CPU-GPU Power Allocation Using Joint Acoustic Feedback Control

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In multi-processor computing devices, such as laptops with both a CPU and a GPU, existing power allocation techniques often result in inefficient use of processor power, leading to suboptimal performance due to arbitrary or 'best guess' based power limits, which can cause one processor to be overpowered while the other is underpowered, reducing overall system efficiency.

Innovation Solution

Implementing a computer-implemented method with two independent power controllers: a joint processor acoustic controller and a joint processor power controller that dynamically adjust power settings for both the CPU and GPU based on operating regimes, workload, and ambient conditions to ensure efficient power allocation and utilization, allowing unused or inefficiently used power to be reallocated.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If processor power limits are selected based on best guess for CPU and GPU allocation, then device complexity is reduced, but processor power utilization efficiency deteriorates

Engineering Contradiction:
Improvepower allocation control complexityVSAvoidprocessor power utilization efficiency
Core Design Contradiction:
Device complexityVSLoss of energy

Solution Approach 1:

The patent implements a feedback mechanism where the system continuously monitors the actual power consumption and performance of both CPU and GPU, then dynamically adjusts their power limits accordingly. The system measures whether one processor is waiting for the other (underutilization) and reallocates power to maximize joint workload performance, thereby resolving the contradiction between simple control and efficient power utilization.

Inventive Principle:
Principle #23Feedback

2Object-affected harmful factors

If CPU core frequency is modulated to prevent fan noise from exceeding specified acoustic level, then acoustic performance is improved, but processing speed deteriorates

Engineering Contradiction:
Improvefan noise levelVSAvoidCPU processing speed
Core Design Contradiction:
Object-affected harmful factorsVSSpeed

Solution Approach 1:

The patent applies dynamic power limit adjustment where the CPU power limit is not fixed but continuously adapted based on acoustic constraints and workload conditions. The system dynamically modulates CPU frequency and power allocation in response to real-time acoustic level measurements and performance requirements, allowing optimal balance between noise control and processing speed under varying operating conditions.

Inventive Principle:
Principle #15Dynamics

3Use of energy by moving object

If CPU core frequency is modulated to ensure specified battery drain rate is met, then battery drain rate control is improved, but processing speed deteriorates

Engineering Contradiction:
Improvebattery drain rateVSAvoidCPU processing speed
Core Design Contradiction:
Use of energy by moving objectVSSpeed

Solution Approach 1:

The patent implements dynamic power management where CPU power limits are continuously adjusted based on battery charge state and drain rate requirements. The system monitors battery status and dynamically modulates CPU frequency and power allocation to maintain acceptable battery drain rates while maximizing processing performance within the available power budget, resolving the contradiction between energy conservation and processing speed.

Inventive Principle:
Principle #15Dynamics

4Productivity

If one processor is allocated more power than needed for joint workload, then that processor's individual performance is improved, but overall system efficiency deteriorates

Engineering Contradiction:
Improveindividual processor performanceVSAvoidoverall system efficiency
Core Design Contradiction:
ProductivityVSLoss of energy

Solution Approach 1:

The patent implements a self-regulating power allocation system where each processor's power limit is automatically adjusted based on the actual joint workload requirements and the other processor's performance capability. The system monitors for idle cycles or waiting states and dynamically reallocates power to eliminate waste, allowing processors to self-adjust their power consumption to match actual workload needs rather than operating under fixed arbitrary limits.

Inventive Principle:
Principle #25Self-service

Data Source

PatentUS12130687B2Techniques for controlling computing performance for power-constrained multi-processor computing systems
Publication Date: 2024.10.29 NVIDIA CORP
  • US12130687B2 patent drawing
  • US12130687B2 patent drawing
  • US12130687B2 patent drawing

AI summary

A computer-implemented method of controlling power consumption in a multi-processor computing device comprises: determining whether a first processor is operating in a high-power regime or a low-power regime; selecting a first set of control rules that includes a first subset of control rules that apply when the first processor is operating in the high-power regime and a second subset of control rules that apply when the first processor is operating in the low-power regime; determining one or more power settings for the first processor based on the first set of control rules; and causing the first processor to perform one or more operations based on the one or more power settings.