Clock-Cycle Voltage Droop Detection for CPU and GPU Mitigation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Computer circuits such as CPUs and GPUs experience performance issues due to voltage droops caused by large changes in current required from the power delivery network, necessitating effective mitigation strategies.

Innovation Solution

The circuitry detects voltage droop events by receiving a clock signal, obtaining predetermined values and threshold counts, and measuring system voltage. When a measurement value reaches a predetermined value, it initiates a count of clock cycles until the value reaches another predetermined value. If the count is below a threshold, the circuit triggers mitigation actions such as decreasing the clock signal frequency or pausing the clock.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If the system operates at high current to improve performance, then productivity is improved, but voltage droop occurs causing reliability to deteriorate

Engineering Contradiction:
Improveprocessing performanceVSAvoidvoltage stability
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The droop detection circuitry monitors voltage conditions in advance and triggers mitigation actions before critical voltage droop occurs. The system uses predetermined thresholds and clock cycle counting to detect early signs of voltage droop and proactively adjusts clock frequency or pauses operation to prevent performance degradation and data loss.

Inventive Principle:
Principle #10Preliminary action

2Reliability

If mitigation action is taken to address voltage droop, then reliability is improved, but productivity deteriorates due to reduced clock frequency or paused operation

Engineering Contradiction:
Improvesystem stabilityVSAvoidprocessing throughput
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The system applies partial mitigation by adjusting clock frequency rather than complete cessation. The droop detection circuitry can trigger frequency reduction as a less severe mitigation action, maintaining some level of processing while addressing voltage instability. This partial action provides a balance between reliability improvement and productivity preservation.

Inventive Principle:
Principle #16Partial or excessive action

3Reliability

If the system uses complex droop detection and mitigation circuitry, then reliability is improved, but device complexity increases

Engineering Contradiction:
Improvevoltage droop protectionVSAvoidcircuitry complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The droop detection system is self-contained within the processing device, using internal clock signals and predetermined thresholds stored in registers. The circuitry monitors its own voltage conditions and autonomously triggers mitigation actions without requiring external control systems, reducing overall system complexity while maintaining reliability.

Inventive Principle:
Principle #25Self-service

Data Source

PatentUS20250138059A1Voltage droop detection
Publication Date: 2025.05.01 ARM LTD
  • US20250138059A1 patent drawing
  • US20250138059A1 patent drawing
  • US20250138059A1 patent drawing

AI summary

The present techniques relate to voltage droop detection and there is disclosed circuitry for detecting a voltage droop event, the circuitry configured to: receive a clock signal from a clock distribution network; obtain, from a storage, a first predetermined value, a second predetermined value and a predetermined threshold count; obtain one or more measurement values associated with a system voltage; when a first measurement value of the one or more measurement values reaches the first predetermined value, initiate a count of clock cycles until a subsequent measurement value of the one or more measurement values reaches the second predetermined value, the second predetermined value being different from the first predetermined value; and when the count of clock cycles is lower than the predetermined threshold count, cause a control entity to take mitigation action.