CN1297196A - Mapping method for distributed access to memory - Google Patents
Mapping method for distributed access to memory Download PDFInfo
- Publication number
- CN1297196A CN1297196A CN99124824A CN99124824A CN1297196A CN 1297196 A CN1297196 A CN 1297196A CN 99124824 A CN99124824 A CN 99124824A CN 99124824 A CN99124824 A CN 99124824A CN 1297196 A CN1297196 A CN 1297196A
- Authority
- CN
- China
- Prior art keywords
- memory
- speed
- storer
- scope
- access
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Abstract
A method for mapping the distributed caches is disclosed. A flag map table is used to disperse the caches. A group addresses in address space is chosen to associated with the flag map table. According to the possible combination of chosen particular addresses, user can define the blocks in system storage mapped by flag map table as cache blocks or non-cache blocks, so the cache blocks are not continuously distributed.
Description
The present invention relates to a kind of mode that reads of storer, and be particularly related to a kind of mapping mode of fast access to memory.
Be illustrated in figure 1 as a kind of known speed and get the calcspar of device.Speed is got device 110 and is mainly got control circuit 112 by fast access to memory 111 and speed and constitute, speed is got control circuit 112 may command speed access to memory 111, be responsible for the running that whole speed is got device 110, its middling speed access to memory 111 comprises data-carrier store 113 and mark memory (tag RAM) 114 again, data-carrier store (data RAM) 113 is deposited the data of corresponding primary memory 140,114 address dates of depositing corresponding to primary memory 140 of mark memory, and whether an available wherein bit was changed with the data in the data discrimination storer 113 as changing mark (dirty bit).
Fig. 2 A be fast access to memory with primary memory between corresponding situation, because fast access to memory can only be deposited the data of a primary memory part, so actual deposit data corresponding to primary memory is at data-carrier store, then the data deposited of data-carrier store are corresponding to that a part of address of primary memory, its corresponding data then leaves in the mark memory, so the address corresponding data of the index address of data-carrier store and mark memory combines, promptly equal the actual address of primary memory, shown in Fig. 2 B.If in primary memory, the address of each data, the speed access to memory all has a position corresponding with it, and the direct mapping (direct mapped) that the fast access to memory structure of this kind is called is because each position of primary memory all is mapped on a certain position of fast access to memory.
Because only deposited the data of a part in the primary memory in the fast access to memory, and CPU when running, mainly all be to do read-write at fast access to memory, so speed get device handle CPU read and write data require the time, must consideration whether choose fast access to memory to reach the problem that whether will again the data of primary memory be moved to storer, judge that the method choose is that the address that CPU is sent compares with the content of mark memory, as if identical then for choosing when reading or writing of receiving that CPU sends requires.The storer that the method that speed access to memory system shines upon with mark makes is set to can speed get (cachable), promptly aforesaid choosing.So, general market bit is limited, generally only is 8 bits or 7 bits, is divided into so storer is subjected to the influence of market bit can speed to get part and can speed get (non-cachable) part with non-.
If the system storage size is 256M, is 512K and the example that is labeled as with 8 bits with fast access to memory, its can shine upon (mapping) to maximum can speed get the range limited of (cachable) in the size of market bit, that is 512K 2
8Doubly, i.e. 128M, as shown in Figure 3.The mapping method of existing fast access to memory, make can fast access to memory scope be continuous, the 200a of first as the 256M system storage among Fig. 3 (system memory), be 0M to the scope of 128M for can speed getting part, the above storer to the 256M scope of 128M then can speed be got controller (cachablecontroller) and is set at non-scope of can speed getting (non-cachable) storer.Can know that by above-mentioned the mark mapping mode of existing fast access to memory system is the storer separated into two parts, a part be lower floor can fast access to memory 200a, another part then is that the non-of upper strata can fast access to memory 200b.Yet, operating system (operatmg system, ⊙ S) can utilize the storer on upper strata to be keep (the status keeping) of storehouse (stack) or state often, begin to begin up storehouse as address toward push-down stack or from the address of 0M from 256M, therefore, the existing mark mapping method that can fast get scope only can be with the underclad portion 200a of storer, and 0～128M is as can speed getting scope.So this is very inefficent for operating system.Therefore, the 200b of the superiors of system storage and orlop 200a are all can the speed scope of getting be research direction at the fast access to memory of control.
In sum, the mark mapping method of known fast access to memory can only be mapped to one continuous can fast access to memory block, and be the underclad portion of system storage, causing the frequent topmost portion storer of operating system access to become can fast access to memory scope, causes overall system efficiency to reduce.
Therefore purpose of the present invention is exactly that a kind of mapping method of distributed access to memory is being provided, and it can make the superiors' part of system storage and orlop part all can be the storer that can fast get.
Another object of the present invention is exactly that a kind of mapping method of distributed access to memory is being provided, and it can disperse the scope of can speed getting of system storage at demand, and needn't be confined to a continuous memory block.
Another object of the present invention is exactly that a kind of mapping method of distributed access to memory is being provided, and makes the most frequent system storage part of operating system access can be configured to for can speed getting scope, to increase the efficient of system.
For reaching above-mentioned purpose with other, the present invention proposes a kind of mapping method of distributed access to memory, and it is summarized as follows:
The mapping method of distributed access to memory disclosed in this invention, a selected wherein group address is as corresponding use with the mark mapping table in an address space.Define system storage that its corresponding mark mapping table is mapped to for can speed getting block and non-ly can speed getting block according to may making up of selected particular address by the user.Because, the system storage that the mark mapping table was mapped to is for can speed getting block and non-ly can speed getting block system and by the user particular address bit is defined, so the superiors' part of system storage can be defined as simultaneously and can speed be got part with the orlop part, or according to user's requirement definition for which block for can speed getting scope or non-ly can speed getting scope, what make storer can the speed scope of getting no longer be continuous distribution, but with required and dispersedly (scatter) distribute.
By this, the user is according to mapping method proposed by the invention, and the superiors' block of system storage that can the operating system access is frequent and orlop block are defined as and can speed get scope, use the efficient of increase system.
For above-mentioned purpose of the present invention, feature and advantage can be become apparent, preferred embodiment cited below particularly, and conjunction with figs. elaborate:
Fig. 1 represents that a kind of existing speed gets the calcspar of device;
Fig. 2 A represents the corresponding situation between fast access to memory and the primary memory;
Fig. 2 B is fast access to memory localization method;
Fig. 3 represents can speed getting scope and can fast getting the synoptic diagram of scope with non-of existing system storer;
Fig. 4 represents the mapping method according to distributed access to memory of the present invention, the corresponding relation between mark mapping table and storer; And
The synoptic diagram of the corresponding example in one of them address in Fig. 5 presentation graphs 4.
The mapping method of distributed access to memory disclosed in this invention, making that the frequent memory block of system access is defined as can fast access to memory, uses the efficient that increases the system access storer.
In an address space (address space) wherein one group of bit of selected upper bit position as with the corresponding use of mark mapping table (tag mapping table).Define system storage that its corresponding mark mapping table is mapped to for can speed getting block and non-ly can speed getting block according to may making up of one group of selected bit by the user.Because, the system storage that the mark mapping table was mapped to is for can speed getting common piece and non-ly can speed getting block system and by the user this group bit is defined, so the superiors' part of system storage can be defined as simultaneously and can speed be got part with the orlop part, or according to user's requirement definition for which block for can speed getting scope or non-ly can speed getting scope, what make storer can the speed scope of getting no longer be continuous distribution, but with required and dispersedly (scatter) distribute.
Please refer to Fig. 4, it represents the mapping method of distributed access to memory of the present invention, and the corresponding relation between mark mapping table and storer is used explanation method of operating of the present invention and its effect.
In the address space, one group of bit of upper bit position as with the market bit corresponding address, with this group bit via a coded program.Bit group behind the coding, if consistent with market bit, then storer be for can speed getting part, what make storer by this can speed get scope position distributing but not the form of continuous distribution as traditional.
For example, as address A[22:20 with an address space] as with the corresponding bit of mark, what the combination of three bits decided that the mark mapping table corresponded to by this can fast access to memory part.At this is that 8M is an example with the memory span.Speed access to memory size is 512K, mark (tag) three figure places, then according to mark was mapped to can fast access to memory size be 2 of 512K
3Doubly, i.e. the size of 4M.That is, there is 4M speed to get in the storer of 8M.
A[22:20] address bit add up to 8 kinds of different combinations arranged, as shown in Figure 4 in regular turn by (000) to (111), its correspond to respectively storer eight part: 8M～7M, 7M～6M ... eight memory blocks such as 2M～1M and 1M～0M.In this example, storer is divided into eight equal portions.
Because operating system can utilize the storer on upper strata to do keeping of storehouse or state often, so storer that should the zone is frequent with access, therefore the superiors' part and orlop part are set for earlier and can speed be got scope, that is 8M～7M and 1M～0M two parts among Fig. 4 for example, its size is 2M.Remaining 2M scope then can be set according to required.Can speed get scope just as shown in the figure in the last storer, present a kind of distributed distribution, and the most frequent the superiors and the orlop of system access, be set to and speed to get part as two parts of 8M～7M and 1M～0M.Therefore, the efficient of system access storer just can greatly improve.
Fig. 5 then represents A[22:20] be the situation of (011), and see through a coded program with label coding one-tenth (01), and the correspondence that this corresponding memory range is 5M～4M diagram.
Therefore, when system storage has 512M, and when the mark of 8 bits is arranged, the size that it can fast access to memory is 128M.With 8M is a unit, according to above-mentioned method, according to selected address and a coded program market bit being pointed to the scope that storer can fast get is disperseed, the scope of the scope system head usefulness of more storer upper strata and lower layer part can being graded is set for can fast access to memory, to improve the efficient of system.
Therefore, feature of the present invention system with a specific address in the address space as with the market bit corresponding address, with this particular address via a coded program, if it is consistent with market bit, storer becomes and can speed get part, and what make storer by this can the speed scope of getting be distributing but not as the form of the continuous distribution traditional.
Another feature of the present invention is that the superiors of system storage and orlop part are all set and become and can speed get scope.This part is the most frequent part of operating system access, so the efficient of system access storer can significantly improve.
A feature more of the present invention is can the speed scope of getting can setting arbitrarily of system storage, be not confined to traditional continuous distribution form.
In sum, though the present invention with preferred embodiment openly as above, so it is not in order to limit the present invention; present technique field personnel; without departing from the spirit and scope of the present invention, can do various changes and modification, so protection scope of the present invention is determined by the accompanying Claim book.
Claims (4)
1. the mapping method of a distributed access to memory can speed be got scope in order to a storer is set at one, comprising:
With a plurality of bits of the higher address bit in the address space as the bit corresponding with a market bit;
Those bits are encoded via a coded program; And
Those bits behind coding are when consistent with this market bit, and this storer is speed to get part, and wherein those bits the superiors and orlop scope of corresponding to this storer is set in advance and can speed gets scope.
2. the mapping method of a distributed access to memory can speed be got scope in order to a storer is set at one, comprising:
With a plurality of bits of the higher address bit in the address space as the bit corresponding with a market bit;
Those bits are encoded via a coded program; And
When those bits behind the coding were consistent with this market bit, part can speed be got by this storer system,
3. the mapping method of fast access to memory as claimed in claim 2, wherein those bits behind the coding are consistent with this market bit is to be set by system.
4. the mapping method of fast access to memory as claimed in claim 2, wherein those bits correspond to this storer by system the most used scope be set in advance and can speed get scope.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN99124824A CN1125402C (en) | 1999-11-18 | 1999-11-18 | Mapping method for distributed access to memory |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN99124824A CN1125402C (en) | 1999-11-18 | 1999-11-18 | Mapping method for distributed access to memory |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1297196A true CN1297196A (en) | 2001-05-30 |
| CN1125402C CN1125402C (en) | 2003-10-22 |
Family
ID=5283616
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN99124824A Expired - Lifetime CN1125402C (en) | 1999-11-18 | 1999-11-18 | Mapping method for distributed access to memory |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1125402C (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7921262B1 (en) | 2003-12-18 | 2011-04-05 | Symantec Operating Corporation | System and method for dynamic storage device expansion support in a storage virtualization environment |
| CN102508802A (en) * | 2011-11-16 | 2012-06-20 | 刘大可 | Data writing method based on parallel random storages, data reading method based on same, data writing device based on same, data reading device based on same and system |
-
1999
- 1999-11-18 CN CN99124824A patent/CN1125402C/en not_active Expired - Lifetime
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7921262B1 (en) | 2003-12-18 | 2011-04-05 | Symantec Operating Corporation | System and method for dynamic storage device expansion support in a storage virtualization environment |
| CN102508802A (en) * | 2011-11-16 | 2012-06-20 | 刘大可 | Data writing method based on parallel random storages, data reading method based on same, data writing device based on same, data reading device based on same and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1125402C (en) | 2003-10-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1105973C (en) | Flash EEPROM main memory in computer system | |
| JP4155463B2 (en) | System having flash management system using only sequential writing and method thereof | |
| CN1317644C (en) | Method and apparatus for multithreaded cache with simplified implementation of cache replacement policy | |
| KR100373313B1 (en) | A method and system for managing a cache memory | |
| US20020133668A1 (en) | Memory device search system and method | |
| CN101241472B (en) | Mapping management process and system | |
| CN1459058A (en) | Cache way prediction based on instruction base register | |
| JPH07334315A (en) | Storage system | |
| WO1996008772B1 (en) | Method of pre-caching data utilizing thread lists and multimedia editing system using such pre-caching | |
| CN1659526A (en) | Method and apparatus for multithreaded cache with cache eviction based on thread identifier | |
| CN104395904A (en) | Efficient data object storage and retrieval | |
| CN101510176B (en) | Control method of general-purpose operating system for accessing CPU two stage caching | |
| DE10055603A1 (en) | Accessing method for data sets of data memory libraries involves providing rest time status to data memory library and accessing database by data memory library having largest rest time status | |
| KR910015988A (en) | Disk drive transition and fault handling method and device | |
| CN1118025C (en) | Method for realizing intelligent card embedded software adopting lagic interval chained list addressing | |
| CN87103748A (en) | Use the data handling system of address mapping | |
| CN1540523A (en) | Quick method for reading/writing buffer in single task | |
| CN101051318A (en) | Method for cluster linking and track hunting magnetic disc data using file distribution table | |
| CN108255742A (en) | A kind of large capacity solid state disk mapping method | |
| WO2001073537A3 (en) | System for and method of accessing blocks on a storage medium | |
| CN1125402C (en) | Mapping method for distributed access to memory | |
| CN102968423A (en) | Design of high-performance private cloud storage node file system on basis of data container | |
| CN1258715C (en) | Virtual shared storage device and method | |
| CN101833514A (en) | Method for realizing virtual internal memory in nand type flash memory medium based on micromemory system | |
| CN203930810U (en) | A kind of mixing storage system based on multidimensional data similarity |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C06 | Publication | ||
| PB01 | Publication | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CX01 | Expiry of patent term |
Granted publication date: 20031022 |
|
| CX01 | Expiry of patent term |