The present invention relates generally to digital computer systems. More specifically, the present invention pertains to efficiently implementing translation between virtual addresses and physical addresses of a memory management system.
Many types of digital computer systems utilize memory caches in order to improve their performance and responsiveness. In typical computer systems, a memory cache typically comprises one or more memory banks that bridge main memory and the CPU. It is faster than main memory and allows instructions to be executed and data to be read at higher speed. The more commonly implemented caches include level 1 caches (e.g., L1), level 2 caches (e.g., L2), and translation look aside buffers (e.g., TLB). Generally, the L1 cache is built into the CPU chip and the L2 cache functions as a secondary staging area that feeds the L1 cache. Increasing the size of the L2 cache may speed up some applications but have no effect on others. The TLB is a cache matching virtual addresses with their corresponding physical address translations. The TLB is typically involved in the execution of most of the applications run on a typical computer system. Modern operating systems maintaining virtual memory make constant use of the TLB as they manage the virtual memory system. Accordingly, it is very important to the performance of the computer system that the data access paths that incorporate the TLB are as thoroughly optimized as possible. Since the TLB often incorporates attribute data in addition to the virtual address to physical address translations, what is required is a solution that can optimize the performance of the TLB with such attribute data in addition to the virtual address to physical address translations.
Embodiments of the present invention provide a method and system for caching attribute data for matching attributes with physical addresses.
The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments of the invention and, together with the description, serve to explain the principles of the invention:
Reference will now be made in detail to the preferred embodiments of the present invention, examples of which are illustrated in the accompanying drawings. While the invention will be described in conjunction with the preferred embodiments, it will be understood that they are not intended to limit the invention to these embodiments. On the contrary, the invention is intended to cover alternatives, modifications and equivalents, which may be included within the spirit and scope of the invention as defined by the appended claims. Furthermore, in the following detailed description of embodiments of the present invention, numerous specific details are set forth in order to provide a thorough understanding of the present invention. However, it will be recognized by one of ordinary skill in the art that the present invention may be practiced without these specific details. In other instances, well-known methods, procedures, components, and circuits have not been described in detail as not to unnecessarily obscure aspects of the embodiments of the present invention.
Embodiments of the present invention implement a method and system for caching attribute data for use with a translation look aside buffer. The method includes storing a plurality of TLB (translation look aside buffer) entries for the virtual address to physical address translations, wherein the entries include respective attributes. A plurality of attribute entries are stored in a memory (e.g., a cache), wherein the memory is configured to provide an attribute entry when that attribute entry is not stored in the TLB. In this manner, embodiments of the present invention reduce the time penalty incurred on a TLB miss, when a page table must be accessed to obtain a physical address and when CPU cycles must be consumed looking up attributes for that physical address. By caching attributes for physical addresses, an attribute cache in accordance with the present invention can significantly reduce the amount of time required to service a TLB miss. Embodiments of the present invention and their benefits are further described below.
Notation and Nomenclature
Some portions of the detailed descriptions which follow are presented in terms of procedures, steps, logic blocks, processing, and other symbolic representations of operations on data bits within a computer memory. These descriptions and representations are the means used by those skilled in the data processing arts to most effectively convey the substance of their work to others skilled in the art. A procedure, computer executed step, logic block, process, etc., is here, and generally, conceived to be a self-consistent sequence of steps or instructions leading to a desired result. The steps are those requiring physical manipulations of physical quantities. Usually, though not necessarily, these quantities take the form of electrical or magnetic signals capable of being stored, transferred, combined, compared, and otherwise manipulated in a computer system. It has proven convenient at times, principally for reasons of common usage, to refer to these signals as bits, values, elements, symbols, characters, terms, numbers, or the like.
It should be borne in mind, however, that all of these and similar terms are to be associated with the appropriate physical quantities and are merely convenient labels applied to these quantities. Unless specifically stated otherwise as apparent from the following discussions, it is appreciated that throughout the present invention, discussions utilizing terms such as “storing” or “accessing” or “providing” or “retrieving” or “translating” or the like, refer to the action and processes of a computer system, or similar electronic computing device, that manipulates and transforms data represented as physical (electronic) quantities within the computer system's registers and memories into other data similarly represented as physical quantities within the computer system memories or registers or other such information storage, transmission or display devices.
Embodiments of the Present Invention
The TLB 100 is used to cache a subset of the translations from a virtual address space to a physical addresses space. As is well known, when a TLB “hit” occurs, the physical address translation is rapidly returned by the TLB since the virtual address-to-physical address translation is stored as an entry in the cache. In addition to caching the physical address, the TLB stores with the physical address a plurality of attributes that are descriptive of the physical address.
The attributes describe different characteristics of the physical address. Such characteristics can include, for example, whether the physical address has previously been stored within the L1 cache 150, whether the physical address has previously been translated, whether the physical address resides within a disk cache, or whether the physical address has been accessed by some other machine process, or the like. By being aware of these attributes, the computer system can tailor its response to the physical address and avoid duplication of work or corruption of the data caches. These functions are performed by the attribute logic 20.
The virtual address to physical address translation process is one of the most critical processes that occur within a computer system. It is very important to the overall performance of the computer system that the data path traversed to obtain a physical address from a virtual address be thoroughly optimized and execute as quickly as possible. Accordingly, it is important to minimize the amount of time consumed by the operation of the attribute logic 20 and the handling physical addresses in accordance with their attributes.
In the present embodiment, the attribute cache 300 is implemented as a “parallel” attribute cache. The attribute cache 300 functions by caching recently accessed attributes associated with the physical addresses stored within the TLB 100. The attribute cache 300 is a parallel attribute cache because it does not reside on the main data path that traverses the TLB 100, attribute logic 20, and the L1 data cache 150 and I/O system 160. Accordingly, the circuitry comprising the attribute cache 300 does not need to be as meticulously optimized, or as expensively implemented, as the circuitry of the other components that are on the main data path. The operation of the parallel attribute cache 300 is further described in
It should be noted that embodiments of the present invention are not limited to any particular 32-bit addressing configuration. For example, embodiments of the present invention are equally applicable to 16-bit, 64-bit, etc. types of addressing configurations. Similarly, although the tags with which the TLB is indexed are shown as being 20 bits in length, embodiments of the present invention are equally applicable to other configurations.
Generally, with virtual addresses comprising incoming 32-bit data words as shown, the most significant 20 bits (e.g., the page name) comprise a tag and are used to search the “x” number of entries of the TLB (e.g., 48 entries, 96 entries, or more) for tag matches (e.g., page name matches). The least significant 12 bits of the incoming virtual address indicate which byte of a page is addressed and become the least significant 12 bits of the physical address, as shown. The context identifier (CID), and other control bits, are included with both the 20 bits of the virtual page address and the bits of the physical address. The output of the TLB is the most significant 20 bits and of the physical address, or sometimes referred to as the page frame address. Generally, the TLB 100 caches the most recent address translations. Thus, TLB misses usually result in the entries of the TLB 100 being updated with the more recent address translations.
As shown in
In the
Upon the occurrence of an attribute cache miss, the attributes are looked up or computed by the logic unit 305. In this case, the required attribute data does not reside in either the TLB 100 or the attribute cache 300. The attributes are looked up or otherwise computed by the logic 305 and then returned to the attribute cache 300 and the TLB 100 along with the physical address.
Thus, the attribute cache 300 provides a number of advantages for the computer system. Since the attribute cache stores only the attributes along with their corresponding physical addresses, as opposed to entire virtual addresses along with corresponding physical addresses (e.g., as in the TLB), the attribute cache can have a much larger number of entries in comparison to the TLB. This increases the chances that the attribute data will reside in the attribute cache even though the attribute data may have been previously flushed from the TLB. Additionally, since the attribute cache 300 is accessed only on TLB misses, the turnover of entries within the attribute cache 300 is less than that of the TLB. To further increase performance, a designer can configure the attribute cache 300 to cache those physical attributes of physical addresses which are most time-consuming to obtain. Thus, the physical attributes that cannot be quickly computed would be the most likely candidates for inclusion in the attribute cache 300. Such examples include, a translation bit indicating whether a translation has been performed on a corresponding entry, or a cache status bit indicating a cache status of the corresponding entry, or the like.
Optionally, in one embodiment, the attribute cache 300 is speculatively loaded to anticipate future TLB misses. For example, upon the occurrence of an attribute cache miss, the logic unit 305 can be consulted to lookup/compute the attributes, and then lookup/compute the attributes for a plurality of additional physical addresses. These additional physical addresses have not yet been requested by the TLB, but are speculatively looked up in anticipation of a subsequent TLB access. In this manner, the attribute cache 300 can optionally speculate on subsequent accesses by the TLB in an attempt to reduce the amount of time in obtaining attributes.
Referring now to
Process 400 begins in step 401, where, upon a TLB miss, a page table 160 is accessed to obtain a new physical address corresponding to a virtual address. In step 402, an attribute cache 300 is accessed to obtain one or more attributes corresponding to the physical address retrieved from the page table 160. In step 403, in the case of the attribute cache miss, process 400 proceeds to step 404 where logic 305 is accessed to lookup/compute the attributes for the physical address. In step 405, the attributes and the physical address are stored within the attribute cache 300 and the TLB 100. In step 406, in the case of speculative loading of the attribute cache, the logic 305 is accessed to lookup/compute a plurality of attributes for plurality of speculative physical addresses. Subsequently, process 400 continues in step 408.
Computer System Platform
With reference now to
In general, system 500 comprises at least one CPU 501 coupled to a North bridge 502 and a South bridge 503. The North bridge 502 provides access to system memory 515 and a graphics unit 510 that drives a display 511. The South bridge 503 provides access to a plurality of coupled peripheral devices 531 through 533 as shown. Computer system 500 also shows a BIOS ROM 540 that stores BIOS initialization software.
In the
The foregoing descriptions of specific embodiments of the present invention have been presented for purposes of illustration and description. They are not intended to be exhaustive or to limit the invention to the precise forms disclosed, and obviously many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the invention and its practical application, to thereby enable others skilled in the art to best utilize the invention and various embodiments with various modifications as are suited to the particular use contemplated. It is intended that the scope of the invention be defined by the claims appended hereto and their equivalents.
This application is a Continuation of U.S. patent application Ser. No. 11/454,355, now issued as U.S. Pat. No. 7,380,098, filed on Jun. 16, 2006, by Anvin et al., and entitled “A Method and System for Caching Attribute Data for Matching Attributes with Physical Addresses,” which is a Continuation of U.S. patent application Ser. No. 10/613,801, now issued as U.S. Pat. No. 7,089,397, filed on Jul. 3, 2003, by Anvin et al., and entitled “A Method and System for Caching Attribute Data for Matching Attributes with Physical Addresses,” which are both incorporated herein by reference in their entirety.
Number | Name | Date | Kind |
---|---|---|---|
5946716 | Karp et al. | Aug 1999 | A |
6012134 | McInerney et al. | Jan 2000 | A |
6189074 | Pedneau | Feb 2001 | B1 |
6275917 | Okada | Aug 2001 | B1 |
6301647 | Green | Oct 2001 | B1 |
6304944 | Pedneau | Oct 2001 | B1 |
6351797 | Beard, Sr. et al. | Feb 2002 | B1 |
6681311 | Gaskins et al. | Jan 2004 | B2 |
6711653 | Quach et al. | Mar 2004 | B1 |
6738864 | Chauvel | May 2004 | B2 |
6769052 | Chauvel et al. | Jul 2004 | B2 |
6779085 | Chauvel | Aug 2004 | B2 |
6826670 | Middleton et al. | Nov 2004 | B2 |
6839813 | Chauvel | Jan 2005 | B2 |
6889330 | Chauvel et al. | May 2005 | B2 |
6901521 | Chauvel et al. | May 2005 | B2 |
7712098 | Chauvel et al. | May 2010 | B2 |
20040193831 | Moyer | Sep 2004 | A1 |
Number | Date | Country | |
---|---|---|---|
Parent | 11454355 | Jun 2006 | US |
Child | 12127648 | US | |
Parent | 10613801 | Jul 2003 | US |
Child | 11454355 | US |