Adaptive GPU Core Virtualization for Multi-VM Resource Utilization

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing data center GPUs face inefficiencies in workload utilization due to time-sliced virtualization, where a single VM may not fully utilize the GPU resources during its dedicated timeframe, leading to underutilization of compute resources.

Innovation Solution

Implementing adaptive virtualization of GPU cores and engine-based virtualization to dynamically allocate and optimize resource utilization across multiple virtual machines, ensuring efficient use of GPU resources.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If time-sliced virtualization is used to dedicate an entire GPU to a single VM for a period of time, then resource isolation and security are improved, but GPU resource utilization deteriorates because a single VM may not utilize all execution resources during its timeframe

Engineering Contradiction:

Improveresource isolationVSAvoidGPU resource utilization

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The GPU is segmented into multiple virtualization engines (VEs), where each VE can be independently allocated to different VMs. This allows fine-grained resource sharing while maintaining isolation, resolving the contradiction between resource dedication and utilization efficiency

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The virtualization engine allocation is dynamic rather than static. The system can reassign VEs to different VMs based on workload demands in real-time, allowing the GPU to adapt to changing utilization needs while maintaining security boundaries

Inventive Principle:

Principle #15Dynamics

2Productivity

If adaptive virtualization of GPU cores is implemented to dynamically allocate resources across multiple VMs, then GPU resource utilization is improved, but system complexity increases due to dynamic resource management requirements

Engineering Contradiction:

ImproveGPU resource utilizationVSAvoidvirtualization management complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

A virtualization engine acts as an intermediary layer between the physical GPU resources and multiple VMs. This intermediary manages the complexity of dynamic resource allocation, presenting a simplified interface to both the hardware and virtual machines while handling the sophisticated resource management internally

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentEP4617872A1Adaptive virtualization of GPU cores and engine based virtualization

Publication Date: 2025.09.17 INTEL CORP

Figure 1
Figure 2A
Figure 2B~2C

AI summary

One embodiment provides a graphics processor comprising a memory interface, a plurality of interfaces to a plurality of compute engines, a processing resource cluster including a plurality of processing resources, the plurality of processing resources configured to execute instructions on behalf of the plurality of compute engines, and virtualization circuitry configured to enable time-sliced virtualization of the plurality of processing resources via the plurality of compute engines, wherein the virtualization circuitry to concurrently process workloads from a plurality of guest software environments during a time-slice via dynamic assignment of the workloads to the plurality of interfaces to the plurality of compute engines.