Compiler Cache Allocation via Partition Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multicore processor systems, shared cache resources are inefficiently allocated due to conflicting demands from processor cores, leading to sub-optimal performance and increased latency and power consumption, especially for applications with varying memory usage patterns.
Innovation Solution
A compiler-generated map is created to dynamically partition the shared cache based on the execution characteristics of individual applications, allowing for optimized allocation of cache resources and reducing interference among processor cores.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If a shared cache is used by multiple processor cores, then cache resource utilization is improved, but cache conflict and processor efficiency deteriorate
Solution Approach 1:
The shared cache is segmented into multiple partitions, with each partition dedicated to a specific processor core or application. The cache management unit maintains a partition map that tracks which core can access which cache partition, eliminating conflicts while maintaining high resource utilization through compiler-directed placement of data structures in appropriate partitions.
2Ease of operation
If cache space is shared equally among applications, then fairness is improved, but allocation optimality deteriorates
Solution Approach 1:
Different cache partitions are allocated to different applications based on their specific needs and characteristics. The compiler analyzes each application's data access patterns and directs data structures to specific cache partitions where they will be most effective, providing locally optimized allocation rather than uniform distribution.
3Productivity
If more cache space is allocated to an application, then performance is improved, but other applications suffer from cache starvation
Solution Approach 1:
The cache is divided into fixed partitions that guarantee minimum resource allocation to each application. This segmentation prevents any single application from monopolizing cache space while ensuring each application has dedicated cache resources, resolving the trade-off between performance and resource availability for other applications.
4Loss of time
If cache conflicts are reduced through partitioning, then latency is reduced, but cache management complexity increases
Solution Approach 1:
A cache management unit acts as an intermediary between processor cores and the partitioned cache. It maintains a partition map and handles address translation and partition management, isolating the complexity from the processor cores themselves while enabling efficient conflict-free cache access.
Data Source
AI summary
Techniques a generally described for creating a compiler determined map for the allocation of memory space within a cache. An example computing system is disclosed having a multicore processor with a plurality of processor cores. At least one cache may be accessible to at least two of the plurality of processor cores. A compiler determined map may separately allocate a memory space to threads of execution processed by the processor cores.