Graphics Processor Read-Only Pipe for Memory Surface Access

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current graphics processing systems face inefficiencies in parallel data processing, particularly in handling read-only memory surfaces, where write operations can hinder read access and lead to increased complexity and overhead.

Innovation Solution

Implementing a separate read-only pipe within the graphics processor to handle read-only memory surfaces, allowing for simultaneous read operations without the need to accommodate write operations, and optimizing cache management by designating specific cache ways for read-only and read-write accesses.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If a unified memory access path is used for both read and write operations, then memory access flexibility is maintained, but read-only access efficiency deteriorates due to write overhead

Engineering Contradiction:
Improveread-only surface access efficiencyVSAvoidmemory access path complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The memory access path is segmented into separate read-only and read-write paths. The read-only pipe is dedicated to reading from read-only surfaces, while the read-write pipe handles both reads and writes. This segmentation allows read-only operations to proceed without interference from write operations, improving access efficiency while maintaining manageable complexity through dedicated paths.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The read-only access functionality is extracted from the unified memory access path and placed into a separate read-only pipe. This extraction removes write operations from the read-only path, eliminating write overhead and enabling optimized read-only processing without compromising the flexibility of the overall memory access system.

Inventive Principle:
Principle #2Taking out (Extraction)

2Productivity

If write operations are accommodated in the same pipe as read operations, then memory access versatility is maintained, but read operation performance deteriorates due to write overhead

Engineering Contradiction:
Improveread operation performanceVSAvoidmemory access versatility
Core Design Contradiction:
ProductivityVSAdaptability or versatility

Solution Approach 1:

The memory access path is segmented into separate read-only and read-write pipes. The read-only pipe is dedicated to reading from read-only surfaces, while the read-write pipe handles both reads and writes. This segmentation allows read-only operations to proceed without interference from write operations, improving access efficiency while maintaining manageable complexity through dedicated paths.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The read-write pipe maintains universality by handling both read and write operations, while the read-only pipe provides specialized functionality for read-only access. This multi-functional design ensures that the system can accommodate various memory access patterns while optimizing for specific workloads.

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

3Adaptability or versatility

If cache management accommodates both read-only and read-write accesses, then cache versatility is maintained, but cache management complexity increases

Engineering Contradiction:
Improvecache versatilityVSAvoidcache management complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The cache is segmented into read-only ways and read-write ways. Read-only cache ways are dedicated to storing data from read-only surfaces, while read-write cache ways handle data that requires write operations. This segmentation simplifies cache management by allowing independent control and optimization of each cache type, reducing the complexity of managing mixed access patterns.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different cache ways are assigned different qualities or properties based on their access patterns. Read-only cache ways are optimized for read operations with appropriate caching strategies, while read-write cache ways are optimized for both reads and writes. This local differentiation allows each cache region to be managed according to its specific requirements, reducing overall management complexity.

Inventive Principle:
Principle #3Local quality

Data Source

PatentUS10901909B2Optimizing read only memory surface accesses
Publication Date: 2021.01.26 INTEL CORP
  • US10901909B2 patent drawing
  • US10901909B2 patent drawing
  • US10901909B2 patent drawing

AI summary

In accordance with some embodiments, a separate pipe is used in graphics processor for handling accesses, namely reads, to read only (RO) surfaces within caches. Moreover, the caches may have defined read only section and defined read write (RW) sections. The read only section may be accessed through a dedicated read only pipe and the read write section may be accessed through a read write pipe for those surfaces that can also be written. Thus, the read only sections are handled in a read only fashion without the need to accommodate writes.