Memory Unit Partitioning for Concurrent Tensor Access in Dataflow Computing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing memory unit management in reconfigurable dataflow computing systems affects performance, particularly in coarse-grained reconfigurable architectures, due to the complexity of mapping operations in both time and space, and the need for efficient resource allocation and communication between compute and memory units.
Innovation Solution
A system for determining and using memory unit partitioning solutions, including a parser, tensor expression extractor, logical memory constraint generator, grouping module, and memory partitioning module, to optimize memory unit partitioning for reconfigurable dataflow processors, enabling efficient execution of tensor indexing expressions and access to memory units.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If memory unit partitioning is optimized for reconfigurable dataflow computing systems, then resource utilization and performance are improved, but device complexity increases due to the need for multiple modules (parser, constraint generator, grouping module, partitioning module)
Solution Approach 1:
The memory management system is divided into distinct functional modules: a parser that extracts tensor indexing expressions from source code, a logical memory constraint generator that converts expressions to constraints, a grouping module that organizes constraints into concurrent access groups, and a memory partitioning module that determines optimal partitioning solutions. This segmentation allows each module to handle specific aspects of memory optimization independently, improving resource utilization while managing complexity through modular design.
Solution Approach 2:
The patent introduces intermediate representation structures (logical memory indexing constraints and concurrent access groups) that serve as mediators between the source code and the final memory partitioning solution. These intermediaries transform complex tensor indexing expressions into structured constraints that can be systematically analyzed and resolved, bridging the gap between high-level programming and low-level memory management.
2Productivity
If tensor indexing expressions are converted to logical memory indexing constraints and grouped into concurrent access groups, then memory access efficiency is improved, but manufacturing precision requirements increase for implementing the partitioning solution
Solution Approach 1:
The system performs preliminary analysis and transformation of tensor indexing expressions into logical memory indexing constraints before actual memory partitioning is implemented. The grouping module pre-organizes these constraints into concurrent access groups, identifying potential conflicts and access patterns in advance. This preliminary action allows the memory partitioning module to make informed decisions about optimal partitioning strategies, improving memory access efficiency while providing clear implementation guidelines that reduce precision requirements.
Data Source
AI summary
A system and method for memory unit partitioning for reconfigurable dataflow computing systems includes a parser that receives and parses source code for a reconfigurable dataflow processor, a tensor expression extractor that extracts tensor indexing expressions from the source code, a logical memory constraint generator that converts the tensor indexing expressions to logical memory indexing constraints, a grouping module that groups the logical memory indexing constraints into concurrent access groups, and a memory partitioning module that determines a memory unit partitioning solution for each concurrent access group.