Manycore Cross-Connect Scheduling for Load-Adaptive Parallel Processing

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing computing architectures face inefficiencies in balancing the need for maximizing single application performance with system utilization efficiency, particularly in high performance computing and cloud computing, leading to sub-optimal resource utilization and compromised performance.

Innovation Solution

A multi-stage, application program load adaptive parallel data processing architecture that dynamically shares resources among software programs, utilizing a hardware-implemented data packet switching cross-connect and receive logic subsystem to enable flexible task execution across multiple processor cores without software involvement, ensuring efficient resource allocation and isolation between programs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If computing resources are dedicated to a single application program, then application performance is maximized, but system utilization efficiency deteriorates

Engineering Contradiction:

Improveapplication performanceVSAvoidsystem utilization efficiency

Core Design Contradiction:

ProductivityVSLoss of energy

Solution Approach 1:

The system dynamically allocates processing cores to application tasks based on real-time workload conditions. The hardware controller monitors task execution states and reassigns cores between applications as needed, transforming the static dedicated allocation into a dynamic shared allocation that adapts to changing performance requirements while maintaining high system utilization

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The same pool of processing cores serves multiple application programs simultaneously through time-multiplexed allocation. The hardware controller acts as a universal resource manager that can assign any core to any application task based on current system state, making the computing resources universally applicable to different workloads rather than dedicated to a single application

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Loss of energy

If computing resources are shared among multiple application programs, then system utilization efficiency is improved, but application performance deteriorates

Engineering Contradiction:

Improvesystem utilization efficiencyVSAvoidapplication performance

Core Design Contradiction:

Loss of energyVSProductivity

Solution Approach 1:

The system uses dynamic workload assessment to adjust resource allocation in real-time. When an application experiences high workload, the hardware controller allocates more cores to it; when workload is low, cores are released to other applications. This dynamic adjustment ensures that performance degradation is minimized while maintaining high overall system utilization

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The hardware controller implements feedback mechanisms by monitoring task execution metrics and using this information to make real-time allocation decisions. The system continuously evaluates the performance state of each application and adjusts core assignment accordingly, creating a closed-loop control system that prevents performance degradation while maximizing resource utilization

Inventive Principle:

Principle #23Feedback

3Adaptability or versatility

If software is involved in resource allocation, then flexibility is improved, but processing overhead increases

Engineering Contradiction:

Improveresource allocation flexibilityVSAvoidprocessing overhead

Core Design Contradiction:

Adaptability or versatilityVSLoss of time

Solution Approach 1:

A hardware-based controller acts as an intermediary between the processing cores and application tasks, handling all resource allocation decisions in hardware. This intermediary eliminates the need for software intervention in scheduling and allocation, providing the flexibility of software-controlled allocation while avoiding the performance overhead of software execution

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system replaces software-based resource allocation mechanisms with hardware-based control logic. The hardware controller implements scheduling algorithms and allocation decisions using digital logic circuits instead of software interpretation, substituting the mechanical/software execution layer with a faster hardware-based control plane that maintains flexibility while reducing overhead

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS20260099367A1System and method for input data load adaptive parallel processing

Publication Date: 2026.04.09 THROUGHPUTER INC

AI summary

Systems and methods provide an extensible, multi-stage, realtime application program processing load adaptive, manycore data processing architecture shared dynamically among instances of parallelized and pipelined application software programs, according to processing load variations of said programs and their tasks and instances, as well as contractual policies. The invented techniques provide, at the same time, both application software development productivity, through presenting for software a simple, virtual static view of the actually dynamically allocated and assigned processing hardware resources, together with high program runtime performance, through scalable pipelined and parallelized program execution with minimized overhead, as well as high resource efficiency, through adaptively optimized processing resource allocation.