GPU Offloading for CPU Pattern Matching Bottleneck
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional pattern matching scanning in computer systems, used for detecting viruses and intrusive attacks, places a significant burden on central processing units (CPUs) due to the large number of patterns that need to be compared, leading to increased resource usage and slower system performance.
Innovation Solution
Offloading pattern matching scanning operations from the CPU to a graphics processing unit (GPU), which utilizes its parallel processing capabilities to perform pattern matching scans, thereby reducing the CPU's workload and improving scanning speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If pattern matching scanning is performed using conventional sequential methods on the CPU, then the scanning can be completed with existing hardware resources, but the CPU resource utilization becomes excessive and system performance deteriorates
Solution Approach 1:
The patent introduces a pattern matching accelerator as an intermediary device between the CPU and the data being scanned. This dedicated hardware component handles the computationally intensive pattern matching operations, freeing the CPU from this burden while maintaining comprehensive virus detection capabilities. The accelerator acts as a specialized mediator that translates scanning requirements into efficient hardware-level operations.
Solution Approach 2:
The patent replaces the software-based sequential pattern matching mechanism running on the CPU with a hardware-based parallel processing system. By implementing pattern matching logic in dedicated hardware circuits rather than software instructions, the system achieves significant performance improvement while maintaining the same detection functionality.
2Reliability
If the number of patterns to be scanned is increased to address new viruses and threats, then the detection coverage is improved, but the scanning time increases significantly
Solution Approach 1:
The patent segments the large set of patterns into multiple smaller groups that can be processed simultaneously by different processing units within the pattern matching accelerator. This segmentation enables parallel processing of pattern groups, allowing the system to handle an expanded number of patterns without proportionally increasing scanning time. Each segment is processed independently and concurrently with other segments.
Solution Approach 2:
The patent transitions from sequential one-dimensional pattern processing to multi-dimensional parallel processing by organizing patterns and data into a two-dimensional processing grid. This dimensional transformation allows multiple patterns to be compared against data simultaneously across different processing elements, dramatically reducing the time required to scan through expanded pattern sets.
Data Source
AI summary
Diagnostic software often requires pattern matching scanning to be performed to detect problems such as computer viruses or unwanted intruders. A computing system offloads pattern matching scanning from a central processing unit to a graphics processing unit.