NVIDIA vGPU 19.0 Elevates AI and Graphics Virtualization on Blackwell GPUs

NVIDIA has unveiled its latest virtual GPU (vGPU) software, version 19.0, designed to enhance the virtualization of graphics and artificial intelligence workloads on its Blackwell GPUs. This release aims to address the growing demands for scalability and efficiency in virtualized environments, according to NVIDIA.

Enhanced Capabilities with Blackwell GPUs

The NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs are at the forefront of this release, equipped with 96 GB of ultra-fast GDDR7 memory. These GPUs are engineered to accelerate a wide range of enterprise workloads, from AI inference and scientific computing to graphics and video applications. The vGPU 19.0 provides comprehensive support for these GPUs, significantly boosting the performance of virtual desktops and workloads in the AI era.

Maximizing GPU Utilization

NVIDIA’s Multi-Instance GPU (MIG) technology is integrated with vGPU 19.0 to partition a single GPU into multiple isolated instances, enhancing the Quality of Service (QoS) by ensuring uninterrupted workload execution. This configuration supports up to 48 concurrent virtual machines (VMs) on a single GPU, catering to diverse workloads, including business operations, video streaming, and AI development.

Meeting Modern Application Demands

With the increasing demand for virtual collaboration and rich media, NVIDIA vGPU 19.0 introduces a new 3B profile to enhance user experience with modern applications requiring accelerated graphics. This profile improves scalability and user density on servers, addressing the higher GPU memory usage observed in Windows 11 compared to Windows 10.

Security and Cloud Integration

The latest vGPU version supports Virtualization-Based Security (VBS) with Microsoft Azure Local and Windows Server hypervisors, offering enhanced security for virtualized environments crucial for sectors like finance, healthcare, and government. Additionally, NVIDIA vGPU is now available on AWS, providing fractional vGPU solutions for cost-effective scalability.

Integration with NVIDIA nVector and Login Enterprise

NVIDIA’s nVector benchmark tool, integrated with Login Enterprise, offers automated testing and monitoring for GPU-enabled virtual desktop infrastructure (VDI) environments. This partnership ensures comprehensive performance analysis and graphical responsiveness for mainstream applications, enhancing the user experience at scale.

The NVIDIA vGPU 19.0 software represents a significant step forward in optimizing the use of Blackwell GPUs, offering advanced features to improve performance across a broader range of users and workloads in virtualized settings.

Image source: Shutterstock