Master ' s Thesis CS - 91 - M 12 A Data Cache that Learns to Fetch
نویسندگان
چکیده
This paper describes Fido, a predictive cache [palmer 1990] that anticipates access by em ploying an associative memory to recognize regularities in access pattern for each isolated access context over time. Continual training adapts the associative memory contents to data base and access pattern changes, allowing on line access predictions for prefetcbing. We dis cuss two salient components of Fido MLP, a replacement policy for managing prefetched ob jects, and Estimating Prophet, the component that recognizes patterns and predicts access. We tben present some early simulation results which suggest that predictive caching worlcs well and conr:lude that it is a promising method.
منابع مشابه
Design of Trace Caches for High Bandwidth Instruction Fetching
In modern high performance microprocessors, there has been a trend toward increased superscalarity and deeper speculation to extract instruction level parallelism. As issue rates rise, more aggressive instruction fetch mechanisms are needed to be able to fetch multiple basic blocks in a given cycle. One such fetch mechanism that shows a great deal of promise is the trace cache, originally propo...
متن کاملDesign of a Predictive Filter Cache for Energy Savings in High Performance Processor Architectures
Filter cache has been proposed as an energy saving architectural feature [9]. A filter cache is placed between the CPU and the instruction cache (I-cache) to provide the instruction stream. Energy savings result from accesses to a small cache. There is however loss of performance when instructions are not found in the filter cache. The majority of the energy savings from the filter cache in hig...
متن کاملCache With Adaptive Fetch Size
Current cache designs support only one xed line size Fixed line size limits cache s ability in spatial temporal locality utilization In this report we present a cache design with multiple fetch sizes for better cache performance The optimal fetch size is predicted based on memory access behavior to exploit changing application locality Overall a better performance is achieved by this cache desi...
متن کاملFido: A Cache That Learns to Fetch
This paper describes Fido, a predictive cache [Palmer 19901 that prefetches by employing an associative memory to recognize access patterns within a context over time. Repeated training adapts the associative memory contents to data and access pattern changes, allowing on-line access predictions for prefetching. We discuss two salient elements of Fido MLP, a replacement policy for managing pref...
متن کاملPerformance Limits of Trace Caches
A growing number of studies have explored the use of trace caches as a mechanism to increase instruction fetch bandwidth. The trace cache is a memory structure that stores statically non-contiguous but dynamically adjacent instructions in contiguous memory locations. When coupled with an aggressive trace or multiple branch predictor, it can fetch multiple basic blocks per cycle using a single-p...
متن کامل