CN102681946A - Memory access method and device - Google Patents
Memory access method and device Download PDFInfo
- Publication number
- CN102681946A CN102681946A CN2012101467549A CN201210146754A CN102681946A CN 102681946 A CN102681946 A CN 102681946A CN 2012101467549 A CN2012101467549 A CN 2012101467549A CN 201210146754 A CN201210146754 A CN 201210146754A CN 102681946 A CN102681946 A CN 102681946A
- Authority
- CN
- China
- Prior art keywords
- groups
- grains
- memory
- particle
- access request
- 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
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
The invention discloses a memory access method and device. The method comprises the following steps: receiving an access request which is a request for accessing a memory; and according to the access request, accessing the memory by taking a grain group as a unit, wherein the grain group is obtained by grouping memory grains, and each grain group comprises one or a plurality of memory grains. According to the memory access method and device, the problem in the relevant technique that high power consumption is caused because all memory grains at the same Rank need to open and close a page when page conflict happens in the memory accessing process is solved, the effect of reducing memory power consumption is achieved, and furthermore, the performance of a memory system is improved.
Description
Technical field
The present invention relates to computer realm, in particular to a kind of memory pool access method and device.
Background technology
In existing computer system, the performance of memory system plays a part very important to processor performance.Memory system mainly is made up of Memory Controller Hub and memory chip.Interface between Memory Controller Hub and the memory chip mainly is made up of control bus and data bus.Control bus mainly comprises control command and address information.In order to obtain higher memory access bandwidth (bandwidth is meant that in the unit interval memory system offers the data volume of processor, is unit usually with the byte), the data bus between Memory Controller Hub and the memory chip is generally 64 generally than broad.
Existing internal memory particle adopts the organizational form of Rank usually, and each Memory Controller Hub can connect a plurality of Rank, and each Rank is made up of a plurality of internal memory particles.Fig. 1 is the structural representation of existing memory system, and in Fig. 1, it is the data bus of 16 (being abbreviated as type x16) that each particle provides width, and then one 64 Rank need be made up of the internal memory particle of 4 x16.Each internal memory particle is made up of a plurality of Bank.Specifically, be example with DDR3 SDRAM, each internal memory granulopexy is made up of 8 Bank.Each Bank is made up of a plurality of row (being also referred to as page or leaf) and a row buffering.When Memory Controller Hub need read perhaps write operation to certain data line, the internal memory particle at first needed the data of this row are moved into row buffering, in row buffering, this line data is read or write operation then.Reading or after write operation accomplishes, need the data in the row buffering be carried go back to original position data.
Memory Controller Hub is usually through three types of internal memory order control memory chips: open page operations, the closed page operation and the operation that reads and writes data.For a memory access request, if the data that this memory access request need be visited in row buffering, are then claimed this memory access request " page or leaf hits ".Otherwise, claim this memory access request " page or leaf conflict ".In when, page or leaf taking place hitting when, the Memory Controller Hub operation that can directly in row buffering, read and write data.When the page or leaf conflict took place, Memory Controller Hub needed at first the row in the row buffering to be moved back to original position (closed page), and the page or leaf that will visit is then moved in the row buffering and (opened page or leaf), in row buffering, carried out the data write operation then.Therefore, the memory access request of page or leaf conflict need consume more time than the memory access request that page or leaf hits.
Existing Memory Controller Hub adopts mode shown in Figure 1 usually, is that unit carries out internal storage access with Rank.In order to obtain 64 bit data bus, Memory Controller Hub need be operated all particles on the Rank simultaneously, and in the internal storage access process, a plurality of functional units of possibility are the access memory system simultaneously, the page or leaf that the different functional units visit is different.The internal memory particle needs frequent execution page or leaf to open and the page or leaf shutoff operation in this case, causes serious page or leaf conflict, thereby makes performance of memory system low.
Fig. 2 is a kind of example of internal memory particle control in the prior art, (in Fig. 2, Rank is made up of 4 particles, and in the system of reality, each Rank possibly be made up of the internal memory particle of different numbers).Because Memory Controller Hub is operated all internal memory particles simultaneously, all internal memory particles all are in identical state.When the page or leaf conflict takes place in a memory access request on a particle, the also inevitable generation page or leaf conflict on other particle of this memory access request.During simultaneously owing to the conflict of generation page or leaf, all the internal memory particles on the same Rank all need carry out page or leaf to be opened and a page shutoff operation, has consumed a large amount of power consumptions, influences the performance of memory system.
To in the correlation technique in the process of carrying out internal storage access because when the page or leaf conflict takes place, all the internal memory particles on the same Rank all need carry out page or leaf to be opened with the page or leaf shutoff operation and causes the big problem of power consumption, does not propose effective solution at present as yet.
Summary of the invention
The invention provides a kind of memory pool access method and device; Because when the page or leaf conflict takes place, all the internal memory particles on the same Rank all need carry out page or leaf and open with the page or leaf shutoff operation and cause the big problem of power consumption to solve in the correlation technique in the process of carrying out internal storage access.
According to an aspect of the present invention, a kind of memory pool access method is provided, this method comprises: receive access request, access request is the request of access memory; According to access request is the unit access memory with the groups of grains, and wherein, groups of grains is through dividing into groups the internal memory particle to obtain, and each groups of grains comprises one or more internal memory particles.
Preferably, be that the step of unit access memory comprises with the groups of grains according to access request: obtain the memory address information of carrying in the access request, wherein, memory address information comprises the particle address, and the particle address is used in reference to the groups of grains to required visit; The groups of grains that memory address information memory address information is corresponding according to the particle address lookup; With the groups of grains that inquires is the unit operation that conducts interviews.
Preferably, with the groups of grains that inquires be unit conduct interviews the operation step comprise: with access request be sent to the corresponding command formation of particle address in; Read the indicated accessing operation of storing in the also fill order formation of access request.
Preferably, the internal memory number of particles that is comprised in the dynamic-configuration groups of grains.
Preferably; Through the internal memory number of particles that is comprised in the groups of configuration registers dynamic-configuration groups of grains; Wherein, Groups of configuration registers is made up of some configuration registers, comprises following information in the configuration register: all internal memory particle numberings that comprised in internal memory particle numbering, groups of grains numbering and the present granule group.
Preferably, accessing operation comprises: read operation and write operation.
According to a further aspect in the invention, a kind of internal storage access device is provided, this device comprises: receiving element, be used to receive access request, and access request is the request of access memory; Addressed location, being used for according to access request is the unit access memory with the groups of grains, and wherein, groups of grains is through dividing into groups the internal memory particle to obtain, and each groups of grains comprises one or more internal memory particles.
Preferably, addressed location comprises: acquisition module, be used for obtaining the memory address information that access request is carried, and wherein, memory address information comprises the particle address, and the particle address is used in reference to the groups of grains to required visit; Enquiry module, the groups of grains corresponding according to the particle address lookup; Access modules, being used for the groups of grains that inquires is the unit access memory.
Preferably, access modules comprises: send submodule, be used for access request is sent to and corresponding command formation of particle address and storage; Implementation sub-module is used for reading the indicated accessing operation of access request that also fill order formation is stored.
Preferably; This device also comprises: dispensing unit; The quantity of the internal memory particle that is used for being comprised through configuration register configuration set groups of grains; Wherein, groups of configuration registers is made up of some configuration registers, comprises following information in the configuration register: all internal memory particles that comprised in internal memory particle numbering, groups of grains numbering and the present granule group.
In the present invention, with the form reorganization particle of groups of grains, each groups of grains can have one or more internal memory particles, and Memory Controller Hub can be operated each internal memory groups of grains separately like this.Because a groups of grains is only operated in each internal memory order, makes each groups of grains be in different state.When the page or leaf conflict took place on a groups of grains in a memory access request, Memory Controller Hub only carries out page or leaf to this groups of grains to be opened and the page or leaf shutoff operation, thereby plays the purpose of saving power consumption.In addition, a Rank is divided into a plurality of groups of grains, and the Bank of each groups of grains opens different pages or leaves; Can effectively reduce the page or leaf conflict; Solved in the correlation technique in the process of carrying out internal storage access because when the page or leaf conflict takes place, all the internal memory particles on the same Rank all need carry out page or leaf to be opened with the page or leaf shutoff operation and causes the big problem of power consumption, reaches the effect that reduces power consumption of internal memory; Further, promoted the performance of memory system.
Description of drawings
Accompanying drawing described herein is used to provide further understanding of the present invention, constitutes the application's a part, and illustrative examples of the present invention and explanation thereof are used to explain the present invention, do not constitute improper qualification of the present invention.In the accompanying drawings:
Fig. 1 is a kind of structural representation according to the memory system of correlation technique;
Fig. 2 is a kind of synoptic diagram according to internal memory particle control in the correlation technique;
Fig. 3 is a kind of preferred process flow diagram according to the memory pool access method of the embodiment of the invention;
Fig. 4 is the preferred synoptic diagram according to a kind of groups of grains of the embodiment of the invention;
Fig. 5 is the synoptic diagram of traditional address zoned format;
Fig. 6 is the synoptic diagram according to the address zoned format of the embodiment of the invention;
Fig. 7 is the preferred synoptic diagram according to the another kind of groups of grains of the embodiment of the invention;
Fig. 8 is the preferred synoptic diagram according to the another kind of groups of grains of the embodiment of the invention;
Fig. 9 is a kind of preferred construction synoptic diagram according to the internal storage access device of the embodiment of the invention;
Figure 10 is the another kind of preferred construction synoptic diagram according to the internal storage access device of the embodiment of the invention; And
Figure 11 is another the preferred construction synoptic diagram according to the internal storage access device of the embodiment of the invention.
Embodiment
Hereinafter will and combine embodiment to specify the present invention with reference to accompanying drawing.Need to prove that under the situation of not conflicting, embodiment and the characteristic among the embodiment among the application can make up each other.
The invention provides a kind of memory pool access method, as shown in Figure 3 specifically, this method comprises:
S302 receives access request, and access request is the request of access memory; Preferably, when needs to certain data line conduct interviews (as, read certain data line or write data) time, send the access request of request access memory to Memory Controller Hub.
S304 is the unit access memory with the groups of grains according to access request, and wherein, groups of grains is through dividing into groups the internal memory particle to obtain, and each groups of grains comprises one or more internal memory particles.Preferably, the internal memory particle is divided into several groups of grains, each groups of grains comprises one or more internal memory particles.Specifically, Fig. 4 illustrates a kind of a kind of synoptic diagram with the memory group that obtains after the grouping of internal memory particle, in Fig. 4; A Rank is made up of 4 internal memory particles, and reference numeral is internal memory particle 0, internal memory particle 1, internal memory particle 2 and internal memory particle 3, and per two particles are formed a groups of grains; The numbering of these two groups of grains is followed successively by a, b, and their corresponding internal memory particle numberings are followed successively by (0; 1), (2,3).Can control each internal memory groups of grains separately in order to make Memory Controller Hub.The present invention provides special operation circuit for each groups of grains.When being 1, the groups of grains at internal memory particle 1 place is groups of grains a, thereby the memory access request is read or write operation groups of grains a when the value of MR (the groups of grains numbering that the value of MR will be visited for this memory access request).Through this form, the present invention has reached division variable grain group, and controls the purpose of these groups of grains respectively.
Above-mentioned preferred embodiment in, reorganize particle with the form of groups of grains, each groups of grains can have one or more internal memory particles, Memory Controller Hub can be operated each internal memory groups of grains separately like this.Because a groups of grains is only operated in each internal memory order, makes each groups of grains be in different state.When the page or leaf conflict took place on a groups of grains in a memory access request, Memory Controller Hub only carries out page or leaf to this groups of grains to be opened and the page or leaf shutoff operation, thereby plays the purpose of saving power consumption.In addition, a Rank is divided into a plurality of groups of grains, and the Bank of each groups of grains opens different pages or leaves; Can effectively reduce the page or leaf conflict; Solved in the correlation technique in the process of carrying out internal storage access because when the page or leaf conflict takes place, all the internal memory particles on the same Rank all need carry out page or leaf to be opened with the page or leaf shutoff operation and causes the big problem of power consumption, reaches the effect that reduces power consumption of internal memory; Further, promoted the performance of memory system.
Also providing a kind of in the present embodiment is the preferred scheme of unit access memory with the groups of grains according to access request; Specifically; This scheme comprises the steps: to obtain the memory address information of carrying in the access request; Wherein, memory address information comprises the particle address, and the particle address is used in reference to the groups of grains to required visit; The groups of grains corresponding according to the particle address lookup; With the groups of grains that inquires is the unit operation that conducts interviews.Specifically, after receiving memory access request, at first obtain the memory address information of carrying in the access request; Traditional address zoned format is as shown in Figure 5, and memory address can be divided into CS (Chip Select is used to select Rank); ROW (row address is used for selecting row), BA (Bank address; Be used to select Bank), COL (column address is used for selecting row).According to this address dividing mode, the memory access request is the unit access memory with Rank.The present invention proposes a kind of novel way to manage.As shown in Figure 6, memory address can be divided into CS (Chip Select is used to select Rank), MR (the particle address is used to select groups of grains), ROW (row address is used for selecting row), BA (the Bank address is used to select Bank), COL (column address is used for selecting row).Wherein, the groups of grains numbering that will visit for this memory access request of the value of MR.Confirming the groups of grains that the memory address of required visit belongs to according to the value of MR, is that unit carries out internal storage access with the groups of grains.Carry out bright specifically below for example:
Suppose that Addr represents the memory access address, original Address Mapping need be translated as the Rank address from high to low successively with the memory access address, row address, and the Bank address, column address and byte address are formed.For example, for 64 memory chips, getting its Rank address has two, and row address is 14, and the Bank address is 3, and column address is 10.Then Addr [2:0] is a byte address, and Addr [12:3] is a column address, and Addr [15:13] is the Bank address, and Addr [29:16] is a row address, and Addr [31:30] is the Rank address.
In the present invention, need new address bit to distinguish each particle of forming Rank.For example, for 64 memory chips, getting its Rank address has two, and row address is 14, and the Bank address is 3, and column address is 10.If it is divided into four passages, then Addr [0] is a byte address, and Addr [10:1] is a column address, and Addr [13:11] is the Bank address, and Addr [27:14] is a row address, and Addr [29:28] is the particle address, and Addr [31:30] is the Rank address.Memory Controller Hub mails to corresponding command queue according to the particle address with the memory access request.Command queue is responsible for storing the memory access request of corresponding particle.
The present invention also optimizes said method, proposed a kind of with the groups of grains that inquires be unit conduct interviews the operation preferred scheme, particularly, this scheme comprise the steps: with access request be sent to the corresponding command formation of particle address in; Read the indicated accessing operation of storing in the also fill order formation of access request.Specifically, Memory Controller Hub obtains the internal memory particle address of required visit according to the memory address information of carrying in the access request, and access request is sent to and the corresponding command formation of particle address, and command queue is responsible for storing the access request of corresponding particle.The situation that is one group of groups of grains with single internal memory particle below illustrates order transmission and data transmission procedure.
Order is sent: original Memory Controller Hub is controlled each Rank separately, and Memory Controller Hub must increase each particle control line among the present invention, for Memory Controller Hub designs each particle that Rank is formed in control separately.In addition, Memory Controller Hub for each particle increased independent state machine (Status Machine, SM).The same with original Rank state machine, the control information of graininess machine record internal memory particle is sent corresponding internal memory order to the internal memory particle.
Data transmission: each particle can be simultaneously respectively to Memory Controller Hub transmission data, the memory access request that these data are corresponding different respectively.The reception and the transmission of graininess machine control data.For reading the memory access request, the data that the graininess machine is passed the internal memory particle back are sent to the corresponding command formation, and command queue is responsible for these data are turned back to memory interface.For writing the memory access request, the graininess machine is sent to the internal memory particle with write data.
In the present invention, the internal memory numbers of particles in each groups of grains can be identical, also can be different.Particularly, as shown in Figure 7, a Rank is made up of 4 particles, and particle 0 is formed a groups of grains with particle 1, and particle 2 is formed a groups of grains respectively with particle 3.In order to make Memory Controller Hub can control each internal memory groups of grains separately, the present invention provides special operation circuit for each groups of grains.The numbering of these three groups of grains is followed successively by a, b, and c, their corresponding internal memory particle numberings are followed successively by (0,1), 2,3.According to the groups of grains address zoned format that technical scheme 1 proposes, when the value of MR was 0 or 1, the memory access request was read or write operation groups of grains a.When the value of MR was respectively 2 or 3, the memory access request was read and write perhaps write operation to groups of grains b and groups of grains c respectively.Through this form, the present invention has reached division variable grain group, and numbers of particles is different in each groups of grains, and controls the purpose of these groups of grains respectively.In practical application, the memory access bandwidth demand of application program is different.The higher application program of bandwidth demand can be used the groups of grains of being made up of a plurality of internal memory particles, and the application program that bandwidth demand is lower can be used the groups of grains of being made up of less particle.
The present invention has also carried out further optimization to said method; With the purpose of the quantity that reaches the internal memory particle that the dynamic-configuration groups of grains comprised, in order to achieve the above object, particularly; The quantity of the internal memory particle that is comprised through configuration register configuration set groups of grains; Wherein, groups of configuration registers is made up of some configuration registers, comprises following information in the configuration register: all internal memory particle numberings that comprised in internal memory particle numbering, groups of grains numbering and the present granule group.Further specify below in conjunction with concrete example:
Fig. 8 illustrates a kind of internal memory granulation tissue, is made up of 4 particles, supposes to have among Fig. 84 groups of grains, and a particle is arranged in each groups of grains, and each internal memory particle all has special-purpose control route.Preferably, in Memory Controller Hub, increase groups of configuration registers, this registers group is made up of a plurality of configuration registers, shown in table one:
Table one
| The particle numbering | The groups of grains numbering | All particle numberings in the groups of grains |
Each register is made up of three parts: particle numbering, the groups of grains numbering that this particle is corresponding, all particle numberings in this groups of grains.In the internal storage access process, Memory Controller Hub is based on this configuration register of the value of MR inquiry, knows all particles numberings in groups of grains numbering that this memory access request need visit and the groups of grains based on the value of MR, and all particles in this groups of grains are visited in the memory access request.Table two, table three and table four have been introduced Fig. 4 respectively for example, the settings of particle numbering and groups of grains numbering corresponding relation among Fig. 7 and Fig. 8.Table two, table three and table four are following:
Table two
| The particle numbering | The groups of grains numbering | All particle numberings in the groups of grains |
| 0 | a | 0，1 |
| 1 | a | 0，1 |
| 2 | |
2，3 |
| 3 | |
2，3 |
Table three
| The particle numbering | The groups of grains numbering | All particle numberings in the groups of grains |
| 0 | a | 0，1 |
| 1 | a | 0，1 |
| 2 | |
2 |
| 3 | |
3 |
Table four
| The particle numbering | The groups of grains numbering | All particle numberings in the groups of grains |
| 0 | a | 0 |
| 1 | |
1 |
| 2 | |
2 |
| 3 | |
3 |
In table two, when MR is 0, this memory access request visit memory access group a, memory access group a is made up of internal memory particle 0 and internal memory particle 1, and this memory access request is based on Query Result visit particle 0 and particle 1; In table three, when MR is 0, this memory access request visit memory access group a, memory access group a is made up of internal memory particle 0 and internal memory particle 1, and this memory access request is based on Query Result visit particle 0 and particle 1; In table four, when MR is 0, this memory access request visit memory access group a, memory access group a is made up of internal memory particle 0, and this memory access request is based on Query Result visit particle 0.Can realize internal memory number of particles in the dynamic-configuration groups of grains according to the method described above, in practical application, the memory access bandwidth demand of application program is different.The higher application program of bandwidth demand can be used the groups of grains of being made up of a plurality of internal memory particles, and the application program that bandwidth demand is lower can be used the groups of grains of being made up of less particle.
Based on the memory pool access method that the foregoing description 1 is put down in writing, the present invention also provides a kind of internal storage access device, and is as shown in Figure 9 specifically, and this device comprises: receiving element 902, be used to receive access request, and access request is the request of access memory; Preferably, when needs to certain data line conduct interviews (as, read certain data line or write data) time, send the access request of request access memory to Memory Controller Hub.Addressed location 904, being used for according to access request is the unit access memory with the groups of grains, and wherein, groups of grains is through dividing into groups the internal memory particle to obtain, and each groups of grains comprises one or more internal memory particles.Specifically, Fig. 4 illustrates a kind of a kind of synoptic diagram with the memory group that obtains after the grouping of internal memory particle, in Fig. 4; A Rank is made up of 4 internal memory particles, and reference numeral is internal memory particle 0, internal memory particle 1, internal memory particle 2 and internal memory particle 3, and per two particles are formed a groups of grains; The numbering of these two groups of grains is followed successively by a, b, and their corresponding internal memory particle numberings are followed successively by (0; 1), (2,3).Can control each internal memory groups of grains separately in order to make Memory Controller Hub.The present invention provides special operation circuit for each groups of grains.When being 1, the groups of grains at internal memory particle 1 place is groups of grains a, thereby the memory access request is read or write operation groups of grains a when the value of MR (the groups of grains numbering that the value of MR will be visited for this memory access request).Through this form, the present invention has reached division variable grain group, and controls the purpose of these groups of grains respectively.
Above-mentioned preferred embodiment in, reorganize particle with the form of groups of grains, each groups of grains can have one or more internal memory particles, Memory Controller Hub can be operated each internal memory groups of grains separately like this.Because a groups of grains is only operated in each internal memory order, makes each groups of grains be in different state.When the page or leaf conflict took place on a groups of grains in a memory access request, Memory Controller Hub only carries out page or leaf to this groups of grains to be opened and the page or leaf shutoff operation, thereby plays the purpose of saving power consumption.In addition, a Rank is divided into a plurality of groups of grains, and the Bank of each groups of grains opens different pages or leaves; Can effectively reduce the page or leaf conflict; Solved in the correlation technique in the process of carrying out internal storage access because when the page or leaf conflict takes place, all the internal memory particles on the same Rank all need carry out page or leaf to be opened with the page or leaf shutoff operation and causes the big problem of power consumption, reaches the effect that reduces power consumption of internal memory; Further, promoted the performance of memory system.
The present invention also improves above-mentioned addressed location 904, to realize that root is the unit access memory with the groups of grains according to access request, for realizing above-mentioned purpose; Specifically, shown in figure 10, addressed location 904 comprises: acquisition module 1002; Be used for obtaining the memory address information that access request is carried; Wherein, memory address information comprises the particle address, and the particle address is used in reference to the groups of grains to required visit; Enquiry module 1004 is used for the groups of grains corresponding according to the particle address lookup; Access modules 1006, being used for the groups of grains that inquires is the unit operation that conducts interviews.
Specifically, after receiving memory access request, at first obtain the memory address information of carrying in the access request; Traditional address zoned format is as shown in Figure 5, and memory address can be divided into CS (Chip Select is used to select Rank); ROW (row address is used for selecting row), BA (Bank address; Be used to select Bank), COL (column address is used for selecting row).According to this address dividing mode, the memory access request is the unit access memory with Rank.The present invention proposes a kind of novel way to manage.As shown in Figure 6, memory address can be divided into CS (Chip Select is used to select Rank), MR (the particle address is used to select groups of grains), ROW (row address is used for selecting row), BA (the Bank address is used to select Bank), COL (column address is used for selecting row).Wherein, the groups of grains numbering that will visit for this memory access request of the value of MR.Confirming the groups of grains that the memory address of required visit belongs to according to the value of MR, is that unit carries out internal storage access with the groups of grains.Carry out bright specifically below for example:
Suppose that Addr represents the memory access address, original Address Mapping need be translated as the Rank address from high to low successively with the memory access address, row address, and the Bank address, column address and byte address are formed.For example, for 64 memory chips, getting its Rank address has two, and row address is 14, and the Bank address is 3, and column address is 10.Then Addr [2:0] is a byte address, and Addr [12:3] is a column address, and Addr [15:13] is the Bank address, and Addr [29:16] is a row address, and Addr [31:30] is the Rank address.
In the present invention, need new address bit to distinguish each particle of forming Rank.For example, for 64 memory chips, getting its Rank address has two, and row address is 14, and the Bank address is 3, and column address is 10.If it is divided into four passages, then Addr [0] is a byte address, and Addr [10:1] is a column address, and Addr [13:11] is the Bank address, and Addr [27:14] is a row address, and Addr [29:28] is the particle address, and Addr [31:30] is the Rank address.Memory Controller Hub mails to corresponding command queue according to the particle address with the memory access request.Command queue is responsible for storing the memory access request of corresponding particle.
Preferably, shown in figure 11, access modules 1006 comprises: send submodule 1102, be used for access request is sent to and the corresponding command formation of particle address; Implementation sub-module 1104 is used for reading the indicated accessing operation of access request that also fill order formation is stored.
Specifically, Memory Controller Hub obtains the internal memory particle address of required visit according to the memory address information of carrying in the access request, and access request is sent to and the corresponding command formation of particle address, and command queue is responsible for storing the access request of corresponding particle.The situation that is one group of groups of grains with single internal memory particle below illustrates order transmission and data transmission procedure.
Order is sent: original Memory Controller Hub is controlled each Rank separately, and Memory Controller Hub must increase each particle control line among the present invention, for Memory Controller Hub designs each particle that Rank is formed in control separately.In addition, Memory Controller Hub for each particle increased independent state machine (Status Machine, SM).The same with original Rank state machine, the control information of graininess machine record internal memory particle is sent corresponding internal memory order to the internal memory particle.
Data transmission: each particle can be simultaneously respectively to Memory Controller Hub transmission data, the memory access request that these data are corresponding different respectively.The reception and the transmission of graininess machine control data.For reading the memory access request, the data that the graininess machine is passed the internal memory particle back are sent to the corresponding command formation, and command queue is responsible for these data are turned back to memory interface.For writing the memory access request, the graininess machine is sent to the internal memory particle with write data.
In the present invention, the internal memory numbers of particles in each groups of grains can be identical, also can be different.Particularly, as shown in Figure 7, a Rank is made up of 4 particles, and particle 0 is formed a groups of grains with particle 1, and particle 2 is formed a groups of grains respectively with particle 3.In order to make Memory Controller Hub can control each internal memory groups of grains separately, the present invention provides special operation circuit for each groups of grains.The numbering of these three groups of grains is followed successively by a, b, and c, their corresponding internal memory particle numberings are followed successively by (0,1), 2,3.According to the groups of grains address zoned format that technical scheme 1 proposes, when the value of MR was 0 or 1, the memory access request was read or write operation groups of grains a.When the value of MR was respectively 2 or 3, the memory access request was read and write perhaps write operation to groups of grains b and groups of grains c respectively.Through this form, the present invention has reached division variable grain group, and numbers of particles is different in each groups of grains, and controls the purpose of these groups of grains respectively.In practical application, the memory access bandwidth demand of application program is different.The higher application program of bandwidth demand can be used the groups of grains of being made up of a plurality of internal memory particles, and the application program that bandwidth demand is lower can be used the groups of grains of being made up of less particle.
The present invention also optimizes said apparatus; Particularly; This device also comprises: dispensing unit, and the quantity of the internal memory particle that is used for being comprised through configuration register configuration set groups of grains, wherein; Groups of configuration registers is made up of some configuration registers, comprises following information in the configuration register: all internal memory particles that comprised in internal memory particle numbering, groups of grains numbering and the present granule group.。Further specify below in conjunction with concrete example:
Fig. 8 illustrates a kind of internal memory granulation tissue, is made up of 4 particles, supposes to have among Fig. 84 groups of grains, and a particle is arranged in each groups of grains, and each internal memory particle all has special-purpose control route.Preferably, in Memory Controller Hub, increase groups of configuration registers, this registers group is made up of a plurality of configuration registers; Shown in table; Each register is made up of three parts: particle numbering, the groups of grains numbering that this particle is corresponding, all particle numberings in this groups of grains.In the internal storage access process, Memory Controller Hub is based on this configuration register of the value of MR inquiry, knows all particles numberings in groups of grains numbering that this memory access request need visit and the groups of grains based on the value of MR, and all particles in this groups of grains are visited in the memory access request.Table two, table three and table four have been introduced Fig. 4 respectively for example, the settings of particle numbering and groups of grains numbering corresponding relation among Fig. 7 and Fig. 8.In table two, when MR is 0, this memory access request visit memory access group a, memory access group a is made up of internal memory particle 0 and internal memory particle 1, and this memory access request is based on Query Result visit particle 0 and particle 1; In table three, when MR is 0, this memory access request visit memory access group a, memory access group a is made up of internal memory particle 0 and internal memory particle 1, and this memory access request is based on Query Result visit particle 0 and particle 1; In table four, when MR is 0, this memory access request visit memory access group a, memory access group a is made up of internal memory particle 0, and this memory access request is based on Query Result visit particle 0.Can realize internal memory number of particles in the dynamic-configuration groups of grains according to the method described above, in practical application, the memory access bandwidth demand of application program is different.The higher application program of bandwidth demand can be used the groups of grains of being made up of a plurality of internal memory particles, and the application program that bandwidth demand is lower can be used the groups of grains of being made up of less particle.
From above description, can find out that with the form reorganization particle of groups of grains, each groups of grains can have one or more internal memory particles, Memory Controller Hub can be operated each internal memory groups of grains separately like this.Because a groups of grains is only operated in each internal memory order, makes each groups of grains be in different state.When the page or leaf conflict took place on a groups of grains in a memory access request, Memory Controller Hub only carries out page or leaf to this groups of grains to be opened and the page or leaf shutoff operation, thereby plays the purpose of saving power consumption.In addition, a Rank is divided into a plurality of groups of grains, and the Bank of each groups of grains opens different pages or leaves; Can effectively reduce the page or leaf conflict; Solved in the correlation technique in the process of carrying out internal storage access because when the page or leaf conflict takes place, all the internal memory particles on the same Rank all need carry out page or leaf to be opened with the page or leaf shutoff operation and causes the big problem of power consumption, reaches the effect that reduces power consumption of internal memory; Further, promoted the performance of memory system.
Obviously, it is apparent to those skilled in the art that above-mentioned each module of the present invention or each step can realize with the general calculation device; They can concentrate on the single calculation element; Perhaps be distributed on the network that a plurality of calculation element forms, alternatively, they can be realized with the executable program code of calculation element; Thereby; Can they be stored in the memory storage and carry out, and in some cases, can carry out step shown or that describe with the order that is different from here by calculation element; Perhaps they are made into each integrated circuit modules respectively, perhaps a plurality of modules in them or step are made into the single integrated circuit module and realize.Like this, the present invention is not restricted to any specific hardware and software combination.
The above is merely the preferred embodiments of the present invention, is not limited to the present invention, and for a person skilled in the art, the present invention can have various changes and variation.All within spirit of the present invention and principle, any modification of being done, be equal to replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (10)
1. a memory pool access method is characterized in that, comprising:
Receive access request, said access request is the request of access memory;
According to said access request is the unit access memory with the groups of grains, and wherein, said groups of grains is through dividing into groups the internal memory particle to obtain, and each groups of grains comprises one or more internal memory particles.
2. method according to claim 1 is characterized in that, is that the step of unit access memory comprises with the groups of grains according to said access request:
Obtain the memory address information of carrying in the access request, wherein, said memory address information comprises the particle address, and said particle address is used in reference to the groups of grains to required visit;
The groups of grains corresponding according to said particle address lookup;
With the groups of grains that inquires is the unit operation that conducts interviews.
3. method according to claim 2 is characterized in that, with the groups of grains that inquires be unit conduct interviews the operation step comprise:
With said access request be sent to the corresponding command formation of said particle address in;
Read and carry out the indicated accessing operation of storing in the said command queue of access request.
4. method according to claim 1 is characterized in that, the internal memory number of particles that is comprised in the said groups of grains of dynamic-configuration.
5. method according to claim 4; It is characterized in that; Through the internal memory number of particles that is comprised in the said groups of grains of groups of configuration registers dynamic-configuration; Wherein, said groups of configuration registers is made up of some configuration registers, comprises following information in the said configuration register: all internal memory particle numberings that comprised in internal memory particle numbering, groups of grains numbering and the present granule group.
6. according to claim 2 or 3 described methods, it is characterized in that said accessing operation comprises: read operation and write operation.
7. an internal storage access device is characterized in that, comprising:
Receiving element is used to receive access request, and said access request is the request of access memory;
Addressed location, being used for according to said access request is the unit access memory with the groups of grains, and wherein, said groups of grains is through dividing into groups the internal memory particle to obtain, and each groups of grains comprises one or more internal memory particles.
8. device according to claim 7 is characterized in that, said addressed location comprises:
Acquisition module is used for obtaining the memory address information that access request is carried, and wherein, said memory address information comprises the particle address, and said particle address is used in reference to the groups of grains to required visit;
Enquiry module is used for the groups of grains corresponding according to said particle address lookup;
Access modules, being used for the groups of grains that inquires is the unit access memory.
9. device according to claim 8 is characterized in that, said access modules comprises:
Send submodule, be used for said access request is sent to and corresponding command formation of said particle address and storage;
Implementation sub-module is used for carrying out the indicated accessing operation of access request that said command queue stores.
10. device according to claim 7 is characterized in that, also comprises:
Dispensing unit; The quantity of the internal memory particle that is used for being comprised through the said groups of grains of configuration register configuration set; Wherein, Said groups of configuration registers is made up of some configuration registers, comprises following information in the said configuration register: all internal memory particles that comprised in internal memory particle numbering, groups of grains numbering and the present granule group.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210146754.9A CN102681946B (en) | 2012-05-11 | 2012-05-11 | Memory access method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210146754.9A CN102681946B (en) | 2012-05-11 | 2012-05-11 | Memory access method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102681946A true CN102681946A (en) | 2012-09-19 |
| CN102681946B CN102681946B (en) | 2015-03-11 |
Family
ID=46813908
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210146754.9A Active CN102681946B (en) | 2012-05-11 | 2012-05-11 | Memory access method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102681946B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104252422A (en) * | 2013-06-26 | 2014-12-31 | 华为技术有限公司 | Memory access method and memory controller |
| CN104699638A (en) * | 2013-12-05 | 2015-06-10 | 华为技术有限公司 | Memory access method and memory access device |
| CN104851454A (en) * | 2014-02-13 | 2015-08-19 | 华为技术有限公司 | Method and device for processing control signal and variable memory |
| CN105051729A (en) * | 2013-02-01 | 2015-11-11 | 起元技术有限责任公司 | Data records selection |
| WO2016082763A1 (en) * | 2014-11-28 | 2016-06-02 | 华为技术有限公司 | Memory access method, relevant device and system |
| CN106776173A (en) * | 2016-12-15 | 2017-05-31 | 郑州云海信息技术有限公司 | A kind of internal-memory detection method and device |
| WO2018094620A1 (en) * | 2016-11-23 | 2018-05-31 | 华为技术有限公司 | Memory allocation method and apparatus |
| CN108446241A (en) * | 2017-02-16 | 2018-08-24 | 华为技术有限公司 | Memory pool access method and device |
| CN109284231A (en) * | 2018-07-24 | 2019-01-29 | 江苏微锐超算科技有限公司 | Processing method, device and the Memory Controller Hub of memory access request |
| CN109408407A (en) * | 2018-10-19 | 2019-03-01 | 龙芯中科技术有限公司 | The distribution method and device of display memory |
| US11068540B2 (en) | 2018-01-25 | 2021-07-20 | Ab Initio Technology Llc | Techniques for integrating validation results in data profiling and related systems and methods |
| US11487732B2 (en) | 2014-01-16 | 2022-11-01 | Ab Initio Technology Llc | Database key identification |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030115403A1 (en) * | 2001-12-19 | 2003-06-19 | Bouchard Gregg A. | Dynamic random access memory system with bank conflict avoidance feature |
| JP2005038454A (en) * | 2003-07-15 | 2005-02-10 | Renesas Technology Corp | Memory module and memory system |
| CN101021814A (en) * | 2007-03-16 | 2007-08-22 | 华为技术有限公司 | Storage and polling method and storage controller and polling system |
| CN102402490A (en) * | 2010-09-16 | 2012-04-04 | 苹果公司 | Multi-ported memory controller with ports associated with traffic classes |
-
2012
- 2012-05-11 CN CN201210146754.9A patent/CN102681946B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030115403A1 (en) * | 2001-12-19 | 2003-06-19 | Bouchard Gregg A. | Dynamic random access memory system with bank conflict avoidance feature |
| JP2005038454A (en) * | 2003-07-15 | 2005-02-10 | Renesas Technology Corp | Memory module and memory system |
| CN101021814A (en) * | 2007-03-16 | 2007-08-22 | 华为技术有限公司 | Storage and polling method and storage controller and polling system |
| CN102402490A (en) * | 2010-09-16 | 2012-04-04 | 苹果公司 | Multi-ported memory controller with ports associated with traffic classes |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105051729A (en) * | 2013-02-01 | 2015-11-11 | 起元技术有限责任公司 | Data records selection |
| US11163670B2 (en) | 2013-02-01 | 2021-11-02 | Ab Initio Technology Llc | Data records selection |
| CN104252422A (en) * | 2013-06-26 | 2014-12-31 | 华为技术有限公司 | Memory access method and memory controller |
| WO2014206230A1 (en) * | 2013-06-26 | 2014-12-31 | 华为技术有限公司 | Memory access method and memory controller |
| CN104699638A (en) * | 2013-12-05 | 2015-06-10 | 华为技术有限公司 | Memory access method and memory access device |
| CN104699638B (en) * | 2013-12-05 | 2017-11-17 | 华为技术有限公司 | Memory pool access method and internal storage access device |
| US11487732B2 (en) | 2014-01-16 | 2022-11-01 | Ab Initio Technology Llc | Database key identification |
| CN104851454A (en) * | 2014-02-13 | 2015-08-19 | 华为技术有限公司 | Method and device for processing control signal and variable memory |
| WO2016082763A1 (en) * | 2014-11-28 | 2016-06-02 | 华为技术有限公司 | Memory access method, relevant device and system |
| CN108604206A (en) * | 2016-11-23 | 2018-09-28 | 华为技术有限公司 | A kind of memory allocation method and equipment |
| WO2018094620A1 (en) * | 2016-11-23 | 2018-05-31 | 华为技术有限公司 | Memory allocation method and apparatus |
| US11232031B2 (en) | 2016-11-23 | 2022-01-25 | Huawei Technologies Co., Ltd. | Allocation of memory ranks based on access traffic |
| CN106776173A (en) * | 2016-12-15 | 2017-05-31 | 郑州云海信息技术有限公司 | A kind of internal-memory detection method and device |
| CN108446241A (en) * | 2017-02-16 | 2018-08-24 | 华为技术有限公司 | Memory pool access method and device |
| CN108446241B (en) * | 2017-02-16 | 2020-12-08 | 华为技术有限公司 | Memory access method and device |
| US11068540B2 (en) | 2018-01-25 | 2021-07-20 | Ab Initio Technology Llc | Techniques for integrating validation results in data profiling and related systems and methods |
| CN109284231A (en) * | 2018-07-24 | 2019-01-29 | 江苏微锐超算科技有限公司 | Processing method, device and the Memory Controller Hub of memory access request |
| CN109284231B (en) * | 2018-07-24 | 2023-05-23 | 江苏微锐超算科技有限公司 | Memory access request processing method and device and memory controller |
| CN109408407A (en) * | 2018-10-19 | 2019-03-01 | 龙芯中科技术有限公司 | The distribution method and device of display memory |
| CN109408407B (en) * | 2018-10-19 | 2021-08-03 | 龙芯中科技术股份有限公司 | Display memory allocation method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102681946B (en) | 2015-03-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102681946B (en) | Memory access method and device | |
| CN102177550B (en) | Independently controlled virtual memory devices in memory modules | |
| US9965222B1 (en) | Software mode register access for platform margining and debug | |
| CN100511119C (en) | Method for realizing shadow stack memory on picture and circuit thereof | |
| CN102356385B (en) | Memory access controller, systems, and methods for optimizing memory access times | |
| US11042304B2 (en) | Determining a transfer rate for channels of a memory system | |
| US8639891B2 (en) | Method of operating data storage device and device thereof | |
| CN101836194B (en) | An optimal solution to control data channels | |
| CN101719104B (en) | Control system and control method of synchronous dynamic memory | |
| CN104252422A (en) | Memory access method and memory controller | |
| CN102999441A (en) | Fine granularity memory access method | |
| CN104714898B (en) | A kind of distribution method and device of Cache | |
| US6412039B1 (en) | Cross-bank, cross-page data accessing and controlling system | |
| US20240036754A1 (en) | Adjustable access energy and access latency memory system and devices | |
| CN103019624A (en) | Phase change memory device | |
| CN116257191B (en) | Memory controller, memory component, electronic device and command scheduling method | |
| CN102163320B (en) | Configurable memory management unit (MMU) circuit special for image processing | |
| CN102243611A (en) | Data storage method and system | |
| CN102591816A (en) | Multichannel Nandflash storage system | |
| KR20110066526A (en) | Multi-port memory controller and multi-port cache | |
| TWI447728B (en) | Controlling method and controller for dram | |
| CN107066208B (en) | Asymmetric read-write method of external memory device and NVM (non-volatile memory) external memory device | |
| CN102073604B (en) | Method, device and system for controlling read and write of synchronous dynamic memory | |
| CN101464838B (en) | Data management method and solid state storage system | |
| TWI721660B (en) | Device and method for controlling data reading and writing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CP03 | Change of name, title or address | ||
| CP03 | Change of name, title or address |
Address after: 100095 Building 2, Longxin Industrial Park, Zhongguancun environmental protection technology demonstration park, Haidian District, Beijing Patentee after: Loongson Zhongke Technology Co.,Ltd. Address before: 100190 No. 10 South Road, Zhongguancun Academy of Sciences, Haidian District, Beijing Patentee before: LOONGSON TECHNOLOGY Corp.,Ltd. |