WO2017189033A1

WO2017189033A1 - Selecting cache aging policy for prefetches based on cache test regions

Info

Publication number: WO2017189033A1
Application number: PCT/US2016/051811
Authority: WO
Inventors: Paul MOYER
Original assignee: Advanced Micro Devices Inc
Current assignee: Advanced Micro Devices Inc
Priority date: 2016-04-27
Filing date: 2016-09-15
Publication date: 2017-11-02
Anticipated expiration: 2018-10-27
Also published as: KR20180130536A; US11321245B2; JP6840169B2; CN109074312B; JP2019516188A; US20200081849A1; US20170315932A1; CN109074312A; US10509732B2; KR102470184B1

Abstract

A cache controller [105] applies an aging policy [111, 112] to a portion of a cache [110] based on access metrics for different test regions [115, 116] of the cache, whereby each test region implements a different aging policy. The aging policy for each region establishes an initial age value for each entry of the cache, and a particular aging policy can set the age for a given entry based on whether the entry was placed in the cache in response to a demand request from a processor core or in response to a prefetch request. The cache controller can use the age value of each entry as a criterion in its cache replacement policy

Description

SELECTING CACHE AGING POLICY FOR PREFETCHES BASED ON CACHE TEST REGIONS

BACKGROUND

Field of the Disclosure

The present disclosure relates generally to processors and more particularly to processor caches.

Description of the Related Art

To improve memory access efficiency, many processors employ a memory hierarchy, whereby a processor includes one or more caches to store a subset of data stored in main memory. The criterion for which subset of data from the main memory is stored in the cache can vary depending on the processor design, but typically prioritizes the caching of data that has recently been accessed. In addition, some processors employ one or more prefetchers that predict data that is likely to be accessed by the processor core in the near future, and copies that data from main memory to the one or more caches. In order to move new data into the one or more caches, the processor must typically select previously stored data for eviction based on a specified replacement scheme. For example, some processors employ a least- recently-used (LRU) replacement scheme, whereby the processor evicts the cache entry that stores data that has not been accessed by the processor core for the greatest amount of time. However, in many scenarios the LRU replacement scheme does not correspond with the memory access patterns of instructions executing at the processor core, resulting in undesirably low memory access efficiency.

BRIEF DESCRIPTION OF THE DRAWINGS

The present disclosure may be better understood, and its numerous features and advantages made apparent to those skilled in the art by referencing the

accompanying drawings. The use of the same reference symbols in different drawings indicates similar or identical items. FIG. 1 is a block diagram of a processing system that applies an aging policy to a portion of a cache based on access metrics for different test regions of the cache in accordance with some embodiments.

FIG. 2 is an example of the processing system of FIG. 1 employing different aging policies in different test regions of the cache in accordance with some embodiments.

FIG. 3 is a diagram illustrating an example of the processing system of FIG. 1 changing the aging policy at the non-test region of the cache based on access metrics for test regions of the cache in accordance with some embodiments.

FIG. 4 is a flow diagram of a method of applying an aging policy to a portion of a cache based on access metrics for different test regions of the cache in accordance with some embodiments.

DETAILED DESCRIPTION

FIGs. 1-4 illustrate techniques for applying an aging policy to a portion of a cache based on access metrics for different test regions of the cache, whereby each test region implements a different aging policy. The aging policy for each region establishes an initial age value for each entry of the cache, and a particular aging policy can set the age for a given entry based on whether the entry was placed in the cache in response to a demand request from a processor core or in response to a prefetch request. The cache can use the age value of each entry as a criterion in its cache replacement policy. By selecting a cache aging policy for the non-test portion of the cache based on access metrics for the different test regions, the cache can dynamically change the aging policy to the more efficient policy for instructions currently executing at the processor core.

To illustrate via an example, assume the cache has two test regions and a non-test region, whereby all the regions are employed by the cache to store data in response to demand requests from a processor core and prefetch requests from a prefetches The cache employs different aging policies for the different test regions, such that for one of the test regions the cache stores a specified initial age for entries associated with demand requests and a different specified initial age for entries associated with prefetch requests. For the other test region, the cache stores the same specified initial age for entries associated with demand requests and for entries associated with prefetch requests. Under both aging policies, the cache changes the age of each entry based on whether the entry is accessed by the processor core. In addition, for each test region the cache evicts entries based on least in part on the age of each entry. The processor monitors access metrics for each of the test regions, such as a cache hit rate, cache miss rate, or a combination thereof. Based on the metrics, the processor selects the aging policy for one of the test regions and applies it to the non-test region of the cache. For example, if the cache hit rate is higher for one of the test regions, the processor can apply the aging policy for that test region to the non-test region of the cache, thereby improving the hit rate for the cache overall. In addition, in response to changes in the relative cache hit rates for the different test regions change, the processor can change the aging policy applied to the non-test region apply different aging policies to the non-test region, thereby improving the cache hit rate for the cache overall. As used herein, the term "aging policy" refers to how age values are applied to entries of a cache, as well as to how such age values may be modified once applied. The aging policy can govern an initial age applied to an entry when it is placed in the cache, the modification of the age in response to an entry being accessed or other entries being accessed, or a combination thereof. The age values can be employed the cache as part of its replacement policy for selecting entries to evict when new data is to be stored at the cache.

FIG. 1 illustrates a processing system 100 that applies a cache aging policy to a portion of a cache based on access metrics for different test regions of the cache in accordance with some embodiments. The processing system 100 is generally configured to execute sets of instructions organized in the form of computer programs in order to carry out tasks on behalf of an electronic device. Accordingly, the processing system 100 can be used in any of a variety of electronic devices, such as a desktop or laptop computer, server, smartphone, tablet, game console, and the like. To facilitate execution of instructions, the processing system 100 includes a processor core 102, a cache controller 105, a cache 1 10, and a memory 120. In some embodiments, the processor core 102, cache controller 105, and cache 1 10 are formed on the same semiconductor die to form at least a portion of a processor core 102, and the memory 120 is formed on a different semiconductor die and connected to the processor 101 via one or more busses or signal lines. In some embodiments, at least a portion of the memory 120 is formed on the same semiconductor die as modules of the processor 101.

The cache 1 10 and the memory 120 are each memory devices generally configured to store data, and can therefore be random access memory (RAM) memory modules, non-volatile memory devices (e.g., flash memory), and the like. As described further herein, the cache 1 10 and the memory 120 form a memory hierarchy for the processing system 100. In addition, the memory hierarchy of the processing system 100 may include other memory modules, such as additional caches not illustrated at FIG. 1.

The processor core 102 includes an instruction pipeline having, for example, a fetch stage to fetch instructions, a decode stage to decode each fetched instruction into one or more operations, execution stages to execute the operations, and a retire stage to retire instructions whose operations have completed execution. One set of operations generated based on the instruction are referred to herein as memory access instructions, and are requests to load data (load operations) from the memory hierarchy of the processing system 100, and requests to store data at the memory hierarchy of the processing system 100. Memory access requests generated by the processor core 102 represent requests to load or store data that are being used or are known to be needed by an executing program, and are referred to herein as demand requests.

The cache controller 105 is a module generally configured to manage the memory hierarchy of the processing system 100 by moving data between the cache 1 10 and the memory 120. In particular, the cache 1 10 is a memory device including a plurality of entries, whereby each entry can be associated with one or more addresses of the memory 120. In response to a memory access request, the cache controller 105 reviews the data stored at one or more entries of the cache 1 10 to determine if any of the entries stores data associated with the memory address targeted by the memory access request. If so, the cache controller 105 indicates a cache hit and satisfies the memory access request at the identified entry, either by storing data at the entry (in the case of a store operation) or by providing the data at the identified entry to the processor core 102 (in the case of a load operation). If data associated with the memory address is not stored at an entry of the cache 1 10, the cache controller 105 indicates a cache miss and issues a request to the memory 120 for the data at the memory address. The cache controller 105 then stores the data at an entry of the cache 1 10, and satisfies the memory access request at the entry.

To support improved memory access efficiency, the processor 101 includes a prefetcher 106 that monitors demand requests issued to the cache controller 105 to identify patterns and predict, based on the identified patterns, memory addresses that will be the target of demand requests by the processor core 102 in the near future. The prefetcher 106 can issue to the cache controller memory access requests, referred to herein as prefetch requests, targeting the identified memory addresses. The cache controller 105 treats a prefetch request similar to a demand request by first identifying if data associated with the memory address targeted by the prefetch request is already stored at and entry of the cache 1 10 and, if not already stored, requesting the data from the memory 120 and storing the data at an entry of the cache 1 10.

In some embodiments, each entry of the cache 1 10 is associated with a subset of memory addresses and only data associated with that subset can be stored at the entry. For example, in some embodiments the cache is an N-way set associative cache having M sets and each of the M sets including N entries. Each of the M sets is assigned a different subset of memory addresses, such that the cache controller 105 only stores data at an entry if the data is associated with a memory address in the entry's assigned subset. When all entries of a set store data (referred to herein as "valid data") that may be the target of a memory access, the set is referred to herein as full. Further, when the cache controller 105 receives data that is to be stored at a full set, it selects one of the entries of the set to evict (that is, to remove from the set) and replace with the received data. The scheme that the cache controller 105 uses to select the entry for eviction is referred to as replacement policy for the cache 1 10.

The replacement policy for the cache 1 10 may be based at least in part on age values associated with each entry in a set. To illustrate, in response to storing data at an entry of the cache 1 10, the cache controller 105 stores an initial age value for the entry. In response to specified criteria the cache controller 105 can adjust the age value for an entry. For example, the cache controller 105 can decrease the age value for an entry in response to the entry being the target of a demand request from the processor core 102, and can increase the age value for the entry in response to another entry in the set, or in a different set, being the target of a demand request. The age value therefore reflects an indication of how recently an entry has been accessed relative to other entries of the set, or of the cache 1 10. When selecting an entry of a full set for eviction, the cache controller 05 can base the decision at least in part the age value of each entry in the set. For example, the cache controller 105 can select the entry having the highest age value for eviction. In some embodiments, the cache controller 105 can use the age value as part of weighted sum with other criteria, such as data priority information, data type information (e.g., whether the entry stores instruction data or operand data), or other information to select an entry for eviction.

As described above, the initial age value assigned to an entry by the cache controller 105, the criteria for adjusting the age value, and the amount by which the cache controller 105 adjusts the age value in response to each of the criteria is referred to as the aging policy for the entry. In some embodiments, the cache controller 105 can apply different aging policies to different regions of the cache 1 10, wherein each region includes a plurality of entries. In at least one embodiment, each region only varies as to its aging policy. Thus, the entries of a region may be used for any set, and different regions may include different entries of the same set.

In the depicted example of FIG. 1 , the cache 1 10 includes three regions: test regions 1 15 and 1 16 and non-test region 1 18. In addition, the cache controller 105 stores aging policies 1 1 1 and 1 12 that differ in at least one aspect, such as the initial age value assigned to an entry, the amount by which an age value is adjusted in response to a specified criterion, and the like. In some embodiments, the aging policy 1 1 mandates a specified initial age for entries that store data responsive to a prefetch request and a different specified initial age for entries that store data responsive to non-prefetch requests. As used herein, a non-prefetch request refers to a memory access that is not a prefetch request targeted to the cache level to which the aging policy is applied. Examples of non-prefetch requests include demand requests, memory access requests generated in response to instruction cache misses, hardware page table walk requests, speculative memory access requests generated by the processor core 102 (e.g. , memory access requests generated during speculative execution of instructions after a branch instruction), and prefetch requests generated at other cache levels.

In contrast, the aging policy 1 12 mandates the same initial age for entries that store data responsive to a prefetch request and entries that store data responsive to a non- prefetch request. The aging policy 1 1 1 can therefore be more efficient for memory access patterns that more frequently access prefetched data, and the aging policy 1 12 can be more efficient for memory access patterns that access prefetched data and non-prefetched data at about the same frequency. However, the memory access pattern for the processor core 102 can vary over time, such that the cache 1 10 may operate more efficiently under different aging policies at different times. Accordingly, for the non-test region 118, the cache controller 105 can select between the aging policies 1 1 1 and 1 12 depending on performance characteristics of the test regions 1 15 and 1 16.

To illustrate, the processor 101 includes a performance monitor 104 that can monitor and record performance characteristics, such as cache hit rate, cache miss rate, and the like, for each of the test regions 1 15 and 1 16. Based on one or more of the performance characteristics, or a combination thereof, of the test regions 1 15 and 1 16, the cache controller selects one of the aging policies 1 1 1 and 1 12 and applies it to the non-test region 1 18. For example, if the cache hit rate for the test region 1 15 is higher than the cache hit rate for the test region 1 16, the cache controller 105 can select the aging policy for the test region 1 15 (e.g., aging policy 1 1 1 ) and apply it to the non-test region 1 18. In some embodiments, the non-test region 1 18 includes the majority of the entries of the cache 1 10. Thus, by selecting the aging policy for the non-test region 1 8 as the more efficient aging policy, as indicated by the

performance of the test regions 1 15 and 1 6, the cache controller 105 enhances the overall efficiency of the cache 1 10. Further, as the performance characteristics of the test regions 1 15 and 1 16 change over time, the cache controller can change the aging policy to the policy that is more efficient for a given set of performance characteristics.

FIG. 2 illustrates an example of the cache controller 105 applying different aging policies to the test regions 1 15 and 1 16 of FIG. 1 in accordance with some embodiments. In the depicted example, each of the test regions 1 15 and 1 16 include a plurality of entries, such as entries 240 and 241 of test region 1 15 and entries 242 and 243 of test region 1 16. Each of the entries includes a data portion (e.g., data portion 245 of entry 240) to store data retrieved from the memory 120— that is, the data used to satisfy memory access requests at the entry. Each entry also includes an age field (e.g., age field 246 of entry 240) to store an age value for the entry.

For the example of FIG. 2, it is assumed the cache controller 105 applies an aging policy to the test region 1 15 wherein the initial age value stored at the age field for an entry depends on whether the data for the entry was retrieved from the memory 120 in response to a demand request or in response to a prefetch request. In particular, in response to retrieving data from the memory 120 in response to a prefetch request (e.g., prefetch request 231 ), the cache controller 105 stores an initial age value of "00" at the age field of the entry that stores the data. In contrast, in response to retrieving data from the memory 120 in response to a demand request (e.g., demand request 230), the cache controller 105 stores an initial age value of "01 " at the age field of the entry. Thus, for the aging policy applied to the test region 1 15, the cache controller applies a "younger" initial age to entries storing prefetched data and an "older" initial age to entries storing demand data. This aging policy can be efficient for memory access patterns wherein the processor core 102 does not issue demand requests for prefetched data. For the test region 1 16, the cache controller 105 applies an aging policy wherein the initial age value stored the same initial age at the age field of each entry, without regard to whether the entry is associated with a prefetch request or a demand request. Thus, in response to retrieving data from the memory 120 in response to a prefetch request (e.g., prefetch request 233), the cache controller 105 stores an initial age value of "01" at the age field of the entry that stores the data. Similarly, in response to retrieving data from the memory 120 in response to a demand request (e.g., demand request 232), the cache controller 105 stores an initial age value of "01 " at the age field at the entry. Thus, for the aging policy applied to the test region 1 16, the cache controller applies the same initial age to all entries. This aging policy can be efficient for memory access patterns where the processor core 102 accesses prefetched data and demand data with substantially equal frequency. In other embodiments, the cache controller 105 can apply an aging policy that stores a lower initial age for demand requests and a higher initial age for prefetch requests.

As described above, the cache controller 105 applies different aging policies to the different test regions 1 15 and 1 16. By measuring the performance characteristics for each test region, the cache controller 105 can identify the more efficient aging policy for a recent memory access pattern of the processor core 102, and can apply the more efficient aging policy to the non-test region 1 18, thereby improving overall cache efficiency.

In some embodiments, in addition to or instead of applying different initial age values as described above, the different aging policies can require different adjustments to the age values. For example, for the test region 1 15, the cache controller 105 can apply an aging policy such that, in response to a memory access to an entry associated with a demand request, the cache controller 105 adjusts the age value for all entries storing prefetch data to the maximum age value of "1 1 ". Thus, in response to a memory access to entry 240, the cache controller 105 can adjust the age value for entry 240 to "00" and the age value for entry 241 (associated with prefetch request 231 ) to "1 1." In contrast, in response to a memory access the cache controller 105 adjusts the age values for entries associated with demand data by increasing the age value by one. Thus, in response to a memory access to the entry 241 , the cache controller 105 increases the age value for the entry 240 (associated with demand request 230) by one, to the value "10." The aging policy applied to the test region 1 15 can be more efficient for memory access patterns where prefetch data is accessed relatively quickly after prefetch or is not accessed at all.

In contrast to the aging policy applied to the test region 1 15, the aging policy applied to the test region 1 16 can require the same adjustment of age values for each entry, without regard to whether the entry is associated with a demand request or a prefetch request. For example, in response to a memory access to an entry of the test region 1 16, the cache controller 105 can increase the age values for all other entries in the test region 1 16 by one, up to a maximum value. Thus, the different aging policies applied to the different test regions 1 15 and 1 16 can vary in one or more of several aspects, including the initial age value applied to an entry, the criteria for adjusting the age value of an entry, the amount by which the age value is adjusted in response to each criterion, and the like. Further, in some embodiments the cache 1 10 can include more than two test regions, with different aging policies applied to each test region, thereby giving the cache controller a wide range of possible aging policies to apply to the non-test region 1 18.

FIG. 3 illustrates a diagram 300 depicting an example of the relative performance of the different test regions 1 15 and 1 16 changing over time in accordance with some embodiments. The diagram 300 includes an X-axis 350 representing time and a Y- axis 351 representing cache hit rate. The diagram 300 depicts a curve 352 representing the cache hit rate for the test region 1 16 over time and a curve 353 representing the cache hit rate for the test region 1 15. In the example illustrated by the diagram 300, prior to a time 355 the cache hit rate for the test region 1 15 is greater than the cache hit rate for the test region 1 16.

Accordingly, prior to time 355 the cache controller 105 applies the aging policy for the test region 1 15 to the non-test region 1 18. At time 355, the relationship between the cache hit rates for the test regions 1 15 and 1 16 changes, such that the cache hit rate for the test region 1 16 is greater than the cache hit rate for the test region 1 15. In response, the cache controller 105 changes the aging policy for the non-test region 1 18 to match the aging policy for the test region 1 16. At a later time 356, the relationship between the cache hit rates for the test regions 1 15 and 1 16 again changes, such that the cache hit rate for the test region 1 15 is again greater than the cache hit rate for the test region 1 16. In response, the cache controller 105 changes the aging policy for the non-test region 1 18 to match the aging policy for the test region 1 15. Thus, the cache controller 05 changes the aging policy applied to the non-test region 1 8 based on changes in the relative performance of the test regions 1 15 and 1 16. FIG. 4 is a flow diagram of a method 400 of selecting an aging policy for a non-test region of a cache based on performance characteristics of different test regions of the cache in accordance with some embodiments. For purposes of description, the method 400 is described with respect to an example implementation at the processing system 100 of FIG. 1. At block 402, the cache controller 105 applies the different aging policies 1 1 1 and 1 12 to the different test regions 1 15 and 1 16 of the cache 1 10. Thus, for example, the cache controller 105 can apply the aging policy 1 1 1 to the test region 1 15 and the aging policy 1 12 to the test region 1 16. At block 404, the performance monitor 104 monitors and records performance characteristics for the different test regions 1 15 and 1 16, such as cache hit rate, cache miss rate, a weighted sum of cache hit rate and cache miss rate, and the like.

At block 406, the cache controller 105 compares the performance characteristics for each of the test regions 1 15 and 1 16 and based on the comparison selects one of the aging polices 1 1 1 and 1 12. For example, the cache controller 105 selects the aging policy that is applied to the test region that has a higher performance as indicated by the performance characteristics. At block 408, the cache controller 105 applies the selected aging policy to the non-test region 1 18 of the cache 1 10. That is, the cache controller stores initial age values and adjusts stored age values at the non-test region 1 18 in keeping with the requirements of the selected aging policy. At block 410, the cache controller 105 uses the age values stored at the non-test region 1 18 to identify cache lines for eviction in accordance with a replacement policy for the cache 1 10. The method flow returns to block 404, reflecting that the cache controller 105 continues to monitor the performance of the test regions 1 15 and 16 and based on the performance adjusts the aging policy applied to the non-test region 1 18.

In some embodiments, certain aspects of the techniques described above may implemented by one or more processors of a processing system executing software. The software comprises one or more sets of executable instructions stored or otherwise tangibly embodied on a non-transitory computer readable storage medium. The software can include the instructions and certain data that, when executed by the one or more processors, manipulate the one or more processors to perform one or more aspects of the techniques described above. The non-transitory computer readable storage medium can include, for example, a magnetic or optical disk storage device, solid state storage devices such as Flash memory, a cache, random access memory (RAM) or other non-volatile memory device or devices, and the like. The executable instructions stored on the non-transitory computer readable storage medium may be in source code, assembly language code, object code, or other instruction format that is interpreted or otherwise executable by one or more processors.

Note that not all of the activities or elements described above in the general description are required, that a portion of a specific activity or device may not be required, and that one or more further activities may be performed, or elements included, in addition to those described. Still further, the order in which activities are listed are not necessarily the order in which they are performed. Also, the concepts have been described with reference to specific embodiments. However, one of ordinary skill in the art appreciates that various modifications and changes can be made without departing from the scope of the present disclosure as set forth in the claims below. Accordingly, the specification and figures are to be regarded in an illustrative rather than a restrictive sense, and all such modifications are intended to be included within the scope of the present disclosure. Benefits, other advantages, and solutions to problems have been described above with regard to specific embodiments. However, the benefits, advantages, solutions to problems, and any feature(s) that may cause any benefit, advantage, or solution to occur or become more pronounced are not to be construed as a critical, required, or essential feature of any or all the claims. Moreover, the particular embodiments disclosed above are illustrative only, as the disclosed subject matter may be modified and practiced in different but equivalent manners apparent to those skilled in the art having the benefit of the teachings herein. No limitations are intended to the details of construction or design herein shown, other than as described in the claims below. It is therefore evident that the particular embodiments disclosed above may be altered or modified and all such variations are considered within the scope of the disclosed subject matter. Accordingly, the protection sought herein is as set forth in the claims below.

Claims

WHAT IS CLAIMED IS:

1. A method, comprising:

at a first region [1 15] of a cache [1 10], selecting an age [246] for a first cache entry [240] based on whether data [245] is stored at the first cache entry in response to a prefetch request [231 ] or a demand request [230]; at a second region [1 16] of the cache, storing the same age for cache entries

[242, 243] independent of whether data is stored at the cache entries of the second region in response to a prefetch request or a demand request; and

replacing data at the first region of the cache and the second region of the cache based on ages of cache entries at the first region and the second region.

2. The method of claim 1 , further comprising:

measuring a first performance characteristic for the first region of the cache and a second performance characteristic for the second region of the cache [404];

selecting a first aging policy [1 1 1 ] for a third region [1 18] of the cache based on the first performance characteristic and the second performance characteristic; and

replacing data at cache entries of the third region of the cache based on ages associated with the cache entries of the third region.

3. The method of claim 2, wherein selecting the first aging policy comprises:

selecting an age for a second cache entry at the third region based on whether data is stored at the first cache entry in response to a prefetch request or a demand request. 4. The method of claim 2, wherein selecting the first aging policy comprises:

storing the same age for cache entries of the third region associated with

demand requests and cache entries of the third region associated with prefetch requests. The method of claim 2, wherein selecting the first aging policy comprises selecting the first aging policy at a first time, and the method further comprises:

selecting at a second time a second aging policy [1 12] for the third region of the cache in response to a change in a relationship between the first performance characteristic and the second performance characteristic, the second aging policy being different from the first aging policy.

The method of claim 2, wherein:

the first performance characteristic comprises a cache hit rate [353] at the first region of the cache; and

the second performance characteristic comprises a cache hit rate [352] at a second region of the cache.

The method of claim 2, wherein:

the first performance characteristic comprises a cache miss rate at the first region of the cache; and

the second performance characteristic comprises a cache miss rate at a

second region of the cache.

The method of claim 2, wherein:

the first performance characteristic comprises a first weighted sum of a cache hit rate at the first region of the cache and a cache miss rate at the first region of the cache; and

the second performance characteristic comprises a second weighted sum of a cache hit rate at a second region of the cache and a cache miss rate at the second region of the cache.

The method of claim 1 , wherein selecting an age for the first cache entry

comprises:

selecting a younger age for the first cache entry in response to identifying that the data is stored at the first cache entry in response to a prefetch request; and selecting an older age for the first cache entry in response to identifying that the data is stored at the first cache entry in response to a demand request.

10. A method, comprising:

at a first region [1 15] of a cache [1 10], applying a first aging policy [1 1 1] to store a cache entry at the first region with ages based on whether the cache entry is stored in response to a demand request or a prefetch request;

at a second region [1 16] of the cache, applying a second aging policy [1 12] to store a cache entry with a same age, independent of whether the cache entry is stored in response to a demand request or a prefetch request; selecting between the first aging policy and the second aging policy based on performance characteristics for the first region and the second region to determine a third aging policy; and

replacing data at a third region [1 18] of the cache based on the selected aging policy.

1 1. The method of claim 10, wherein selecting between the first aging policy and the second aging policy comprises:

selecting the first aging policy for the third region at a first time [355]; and selecting the second aging policy for the third region at a second time [356].

12. The method of claim 1 1 , wherein the second time is after the first time.

13. The method of claim 1 1 , wherein the second aging policy for the third region is selected at the second time in response to a change in a relationship between the performance characteristics for the first region and the second region.

14. A processor [100] , comprising:

a cache [1 10] comprising a first region [1 15] and a second region [1 16]; and a cache controller [105] configured to: at the first region, select an age [246] for a first cache entry [240] based on whether data is stored at the first cache entry in response to a prefetch request [231 ] or a demand request [230];

at the second region, store the same age for cache entries independent of whether data is stored at the cache entries of the second region in response to a prefetch request or a demand request; and

replace data at the first region of the cache and the second region of the cache based on ages of cache entries at the first region and the second region.

15. The processor of claim 14, further comprising:

a performance monitor [104] to measure a first performance characteristic for the first region of the cache and a second performance characteristic for the second region of the cache; and

wherein the cache controller is to:

select a first aging policy for a third region [1 18] of the cache based on the first performance characteristic and the second performance characteristic; and

replace data at cache lines of the third region of the cache based on the selected first aging policy.

16. The processor of claim 15, wherein the cache controller is to:

select an age for a second cache entry at the third region based on whether data is stored at the first cache entry in response to a prefetch request or a demand request. 17. The processor of claim 15, wherein the cache controller is to select the first aging policy by:

storing the same age for cache lines of the third region associated with

demand requests and cache lines of the third region associated with prefetch requests. 18. The processor of claim 15, wherein the cache controller is to: select the first aging policy at a first time [355]; and

select at a second time [356] a second aging policy for the third region of the cache in response to a change in a relationship between the first performance characteristic and the second performance characteristic, the second aging policy being different from the first aging policy.

The processor of claim 15, wherein:

the first performance charactenstic comprises a cache hit rate [352] at the first region of the cache; and

the second performance characteristic comprises a cache hit rate[353] at a second region of the cache.

The processor of claim 15, wherein:

the second performance characteristic comprises a cache miss rate at a

second region of the cache.

The processor of claim 15, wherein:

the first performance characteristic comprises a first weighted sum of a

number of cache hits at the first region of the cache and a number of cache misses at the first region of the cache; and

the second performance characteristic comprises a second weighted sum of a number of cache hits at a second region of the cache and a number of cache misses at the second region of the cache.

The processor of claim 14, wherein the cache controller is to select an age for the first cache entry by:

selecting a younger age for the first cache entry in response to identifying that the data is stored at the first cache entry in response to a prefetch request; and

selecting an older age for the first cache entry in response to identifying that the data is stored at the first cache entry in response to a demand request.