Graphics Pipeline Atomic Post-Synchronization Operations

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional graphics processing architectures face sub-optimal performance and power consumption due to limited ability of software drivers to track GPU progress through post-synchronization operations, which are restricted to simple memory writes.

Innovation Solution

Implementing post-synchronization operations that include atomic operations and using a graphics pipeline to store and process these operations, allowing for more complex data modifications and tracking resources, with a flush marker triggering the execution of these operations after completion of corresponding commands.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If simple memory writes are used for post-synchronization operations, then the system is simple to implement, but performance and power efficiency are sub-optimal

Engineering Contradiction:
ImproveperformanceVSAvoidcomplexity of post-synchronization operations
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent changes the operational parameters of post-synchronization from simple memory writes to atomic operations (read-modify-write sequences). This parameter change enables sophisticated tracking and resource management while maintaining system efficiency, directly addressing the performance limitation without requiring complete architectural redesign

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The graphics processing unit is enhanced to perform multiple functions within the post-synchronization operation framework: simple memory writes, atomic read operations, atomic modify operations, and atomic write operations. This multi-functionality allows the same hardware infrastructure to handle both simple and complex synchronization needs, improving performance without proportionally increasing complexity

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

2Use of energy by stationary object

If simple memory writes are used for post-synchronization operations, then the implementation is straightforward, but power consumption is sub-optimal

Engineering Contradiction:
Improvepower consumptionVSAvoidtracking capability
Core Design Contradiction:
Use of energy by stationary objectVSAdaptability or versatility

Solution Approach 1:

The system enables self-service tracking where the graphics processing unit automatically performs atomic operations to track its own execution progress and resource status. This eliminates the need for continuous external polling by the software driver, reducing unnecessary power consumption while maintaining sophisticated tracking capability through the atomic read-modify-write operations

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent transforms the tracking mechanism from simple memory writes to atomic operations with multiple parameter states (read values, modified values, write values). This parameter enrichment allows precise tracking of GPU progress and resource status, improving adaptability while the efficient atomic operation execution model reduces overall power consumption compared to continuous simple writes

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS9626732B2Supporting atomic operations as post-synchronization operations in graphics processing architectures
Publication Date: 2017.04.18 INTEL CORP
  • US9626732B2 patent drawing
  • US9626732B2 patent drawing
  • US9626732B2 patent drawing

AI summary

Methods and systems may provide for storing a set of post-synchronization operations to a graphics memory and sending a flush marker to a graphics pipeline. Additionally, the set of post-synchronization operations may be processed in response to the flush marker exiting the graphics pipeline. In one example, the set of post-synchronization operations includes one or more atomic operations. Moreover, the set of post-synchronization operations may be obtained from an inline portion of an atomics command.