory bandwidth suited to latency-sensitive inference pipelines, media transcoding workloads, and cloud gaming deployments. Its low-profile, single-slot form factor and PCIe 4.0 x16 host interface make it well-suited for dense rack configurations where space, power, and thermal efficiency are primary design constraints.

The GPU incorporates dedicated hardware AV1 encode and decode engines alongside support for H.264, H.265/HEVC, and VP9 codecs, enabling high-density media delivery at scale without relying on CPU resources. Intel oneAPI software stack compatibility provides developers with a unified programming model across heterogeneous compute assets, while the GPU also supports open standards including OpenCL, OpenGL, Vulkan, and DirectX 12 Ultimate. The Flex 170 operates within a 75 W TDP envelope, drawing power solely from the PCIe slot without requiring auxiliary power connectors, which simplifies integration into existing server platforms from major brands.

As an tray unit, the GDCFLX170T is intended for system integrators and data center operators building or refreshing hyperscale or edge server platforms at volume. Its thermal design is validated for continuous operation in rack-mount server environments, and the card supports up to four GPUs per host system, enabling flexible scaling of inference and media workloads. Intel's Data Center GPU management tooling provides telemetry, health monitoring, and remote management capabilities consistent with enterprise operational requirements.

Ideal for

• High-density cloud gaming streaming services requiring low-latency GPU virtualization across multiple concurrent user sessions
• AI inference acceleration for deployed machine learning models in cloud and edge data center environments
• Enterprise media transcoding at scale, leveraging dedicated AV1, H.265, and H.264 hardware encode/decode engines for OTT and broadcast workflows
• Virtual desktop infrastructure (VDI) deployments where multiple users require GPU-accelerated graphical workloads on shared server hardware
• Streaming and rendering workloads for cloud workstation solutions targeting remote creative and engineering teams
• Dense GPU server builds where power and space constraints demand a low-profile, no-auxiliary-power, high-throughput accelerator

Technical specifications

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