Multithreaded Processor Throughput Estimation via Cache Miss Rates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The relationship between the performance of on-chip processor cache memories and overall processor performance is not well understood for multithreaded processors, which hinders efficient scheduling and optimization in memory-intensive applications.
Innovation Solution
An estimate of multithreaded processor throughput is calculated based on cache miss rates, considering probabilities of different thread states and the effects of memory bus delays, using a hierarchical cache memory system with read and write miss rates, and accounting for the impact of a write buffer.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If multithreaded processor architecture is used to hide memory latency, then processor throughput is improved, but understanding the relationship between cache performance and overall processor performance becomes difficult
Solution Approach 1:
The patent introduces cache miss rate as an intermediary metric that mediates between complex multithreaded processor operations and overall throughput measurement. By focusing on cache miss rate as a key indicator, the system simplifies the complex relationship between multiple threads, cache hierarchy, and processor throughput into a manageable measurement and estimation framework.
2Measurement precision
If cache miss rate measurement is used to estimate throughput, then measurement precision is improved, but device complexity increases due to hierarchical cache and multiple thread states
Solution Approach 1:
The patent extracts the essential performance indicator (cache miss rate) from the complex multithreaded processor system and uses it as the primary basis for throughput estimation. By taking out cache miss rate measurement as the focal point and building the estimation model around it, the system achieves accurate throughput measurement without being overwhelmed by the full complexity of the hierarchical cache and multithreaded architecture.
3Productivity
If probabilities of multiple thread states are calculated to estimate throughput, then productivity measurement is improved, but calculation complexity increases
Solution Approach 1:
The patent segments the complex multithreaded processor state into discrete, manageable thread execution states (e.g., number of active threads, cache hit/miss states). By dividing the continuous complex state space into segmented categories and calculating probabilities for each segment, the system achieves accurate throughput measurement while keeping calculations tractable through structured state classification.
Data Source
AI summary
An estimate is calculated of the throughput of a multi-threaded processor having N threads based on measured miss rates of a cache memory associated with the processor by calculating, based on the cache miss rates a probability that the processor is in a state with one thread running, a probability that the processor is in a state with two threads are running and continuing to a probability that the processor is in a state with N threads running, multiplying each probability by a measured throughput of the processor when it is in the corresponding state and summing the resulting products. This estimate may also be corrected for bus delays in transferring information between the cache memory and main memory. The estimate can be used for thread scheduling in a multiprocessor system.