Partial Line-Based Probe Filters Bypass Cache Multicast Probes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The exponential growth of CPU cache size outpaces the capacity of line-based cache directories, leading to inefficiencies in tracking cache lines and increased multicast probes, while region-based probe filters lack precision in identifying cache line subscriptions.
Innovation Solution
Implementing a partial line-based probe filter with spare state encoding to indicate recently invalidated cache lines, allowing transactions to bypass multicast probes and efficiently manage cache array entries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a line-based cache directory scheme is used to track individual cache lines, then tracking precision is improved, but the directory size increases linearly with total cache capacity, failing to keep up with exponential cache growth
Solution Approach 1:
The patent segments the cache directory into region-based probe filters that track groups of cache lines (sectors) rather than individual lines. This segmentation allows the directory to scale with logarithmic complexity relative to cache capacity while maintaining adequate tracking precision through sector-level granularity.
Solution Approach 2:
The patent introduces a hierarchical dimension to the cache directory structure by combining region-based tracking (coarse granularity) with partial line-based probe filters (fine granularity for communication lines). This multi-dimensional approach allows efficient scaling while preserving precision where needed.
2Device complexity
If a region-based probe filter is used to track cache regions instead of individual lines, then directory size is reduced, but the ability to identify specific cache line subscriptions is lost
Solution Approach 1:
The patent applies local quality by enhancing the region-based probe filter with partial line-based probe filters specifically for communication lines. This selective refinement provides precise subscription identification only where needed (for communication lines) while maintaining coarse-grained region tracking for other cache lines, optimizing the balance between directory size and tracking precision.
3Productivity
If partial line-based probe filter entries are reclaimed upon invalidation to free space, then directory space utilization is improved, but subsequent accesses require multicast probes increasing latency
Solution Approach 1:
The patent applies preliminary action by pre-allocating multiple probe filter entries for communication lines and maintaining them even after invalidation. This ensures that subsequent accesses can immediately determine subscription status without requiring multicast probes, thereby reducing access latency while maintaining efficient space utilization through controlled entry retention.
Data Source
AI summary
A computing device includes detection circuitry configured to detect invalidation of a line of a cache array. The computing device additionally includes setting circuitry configured to set, in response to the detected invalidation, a spare state encoding in an entry of a partial line-based probe filter that indicates recent invalidation of the line of the cache array. The computing device also includes processing circuitry configured to process a transaction that hits on the entry of the partial line-based probe filter by avoiding a multicast probe of the cache array. Various other methods, systems, and computer-readable media are also disclosed.