Iris Coleman
March 20, 2025 03:44
NVIDIA’s VGPU 18.0 release enhances AI capabilities on virtual platforms, supports Microsoft Windows Server 2025 and Proxmox VE, and introduces a new AI toolkit for developers.
Nvidia has announced Virtual GPU (VGPU) 18.0 technology, demonstrating significant advances in AI capabilities within Virtual Desktop Infrastructure (VDI). According to Nvidia, the release aims to improve productivity, security and performance by leveraging AI-driven workloads across virtualized environments.
Enhanced support for Microsoft Windows and Proxmox Ve
One of the main features of the VGPU 18.0 update is support for Microsoft Windows Server 2025. This integration allows users to take advantage of advanced management capabilities such as GPU partitions and live migration to facilitate efficient workload management and maintenance without destroying virtual machines. Additionally, this update supports the Windows subsystem for Linux (WSL) and allows for seamless operation of Linux distributions along with Windows.
Additionally, this release includes support for Proxmox VE, an open source server management platform. Once added to the NVIDIA VGPU Support Hypervisor, users can use software-defined GPU resources to manage workloads for stringent Linux and Windows applications. Daire Byrne, DNEG’s Systems Head, praised the seamless integration of Nvidia VGPUs and Proxmox VEs, highlighting its ability to efficiently manage GPU usage and support computationally intensive workloads.
New AI Toolkit and Industry Collaboration
NVIDIA is also introducing a new AI Virtual Workstation (VWS) toolkit aimed at accelerating AI development through VDI. The latest addition, LLMS toolkit tweaks and customization allows developers to install custom adapters on a single leading language model (LLM) for a variety of use cases using parameter-efficient fine-tuning techniques such as Low Rank Adaptation (LORA).
The company partnered with Citrix to provide AI virtual workstations, enabling industries such as healthcare, financial services and government to expand their AI-driven business goals. This collaboration improves the ability of Citrix-powered virtualized environments to run AI workloads, ensuring centralized data management and security.
Wide range of GPU compatibility and upcoming enhancements
The VGPU 18.0 release also extends uneven VGPU support, allowing a single GPU to process multiple VGPU profiles simultaneously with different frame buffer sizes. This feature has been extended to Nvidia Turing and Volta GPUs to optimize GPU utilization across a variety of generations of accelerated computing.
NVIDIA has announced upcoming support for the RTX Pro 6000 Blackwell Server Edition GPU, which is expected for the second half of 2025. This GPU, revealed in NVIDIA GTC 2025, promises to dynamically split resources for diverse workloads, from graphics to AI, and becomes a generic solution for data centers.
Overall, NVIDIA’s VGPU 18.0 release is set up to transform virtualized environments by enhancing AI capabilities and supporting a wide range of platforms and applications. Companies and developers are encouraged to explore these new features to maximize the potential of accelerated computing.
Image source: ShutterStock
