Memory device and operation method thereof

By adopting two interlaced configurations of memory and instruction pre-read technology in memory, the problem of multi-cycle delay in memory during access operations is solved, improving processor performance and reducing hardware costs.

CN113110878BActive Publication Date: 2024-11-08REALTEK SEMICON CORP
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202010022523.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-01-09
Publication Date
2024-11-08
Estimated Expiration
2040-01-09

AI Technical Summary

Technical Problem

When existing memory is accessed, multiple frequency cycles are required, resulting in an increase in the processor's waiting time, and increasing memory bandwidth or adding write buffers to make up for speed differences will increase hardware cost and reduce performance.

Method used

Using the interleaved configuration of two memories and pre-read instruction, the reference address of the processing circuit is converted into the actual address through the access circuit, and the data is read in parallel and pre-read, reducing multi-cycle access delays, and avoiding branch instructions. Pause period caused by address misalignment.

Benefits of technology

Effectively eliminate read delay caused by multi-cycle access, reduce the pause period of branch instructions, reduce the processor access time, reduce hardware costs, and improve the processor's performance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113110878B_ABST
    Figure CN113110878B_ABST
Patent Text Reader

Abstract

A memory device and an operation method thereof, comprising: a first memory and a second memory and an access circuit. A reference address of a processing circuit corresponds to an actual address of the first memory and the second memory in an interlaced manner. The access circuit is configured to: receive a read instruction corresponding to a reference read address from the processing circuit to convert the reference read address into an actual read address of the first memory and the second memory; simultaneously read a first set of read data from the first memory and pre-read a second set of read data from the second memory according to the actual read address and the next actual read address; respond to the first set of read data to the processing circuit; and respond to the second set of read data to the processing circuit when a next read instruction corresponding to a next reference read address is received by the processing circuit and the next reference read address corresponds to a next actual read address.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to a memory technology, and more particularly to a memory device and an operating method thereof. Background Art

[0002] When accessing memory, it often takes multiple frequency cycles to perform. In order to reduce the waiting time and maintain a higher operating frequency, the processor usually adds a cache to make up for the speed difference. However, this approach will greatly increase the area and cost of the processor.

[0003] Another way to compensate is to increase the bandwidth of the memory and read multiple word groups of data at a time. However, when the read instruction encounters an address branch, if the multiple word groups read are not aligned, the target address reading time and the acquisition of the next continuous instruction require additional cycles, reducing the processor access efficiency. In addition, when writing, it is often necessary to add additional write buffers to combine multiple word groups into one write, which increases the hardware cost.

[0004] Therefore, how to design a new memory device and its operating method to solve the above-mentioned deficiencies is an urgent problem to be solved in the industry. Summary of the invention

[0005] The invention summary is intended to provide a simplified summary of the present disclosure so that readers can have a basic understanding of the present disclosure. This invention summary is not a complete overview of the present disclosure, and it is not intended to point out the important / critical components of the embodiments of the present invention or to define the scope of the present invention.

[0006] An object of the present invention is to provide a memory device and an operating method thereof, thereby improving the problems of the prior art.

[0007] To achieve the above-mentioned purpose, a technical solution of the present invention is related to a memory device, comprising: a first memory, a second memory and an access circuit. A plurality of reference addresses of the processing circuit correspond to a plurality of actual addresses of the first memory and the second memory in an interlaced manner. The access circuit is configured to: receive a read instruction corresponding to a reference read address from the processing circuit to convert the reference read address into an actual read address of the first memory and the second memory; simultaneously read a first set of read data from the first of the first memory and the second memory and pre-read a second set of read data from the second memory according to the actual read address and the next actual read address; respond to the first set of read data to the processing circuit; and respond to the second set of read data to the processing circuit when the processing circuit receives a next read instruction corresponding to the next reference read address and the next reference read address corresponds to the next actual read address.

[0008] Another technical solution of the present invention is related to a memory device operation method, including: enabling an access circuit to receive a read instruction corresponding to a reference read address from a processing circuit to convert the reference read address into actual read addresses of a first memory and a second memory, wherein a plurality of reference addresses of the processing circuit correspond to a plurality of actual addresses of the first memory and the second memory in an interlaced manner; enabling the access circuit to simultaneously read a first set of read data from the first of the first memory and the second memory and pre-read a second set of read data from the second according to the actual read address and the next actual read address; enabling the access circuit to respond to the first set of read data to the processing circuit; and enabling the access circuit to respond to the second set of read data to the processing circuit when the processing circuit receives a next read instruction corresponding to a next reference read address and the next reference read address corresponds to a next actual read address.

[0009] The memory device and its operation method of the present invention eliminate the read delay caused by multi-cycle access by setting two memories in parallel reading and pre-reading instructions. In addition, the address configuration of the two memories is staggered and can be controlled separately, so that the branch instruction is not restricted by the address bit alignment, reducing the pause cycle caused by the branch instruction. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to make the above and other objects, features, advantages and embodiments of the present invention more clearly understood, the accompanying drawings are described as follows:

[0011] Figure 1 A block diagram of a computer system according to an embodiment of the present invention;

[0012] Figure 2 A partial comparison table of a reference address of a processing circuit and actual addresses of a first memory and a second memory in one embodiment of the present invention;

[0013] Figure 3 A timing diagram of a memory device performing a read operation in one embodiment of the present invention;

[0014] Figure 4 A timing diagram of a memory device performing a read operation in another embodiment of the present invention;

[0015] Figure 5 A timing diagram of a memory device performing a write operation in one embodiment of the present invention;

[0016] Figure 6 is a flow chart of a memory device operating method according to an embodiment of the present invention; and

[0017] Figure 7 The figure is a flow chart of a memory device operating method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0018] Please refer to Figure 1 . Figure 1 FIG. 1 is a block diagram of a computer system 1 according to an embodiment of the present invention. The computer system 1 includes a processing circuit 100 and a memory device 110 .

[0019] The processing circuit 100 can access the memory device 110 by sending access commands, such as but not limited to a read command RC and a write command WC. For example, when the processing circuit 100 sends a read command RC, read data RD can be read from the memory device 110 according to the corresponding address. When the processing circuit 100 sends a write command WC, write data WD can be written to the memory device 110 according to the corresponding address.

[0020] In one embodiment, the transmission of commands and data between the processing circuit 100 and the memory device 110 may be performed via a bus 120 therebetween.

[0021] It should be noted that the computer system 1 may actually include other components that can interact with the processing circuit 100 and the memory device 110, and is not intended to be construed as Figure 1 Limited to the number of components shown.

[0022] The memory device 110 includes a first memory SRAM1 , a second memory SRAM2 , and an access circuit FET.

[0023] In one embodiment, the first memory SRAM1 and the second memory SRAM2 are both static random access memories, but the present invention is not limited thereto.

[0024] The access circuit FET is configured to access the first memory SRAM1 and the second memory SRAM2 according to the read command RC and the write command WC sent by the processing circuit 100 .

[0025] When the memory device 110 performs a read operation, the access circuit FET receives the read command RC from the processing circuit 100, and converts the reference read address corresponding to the read command RC into the actual read address of the first memory SRAM1 and the second memory SRAM2, so as to read the corresponding data from the first memory SRAM1 and the second memory SRAM2. The memory device 110 further includes a first read buffer BUF1 and a second read buffer BUF2, which respectively temporarily store the data read by the access circuit FET, and then the access circuit FET responds to the data to the processing circuit 100.

[0026] On the other hand, the memory device 110 does not include a write buffer. When the memory device 110 performs a write operation, the access circuit FET receives a write instruction WC from the processing circuit 100, converts the reference write address corresponding to the write instruction WC into an actual write address of the first memory SRAM1 and the second memory SRAM2, and writes the corresponding data into the first memory SRAM1 and the second memory SRAM2 without temporary storage.

[0027] The structure and operation of the memory device 1 will be further described below through more detailed examples.

[0028] The multiple reference addresses of the processing circuit 100 correspond to the multiple actual addresses of the first memory SRAM1 and the second memory SRAM2 in an interlaced manner. More specifically, in one embodiment, the actual address of the Mth first memory corresponds to the reference address of the 2M-1th processing circuit, the actual address of the Mth second memory corresponds to the reference address of the 2Mth processing circuit, and M is a positive integer greater than or equal to 1.

[0029] Please refer to Figure 2 . Figure 2 It is a partial comparison table of the reference address of the processing circuit 100 and the actual addresses of the first memory SRAM1 and the second memory SRAM2 in one embodiment of the present invention.

[0030] like Figure 2 As shown, the actual addresses of the first memory SRAM1 from the first to the fourth are expressed in hexadecimal, which are consecutive 0x0, 0x1, 0x2, and 0x3. The actual addresses of the second memory SRAM2 from the first to the fourth are expressed in hexadecimal, which are consecutive 0x0, 0x1, 0x2, and 0x3. The actual addresses of the first memory SRAM1 and the second memory SRAM2 correspond to a word length, that is, 32 bits of data.

[0031] On the other hand, the processing circuit 100 refers to addresses 1 to 8 in hexadecimal format, which are consecutive 0x0, 0x4, 0x8, 0xC, 0x10, 0x14, 0x18, and 0x1C, respectively.

[0032] Therefore, after the above calculation, the 1st to 4th actual addresses of the first memory SRAM1 correspond to the 1st, 3rd, 5th, and 7th reference addresses of the processing circuit 100, respectively. And the 1st to 4th actual addresses of the second memory SRAM2 correspond to the 2nd, 4th, 6th, and 8th reference addresses of the processing circuit 100, respectively, after the above calculation.

[0033] It should be noted that Figure 2The table shown is only a part of the comparison table. The processing circuit 100 and the first memory SRAM1 and the second memory SRAM2 may actually include more corresponding reference addresses and actual addresses.

[0034] Taking the read operation as an example, when the access circuit FET receives the read command RC from the processing circuit 100 , it converts the corresponding reference read address into the actual read address of the first memory SRAM1 and the second memory SRAM2 .

[0035] Please refer to Figure 3 . Figure 3 FIG. 1 is a timing diagram of a read operation performed by the memory device 110 according to an embodiment of the present invention.

[0036] exist Figure 3 In the figure, CPUADD represents the reference read address corresponding to the read instruction RC of the processing circuit 100, CPUDA represents the data of the processing circuit 100 responded by the access circuit FET, SRAM1 represents the data read from the first memory SRAM1, SRAM2 represents the data read from the second memory SRAM2, BUF1 represents the data temporarily stored in the first read buffer BUF1, and BUF2 represents the data temporarily stored in the second read buffer BUF2.

[0037] like Figure 3 As shown, the signal and data transmission between the processing circuit 100, the first memory SRAM1 and the second memory SRAM2 are all operated according to the clock signal CLK. Among them, the read command RC of the processing circuit 100 is transmitted at least one unit frequency. And the data of the first memory SRAM1 and the second memory SRAM2 are read at two unit frequencies.

[0038] At the unit frequency T0 , the access circuit FET receives the first read command RC1 from the processing circuit 100 via, for example but not limited to, the bus 120 , and the read command RC1 corresponds to the reference read address 0x0.

[0039] The access circuit FET then simultaneously reads the first set of read data RD1 from the first memory SRAM1 and pre-reads the second set of read data RD2 from the second memory SRAM2 according to the actual read address (the actual address 0x0 of the first memory SRAM1, and corresponds to the reference read address 0x0 of the processing circuit 100) and the next actual read address (the actual address 0x0 of the second memory SRAM2, and corresponds to the reference read address 0x4 of the processing circuit 100).

[0040] At the unit frequency T1 , the processing circuit 100 is suspended because the first memory SRAM1 and the second memory SRAM2 require two unit frequencies to read.

[0041] At the unit frequency T2, the first set of read data RD1 and the second set of read data RD2 are temporarily stored in the first read buffer BUF1 and the second read buffer BUF2. At this time, the access circuit FET will respond to the first set of read data RD1 to the processing circuit 100 at a unit frequency.

[0042] It should be noted that, in practice, since the first memory SRAM1 requires two unit frequencies to read, at the unit frequency T2, the first set of read data RD1 can also be selectively responded to directly by the access circuit FET to the processing circuit 100 without being temporarily stored in the first read buffer BUF1. However, the present invention is not limited thereto.

[0043] At the same time, the access circuit FET receives the next read command RC2 from the processing circuit 100. In this embodiment, the next reference read address corresponding to the read command RC2 is 0x4.

[0044] The access circuit FET will determine whether this next reference read address corresponds to the aforementioned next actual read address, so as to pre-read the third group of read data RD3 from the first memory SRAM1 and pre-read the fourth group of read data RD4 from the second memory SRAM2 according to the next two actual read addresses (the actual address 0x1 of the first memory SRAM1, and the corresponding reference read address 0x8 of the processing circuit 100) and the next three actual read addresses (the actual address 0x1 of the second memory SRAM2, and the corresponding reference read address 0xC of the processing circuit 100).

[0045] At the unit frequency T3, since the reference read address corresponding to the read instruction RC2 received by the access circuit FET at the unit frequency T2 is 0x4, which matches the second set of read data RD2 pre-read by the access circuit FET from the second memory SRAM2, the access circuit FET will respond to the second set of read data RD2 from the second read buffer BUF2 to the processing circuit 100.

[0046] At the same time, the access circuit FET receives the next read command RC3 from the processing circuit 100. In this embodiment, the reference read address corresponding to the read command RC3 is 0x8. The access circuit FET determines that the reference read address corresponds to the third set of read data RD3 previously read without performing other actions.

[0047] At the unit frequency T4, since the reference read address corresponding to the read instruction RC3 received by the access circuit FET at the unit frequency T3 is 0x8, which matches the third set of read data RD3 pre-read by the access circuit FET from the first memory SRAM1, the access circuit FET will respond to the third set of read data RD3 from the first read buffer BUF1 to the processing circuit 100.

[0048] Similarly, as mentioned above, in practice, since the first memory SRAM1 requires two unit frequencies to read, at the unit frequency T4, the third set of read data RD3 can also be selectively responded to directly by the access circuit FET to the processing circuit 100 without being temporarily stored in the first read buffer BUF1. However, the present invention is not limited thereto.

[0049] At the same time, the access circuit FET receives the next read command RC4 from the processing circuit 100. In this embodiment, the reference read address corresponding to the read command RC4 is 0xC. The access circuit FET determines that the reference read address corresponds to the fourth set of read data RD4 pre-read above, and pre-reads two sets of read data from the first memory SRAM1 and the second memory SRAM2 according to the next actual read address (the actual address 0x2 of the first memory SRAM1, corresponding to the reference read address 0x10 of the processing circuit 100) and the next actual read address (the actual address 0x2 of the second memory SRAM2, corresponding to the reference read address 0x14 of the processing circuit 100).

[0050] Therefore, when the access circuit FET receives a plurality of read commands RC from the processing circuit 100 that correspond to consecutive reference read addresses, data can be read continuously without interruption until the unit frequency T5.

[0051] At the unit frequency T6, due to the idleness of the processing circuit 100, the access circuit FET does not receive the read command RC from the processing circuit 100, but continues to pre-read two sets of read data. Therefore, when the reference read address of the read command RC received by the access circuit FET at the unit frequency T7 is still continuous with the previous read address, it will benefit from the pre-read mechanism without unnecessary pause.

[0052] Please refer to Figure 4 . Figure 4 FIG. 4 is a timing diagram of a read operation performed by the memory device 110 according to another embodiment of the present invention. Figure 4 The labels and Figure 3 The same, therefore, will not be repeated here. Further, Figure 4 The operation of unit frequency T0-T3 in Figure 3 The unit frequencies T0-T3 in are the same, so they are not repeated here.

[0053] In this embodiment, at the unit frequency T4, the reference read address corresponding to the read instruction RC4 received by the access circuit FET is 0x84, not 0xC. Such a reference read address is not continuous with the reference read address 0x8 corresponding to the previous read instruction RC3. Therefore, the read instruction RC4 is a branch instruction that jumps out of the original corresponding continuous address read.

[0054] Although the first read buffer BUF1 and the second read buffer BUF2 temporarily store the third group of read data RD3 and the fourth group of read data RD4 that are pre-read and correspond to the reference read addresses 0x8 and 0xC respectively, the access circuit FET can only respond to the third group of read data RD3 to the processing circuit 100 according to the read instruction RC3 at the unit frequency T4, but cannot respond to the fourth group of read data RD4 to the processing circuit 100 according to the read instruction RC3 at the next unit frequency T5.

[0055] Therefore, at the unit frequency T4, the access circuit FET will read the fifth set of read data RD5 (corresponding to the reference read address 0x84 of the processing circuit 100) according to the read command RC4, and pre-read the sixth set of read data RD6 (corresponding to the reference read address 0x88 of the processing circuit 100). Figure 2 According to the corresponding relationship in the comparison table shown, the fifth set of read data RD5 will be read from the second memory SRAM2, and the sixth set of read data RD6 will be read from the first memory SRAM1.

[0056] Since reading data requires two unit frequencies, the processing circuit 100 will be suspended at the unit frequency T5.

[0057] At the unit frequency T6 , the access circuit FET receives the read command RC5 from the processing circuit 100 , and the reference read address corresponding to the read command RC5 is 0x88, corresponding to the sixth set of read data RD6 read in advance.

[0058] Therefore, in addition to responding the fifth set of read data RD5 to the processing circuit 100, the access circuit FET also pre-reads the seventh set of read data RD7 and the eighth set of read data RD8 corresponding to the reference read addresses 0x8C and 0x90 respectively from the second memory SRAM2 and the first memory SRAM1. The subsequent operations of each unit frequency are the same as the situation where the reference read addresses of the read instruction are continuous with each other, so they are not repeated here.

[0059] In some technologies, although the access width of a single memory is increased to read two words of data at a time, the target address to be read must be aligned to 64 bits. When a branch instruction is encountered during the memory reading process, and the target address of the branch is not aligned to 64 bits, the reading time of the target address and the next continuous instruction acquisition time of the target address both require two unit frequencies, and the processing circuit must be suspended for multiple cycles.

[0060] The memory device 110 of the present invention eliminates the read delay caused by multi-cycle access by setting two memories in parallel reading and pre-reading instructions. In addition, the two memories are staggered and can be controlled separately, so that branch instructions are not subject to the limitation of 64-bit alignment, reducing the pause cycle caused by branch instructions.

[0061] On the other hand, taking the write operation as an example, when the access circuit FET receives a read instruction WC corresponding to a reference write address from the processing circuit 100, it converts the reference write address into the actual write address of the first memory SRAM1 and the second memory SRAM2 to determine the address of the actual data to be written.

[0062] Please refer to Figure 5 . Figure 5 FIG. 1 is a timing diagram of a memory device 110 performing a write operation according to an embodiment of the present invention.

[0063] like Figure 5 As shown, during the unit frequency T0-T3, the access circuit FET receives write commands WC1-WC4 in sequence. The write commands WC1-WC4 are each received at a unit frequency, and the write commands WC1-WC4 correspond to consecutive reference write addresses 0x0, 0x4, 0x8 and 0xC, respectively.

[0064] The reference write addresses 0x0, 0x4, 0x8 and 0xC correspond to the actual address 0x0 of the first memory SRAM1, the actual address 0x0 of the second memory SRAM2, the actual address 0x1 of the first memory SRAM1 and the actual address 0x1 of the second memory SRAM2 respectively.

[0065] Since the first memory SRAM1 and the second memory SRAM2 require two unit frequencies to be written, and the first memory SRAM1 and the second memory SRAM2 can be accessed independently, the access circuit FET sequentially writes the write data WD1-WD4 corresponding to the write instructions WC1-WC4 into the above addresses of the first memory SRAM1 and the second memory SRAM2 at the unit frequencies T0-T3. Similarly, the unit frequencies T4 and later can also be operated according to the same writing method, so they are not repeated here.

[0066] Therefore, the access circuit FET can write a set of data in each unit frequency without any pause and without setting a write buffer.

[0067] Figure 6 FIG. 6 is a flow chart of a memory device operating method 600 according to an embodiment of the present invention.

[0068] The memory device operation method 600 may be applied to Figure 1 In the memory device 110 shown, the memory device operation method 600 is used to perform a read operation of the access circuit FET on the first memory SRAM1 and the second memory SRAM2. The memory device operation method 600 includes the following steps (it should be understood that the steps mentioned in this embodiment, except for those whose order is specifically described, can be adjusted according to actual needs, and can even be executed simultaneously or partially simultaneously).

[0069] In step 601 , the access circuit FET receives a read command RC corresponding to a reference read address from the processing circuit 100 .

[0070] In step 602 , the access circuit FET converts the reference read address into the actual read address of the first memory SRAM1 and the second memory SRAM2 , wherein the multiple reference addresses of the processing circuit 100 correspond to the multiple actual addresses of the first memory SRAM1 and the second memory SRAM2 in an alternating manner.

[0071] In step 603 , the access circuit FET is enabled to simultaneously read the first set of read data RD1 and pre-read the second set of read data RD2 from the first memory SRAM1 and the second memory SRAM2 according to the actual read address and the next actual read address.

[0072] In step 604 , the access circuit FET is enabled to respond to the first set of read data RD1 to the processing circuit 100 .

[0073] In step 605 , the access circuit FET receives a next read command RC corresponding to a next reference read address and determines whether the next reference read address corresponds to a next actual read address.

[0074] In step 606 , when the next reference read address of the next read command RC corresponds to the next actual read address, the access circuit FET is enabled to respond to the pre-read read data, such as the second set of read data RD2 , to the processing circuit 100 .

[0075] In step 607 , the access circuit FET is enabled to pre-read the next two sets of read data and store them in the first read register BUF1 and the second read register BUF2 , respectively.

[0076] More specifically, the access circuit FET pre-reads the third set of read data RD3 from the first memory SRAM1 and the fourth set of read data RD4 from the second memory SRAM2 according to the next two actual read addresses and the next three actual read addresses, and stores them in the first read buffer BUF1 and the second read buffer BUF2 respectively.

[0077] After step 607, the process returns to step 605 to continue to judge the next received read instruction.

[0078] When the access circuit FET determines in step 605 that the next reference read address does not correspond to the next actual read address, in step 608 , the access circuit FET determines whether the processing circuit 100 has not issued a read instruction.

[0079] When the access circuit FET determines that the processing circuit 100 has not issued a read instruction, the process will continue to step 607 to continue pre-reading. When the access circuit FET determines that the processing circuit 100 has issued a read instruction, the process will return to step 602 to read the first memory SRAM1 and the second memory SRAM2 according to the new read address.

[0080] Figure 7 FIG. 7 is a flow chart of a memory device operating method 700 according to an embodiment of the present invention.

[0081] The memory device operation method 700 may be applied to Figure 1 In the memory device 110 shown, the memory device operation method 700 is used to perform the write operation of the access circuit FET to the first memory SRAM1 and the second memory SRAM2. The memory device operation method 700 includes the following steps (it should be understood that the steps mentioned in this embodiment, except for those whose order is specifically described, can be adjusted according to actual needs, and can even be executed simultaneously or partially simultaneously).

[0082] In step 701 , the access circuit FET receives a write command WC corresponding to a reference read address from the processing circuit 100 .

[0083] In step 702, the access circuit FET is made to convert the reference write address into the actual write address of the first memory SRAM1 and the second memory SRAM2.

[0084] In step 703, the access circuit FET is enabled to write the write data WD into the actual write address according to the write command WC.

[0085] Although specific embodiments of the present invention are disclosed in the above implementation modes, they are not intended to limit the present invention. Those with ordinary knowledge in the technical field to which the present invention belongs can make various changes and modifications without departing from the principles and spirit of the present invention. Therefore, the scope of protection of the present invention shall be based on that defined in the appended claims.

[0086]

Explanation of symbols

[0087] 1…Computer Systems

[0088] 110…Memory device

[0089] 600…Memory device operation method

[0090] 700…Memory device operation method

[0091] BUF1…First read buffer

[0092] CLK...frequency signal

[0093] CPUDA...Data

[0094] RC, RC1-RC6...Read instructions

[0095] SRAM1…First memory

[0096] T0-T7...Unit frequency

[0097] WD, WD1-WD4...write data 100...processing circuit

[0098] 120…Bus

[0099] 601-608…Steps

[0100] 701-703…Steps

[0101] BUF2…Second read buffer

[0102] CPUADD...refer to read address

[0103] FET…Access Circuit

[0104] RD, RD1-RD8... read data

[0105] SRAM2…Second memory

[0106] WC, WC1-WC4...write instructions.

Claims

1. A memory device, comprising: A first memory and a second memory, wherein a plurality of reference addresses of a processing circuit correspond to a plurality of actual addresses of the first memory and the second memory in an interlaced manner, comprising: The actual address of the Mth first memory corresponds to the reference address of the 2M-1th processing circuit, the actual address of the Mth second memory corresponds to the reference address of the 2Mth processing circuit, and M is a positive integer greater than or equal to 1; and An access circuit is configured to: receiving a read instruction corresponding to a reference read address from the processing circuit to convert the reference read address into an actual read address of the first memory and the second memory; Simultaneously reading a first set of read data from a first one of the first memory and the second memory and pre-reading a second set of read data from a second one according to the actual read address and an actual read address next to the actual read address; responding to the first set of read data to the processing circuit; as well as When the processing circuit receives a next read instruction corresponding to a next reference read address and when the next reference read address corresponds to the next actual read address, the access circuit pre-reads a third group of read data from the first of the first memory and the second memory and pre-reads a fourth group of read data from the second memory according to a second actual read address and a third actual read address at the same time; and when the next reference read address does not correspond to the next actual read address, the access circuit reads a fifth group of read data from the first of the first memory and the second memory and pre-reads a sixth group of read data according to a memory read address corresponding to the next reference read address conversion and a next memory read address adjacent to the memory read address.

2. The memory device according to claim 1, wherein: The read instruction is received from the processing circuit at at least one unit frequency, the first set of read data and the second set of read data are read from the first memory and the second memory at two of the unit frequencies, and the first set of read data and the second set of read data are read by the processing circuit at one of the unit frequencies. 3 . The memory device according to claim 1 , further comprising a first read register and a second read register, respectively configured to temporarily store the first set of read data and the second set of read data.

4. The memory device according to claim 1, wherein: The actual addresses of the first memory and the second memory correspond to data of a word group length.

5. The memory device according to claim 1, wherein: The access circuit is also configured to receive a write instruction corresponding to a reference write address from the processing circuit to convert the reference write address into an actual write address of the first memory and the second memory, and further write a set of write data into the actual write address according to the write instruction.

6. The memory device according to claim 1, wherein: The write command is received from the processing circuit at at least one unit frequency, and the write data is written at two of the unit frequencies.

7. The memory device according to claim 1, wherein: The memory device does not include a write buffer.

8. A method for operating a memory device, comprising: An access circuit receives a read instruction corresponding to a reference read address from a processing circuit to convert the reference read address into an actual read address of a first memory and a second memory, wherein a plurality of reference addresses of the processing circuit correspond to a plurality of actual addresses of the first memory and the second memory in an interlaced manner, comprising: The actual address of the Mth first memory corresponds to the reference address of the 2M-1th processing circuit, the actual address of the Mth second memory corresponds to the reference address of the 2Mth processing circuit, and M is a positive integer greater than or equal to 1; The access circuit simultaneously reads a first set of read data from a first one of the first memory and the second memory and pre-reads a second set of read data from a second one of the first memory and the second memory according to the actual read address and an actual read address next to the actual read address; causing the access circuit to respond to the first set of read data to the processing circuit; as well as When the processing circuit receives a next read instruction corresponding to a next reference read address and when the next reference read address corresponds to the next actual read address, the access circuit simultaneously pre-reads a third group of read data from the first of the first memory and the second memory and pre-reads a fourth group of read data from the second memory according to a second actual read address and a third actual read address; and when the next reference read address does not correspond to the next actual read address, the access circuit reads a fifth group of read data from the first of the first memory and the second memory and pre-reads a sixth group of read data according to a memory read address corresponding to the next reference read address conversion and a next memory read address adjacent to the memory read address.