Cache Controller Mode Switching for Coherence and Stall Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing memory systems face inefficiencies in cache allocation and transaction processing due to cache coherence states and varying priority levels, leading to stalled transactions and suboptimal performance in multi-level cache hierarchies.
Innovation Solution
Implementing dynamic transaction arbitration and cache allocation strategies, including random selection of cache ways, priority-based transaction processing, and mode-switchable caching, to optimize cache utilization and reduce transaction stalls.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If cache allocation follows strict coherence state checking, then cache consistency is maintained, but transaction processing speed decreases due to stalls
Solution Approach 1:
The patent implements dynamic cache allocation by allowing the cache controller to adaptively select between strict coherence checking and more permissive allocation based on current system state. The controller can dynamically adjust allocation policies for different cache ways, enabling faster processing when consistency risks are low while maintaining reliability when needed.
Solution Approach 2:
The system changes the parameter of cache allocation policy from static to dynamic. By modifying allocation parameters based on coherence states and pending requests, the system can optimize between consistency and speed. The controller adjusts allocation behavior by changing parameters such as whether to check coherence states or allow speculative allocation in different contexts.
2Measurement precision
If all cache ways are monitored for coherence states, then allocation accuracy is improved, but controller complexity increases
Solution Approach 1:
The patent extracts the coherence checking function from every cache way and consolidates it into a centralized arbitration logic. Instead of each way independently monitoring its own state, the controller extracts and processes coherence information centrally, reducing redundant complexity while maintaining allocation accuracy.
Solution Approach 2:
The cache controller is designed with universal arbitration logic that handles multiple functions: coherence state monitoring, random selection, pending request detection, and allocation decision-making. This multi-functional approach reduces overall system complexity by consolidating control functions into a single versatile controller rather than separate specialized units for each function.
3Productivity
If random way selection is used when no invalid ways are available, then cache utilization is improved, but risk of allocating to pending requests increases
Solution Approach 1:
The patent applies preliminary action by checking for pending requests before performing random allocation. The controller first identifies which cache ways have pending requests and excludes them from the random selection pool. This preliminary filtering ensures that random allocation only occurs among safe candidates, maintaining reliability while still achieving good utilization.
Solution Approach 2:
The arbitration logic serves as an intermediary between the random selection mechanism and the cache ways. It mediates the allocation process by filtering out ways with pending requests before allowing random selection to occur. This intermediary layer ensures that random allocation improves utilization without compromising correctness.
Data Source
AI summary
An apparatus includes a CPU core, a first cache subsystem coupled to the CPU core, and a second memory coupled to the cache subsystem. The first cache subsystem includes a configuration register, a first memory, and a controller. The controller is configured to: receive a request directed to an address in the second memory and, in response to the configuration register having a first value, operate in a non-caching mode. In the non-caching mode, the controller is configured to provide the request to the second memory without caching data returned by the request in the first memory. In response to the configuration register having a second value, the controller is configured to operate in a caching mode. In the caching mode the controller is configured to provide the request to the second memory and cache data returned by the request in the first memory.