SoC Memory Bandwidth Partitioning for Mixed-Criticality Workloads

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Systems-on-a-chip (SoCs) face performance interference due to shared memory sub-system resources, particularly when processes with varying criticality are deployed, leading to potential errors or failures.

Innovation Solution

Implement performance partitioning by assigning thresholds for resource access based on workload criticality, ensuring higher priority safety-related processes are protected from interference by lower priority processes through minimum and maximum bandwidth allocations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If memory sub-system resources are shared among multiple processes, then resource utilization efficiency is improved, but performance interference between processes with different criticality occurs

Engineering Contradiction:

Improveresource utilization efficiencyVSAvoidperformance interference

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent segments the shared memory sub-system resources into multiple partitions, where each partition is assigned to workloads of a specific criticality level. This segmentation allows high-criticality workloads to have dedicated resource guarantees while low-criticality workloads can utilize remaining resources, thereby preventing performance interference while maintaining high overall resource utilization.

Inventive Principle:

Principle #1Segmentation

2Reliability

If separate hardware resources are allocated to safety-related and non-safety related functions, then interference prevention is improved, but hardware complexity and requirements increase

Engineering Contradiction:

Improveinterference preventionVSAvoidhardware requirements

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent implements dynamic resource allocation within a unified hardware architecture, where resource partitions can be flexibly assigned and reconfigured based on workload criticality levels. This dynamic approach enables interference prevention through logical isolation rather than physical separation, reducing hardware complexity while maintaining safety requirements.

Inventive Principle:

Principle #15Dynamics

3Reliability

If bandwidth thresholds are assigned to control resource access, then performance partitioning is improved, but resource allocation flexibility decreases

Engineering Contradiction:

Improveperformance partitioningVSAvoidresource allocation flexibility

Core Design Contradiction:

ReliabilityVSAdaptability or versatility

Solution Approach 1:

The patent uses bandwidth thresholds as controllable parameters to manage resource access for different criticality levels. By adjusting these threshold parameters dynamically, the system can maintain performance partitioning guarantees while adapting resource allocation to changing workload demands, thus preserving both reliability and flexibility.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS20260079758A1Hardware resource partitioning for safety-related applications

Publication Date: 2026.03.19 NVIDIA CORP

AI summary

Systems and methods in accordance with the present disclosure can prevent interference of memory operations being performed, for example, by partitioning memory access to safety related applications and non-safety related applications. In various examples, one or more circuits can assign, to a workload for execution on a system-on-a-chip (SoC) and according to a criticality of the workload, at least one threshold for bandwidth of a resource of the SoC. The one or more circuits can control execution of the workload according to the at least one threshold.