Core Level Predication in Dense Algorithm Processing Unit

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current integrated circuit architectures, particularly GPUs, lack the necessary processing capabilities to handle complex machine learning algorithms and real-time sensor data processing for autonomous robotics and vehicles, leading to inefficiencies in sensor signal processing and computation tasks.

Innovation Solution

A dense algorithm processing unit (DAPU) with a mesh architecture and core-level predication, featuring a predicate stack of single-bit registers that controls instruction execution based on conditional clauses, enabling efficient processing of perception data and algorithms.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If general purpose integrated circuits including CPUs and GPUs are used for sensor data processing, then the system can handle basic computation tasks, but the processing performance is insufficient for complex machine learning algorithms and real-time sensor data processing

Engineering Contradiction:
Improveprocessing performanceVSAvoidcomputational capability
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The system segments processing tasks into two distinct pathways: a perception processing unit optimized for machine learning algorithms and sensor data processing, and a separate control unit for other computational tasks. This segmentation allows each unit to be specialized for its specific function, achieving high performance for perception tasks without requiring the entire system to be overly complex.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The perception processing unit employs specific architectural features tailored for machine learning workloads, including tensor processing units with specialized activation functions, convolutional neural network accelerators, and optimized memory hierarchies. These local quality enhancements provide the necessary computational power for complex algorithms while keeping the rest of the system design manageable.

Inventive Principle:
Principle #3Local quality

2Adaptability or versatility

If additional and/or disparate circuitry is assembled to a traditional GPU to handle path planning and sensor fusion, then the perception processing needs are met, but the system becomes fragmented and inefficient

Engineering Contradiction:
Improveperception processing capabilityVSAvoidsystem fragmentation
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent merges path planning, sensor fusion, and machine learning processing into a single integrated perception processing unit. This unified architecture eliminates the need for disparate circuitry assemblies, reducing system fragmentation while maintaining full adaptability for all perception processing needs through shared hardware resources and coordinated processing pipelines.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The perception processing unit is designed as a universal platform capable of handling multiple perception tasks including machine learning inference, path planning, sensor fusion, and obstacle detection. By implementing a multi-functional architecture with reconfigurable processing elements, the system achieves high adaptability without requiring separate specialized circuitry for each function.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Speed

If a unified integrated circuit architecture is designed for high performance perception processing, then real-time processing capability is achieved, but the device complexity increases

Engineering Contradiction:
Improvereal-time processing speedVSAvoidintegrated circuit architecture
Core Design Contradiction:
SpeedVSDevice complexity

Solution Approach 1:

The integrated circuit employs dynamic voltage and frequency scaling, along with runtime reconfiguration capabilities that allow the perception processing unit to adapt its operational characteristics based on task requirements. This dynamic behavior enables high-speed processing when needed while reducing complexity and power consumption during lower-demand operations, effectively managing the complexity-performance tradeoff.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS10761848B1Systems and methods for implementing core level predication within a machine perception and dense algorithm integrated circuit
Publication Date: 2020.09.01 QUADRIC IO INC
  • US10761848B1 patent drawing
  • US10761848B1 patent drawing
  • US10761848B1 patent drawing

AI summary

Systems and methods for implementing an integrated circuit with core-level predication includes: a plurality of processing cores of an integrated circuit, wherein each of the plurality of cores includes: a predicate stack defined by a plurality of single-bit registers that operate together based on one or more of logical connections and physical connections of the plurality of single-bit registers, wherein: the predicate stack of each of the plurality of processing cores includes a top of stack single-bit register of the plurality of single-bit registers having a bit entry value that controls whether select instructions to the given processing core of the plurality of processing cores is executed.