Search Images Maps Play YouTube News Gmail Drive More »
Sign in
Screen reader users: click this link for accessible mode. Accessible mode has the same essential features but works better with your reader.

Patents

  1. Advanced Patent Search
Publication numberUSRE41589 E1
Publication typeGrant
Application numberUS 10/290,367
Publication dateAug 24, 2010
Filing dateNov 8, 2002
Priority dateSep 8, 1993
Fee statusPaid
Also published asUS5873122, US6154807, US6292867
Publication number10290367, 290367, US RE41589 E1, US RE41589E1, US-E1-RE41589, USRE41589 E1, USRE41589E1
InventorsOsamu Nishii, Nobuyuki Hayashi, Noriharu Hiratsuka, Tetsuhiko Okada, Hiroshi Takeda
Original AssigneeRenesas Technology Corp.
Export CitationBiBTeX, EndNote, RefMan
External Links: USPTO, USPTO Assignment, Espacenet
Memory system performing fast access to a memory location by omitting the transfer of a redundant address
US RE41589 E1
Abstract
A data processing system including a processor LSI and a DRAM divided into banks, for increasing a ratio of using a fast operation mode for omitting transfer of a row address to the DRAM and for minimizing the amount of logics external to the processor LSI. The processor LSI includes row address registers for holding recent row addresses corresponding to the banks. The contents of the row address registers are compared with an accessed address by a comparator to check for each bank whether the fast operation mode is possible. As long as the row address does not change in each bank, the fast operation mode can be used, thus making it possible to speed up operations, for example in block copy processing.
Images(9)
Previous page
Next page
Claims(47)
1. A data processing system, comprising:
a data processing unit;
a memory which is divided into a plurality of banks such that an access is made to one of said banks at a time;
a plurality of address registers coupled to said data processing unit, and corresponding to said plurality of banks;
a comparator coupled to at least one of outputs from said address registers; and
a controller coupled to said comparator, ;
a first line for sending a row address strobe signal, which is shared with the plurality of banks, from the controller to the memory;
a second line for sending a column address strobe signal, which is shared with the plurality of banks, from the controller to the memory;
a third line for sending a bank signal, which is shared with the plurality of banks, from the controller to the memory;
wherein said data processing unit accesses one of said banks, which is specified by the bank signal of the third line, for reading out data, and accesses another bank of said banks for writing data,
wherein each of said plurality of address registers holds recently accessed addresses corresponding to said plurality of banks,
wherein said comparator compares an access address for a bus access with contents of at least one of said address registers, when said data processing unit issues said bus access, and
wherein said controller omits transfer of said access address to said memory in response to an indication of a coincidence by said comparator, when said bus access is a next bus access after a bus access to a different bank, said indication indicating a coincidence between said access address and said contents of said at least one of said address systemregisters.
2. A data processing system according to claim 1, wherein said data processing unit, and said plurality of address registers are integrated in a single processor LSI (Large Scale Integration).
3. A data processing system according to claim 1, wherein said access address is a row address.
4. A data processing system according to claim 1, wherein said memory is a dynamic memory LSI.
5. A processor LSI comprising:
a plurality of address terminals for connecting with a main memory LSI, the main memory LSI using a synchronous dynamic memory having a plurality of banks such that an access is made to one of the plurality of banks at a time;
a first external terminal for outputting a row address strobe signal, which is shared with the plurality of banks, to the main memory LSI;
a second external terminal for outputting a column address strobe signal, which is shared with the plurality of banks, to the main memory LSI;
a third external terminal for outputting a bank signal, which is shared with the plurality of banks, to the main memory LSI;
a data processing unit;
a plurality of address registers coupled to said data processing unit and provided corresponding to the banks;
a selector coupled to said plurality of address registers;
a comparator coupled to said selector and said data processing unit; and
a controller coupled to said comparator,
wherein one of the plurality of the banks is specified by the bank signal, which is determined based on information included in a currently accessed address output by said data processing unit, and accessed at one time;
wherein said address registers hold recently accessed addresses,
wherein said selector selects one of said address registers based on the information included in the currently accessed address,
wherein said comparator compares a row address of the currently accessed address with a recently accessed row address held in the address register selected by said selector, and
wherein said controller, responsive to an output from said comparator indicating that the row address of the currently accessed address is the same as the recently accessed row address, omits transfer of said row address of the currently accessed address and allows output of a column address of the currently accessed address from said first terminals.
6. A processor LSI according to claim 5, wherein the information included in the currently accessed address is a bank bit for specifying one of the banks.
7. A processor LSI according to claim 5, wherein said processor LSI outputs the row address and the column address to the main memory LSI via said terminals by an address multiplex system.
8. A processor LSI according to claim 5, wherein said processor LSI arbitrates so that the data processing unit accesses one of the banks for reading out data and accesses the another one of the banks for writing data.
9. A processor LSI according to claim 5, wherein said main memory LSI is a synchronous dynamic memory.
10. The processor LSI according to claim 5, wherein when the recently accessed address held in one of said plurality of address registers which is selected by said selector is valid and the row address of the currently accessed address is the same as the recently accessed row address, said controller omits transfer of the row address of the currently accessed address and allows outputs of a column address of the currently accessed address from first terminals, and
wherein when the recently accessed address held in one of said plurality of address registers is invalid, said controller allows output of the row address and column address of the currently accessed address from said first terminals.
11. The processor LSI according to claim 10, further comprising:
a plurality of valid bits provided corresponding to said plurality of address registers, each of said plurality of valid bits having an information which indicates whether the recently accessed address held in each of said plurality of address registers is valid or invalid.
12. The processor LSI according to claim 5, wherein said address registers hold the row address of the recently accessed addresses.
13. A processor LSI comprising:
a plurality of address terminals for connecting with a main memory LSI, the main memory LSI having at least a first bank and a second bank such that an access is made to one of the first and second banks at a time,
a first external terminal for outputting a row address strobe signal, which is shared with the plurality of banks, to the main memory LSI;
a second external terminal for outputting a column address strobe signal which is shared with the plurality of banks, to the main memory LSI;
a third external terminal for outputting a bank signal, which is shared with the plurality of banks, to the main memory LSI;
wherein one of the plurality of the banks to be accessed is specified by the bank signal,
wherein said processor LSI is enabled to perform a first access to the first bank and then perform a second access to the first bank,
wherein when the value of a row address in the first access is different from the value of a row address in the second access, then said processor LSI outputs a row address and a column address from said address terminals, and
wherein when the value of a row address in said first access is identical with the value of a row address in second access, then said processor LSI outputs a column address from said first terminals without outputting a row address.
14. A processor LSI according to claim 13, wherein said processor LSI outputs the row address and the column address to the main memory LSI via said terminals by an address multiplex system.
15. A processor LSI according to claim 13, wherein the second access immediately follows the first access.
16. A processor LSI according to claim 13, wherein said processor LSI is enabled to perform an access to the second bank between the first access and the second access.
17. A processor LSI according to claim 13, wherein said processor LSI is enabled to perform a third access to the first bank, then perform a fourth access to the second bank, and thereafter perform a fifth access to the first bank,
wherein when the value of a row address in the third access is different from the value of a row address in the fifth access, then said processor LSI outputs a row address and a column address from said first terminals in the fifth access, and
wherein when the value of a row address in said third access is identical with the value of a row address in fifth access, then said processor LSI outputs a column address from said first terminals in the fifth access without outputting a row address.
18. A processor LSI according to claim 17, wherein said processor LSI outputs the row address and the column address to the main memory LSI via said second terminals by an address multiplex system.
19. A processor LSI according to claim 13, further comprising:
a data processing unit,
wherein said processor LSI is enabled to perform the first and second accesses according to requests by said processor unit.
20. A processor LSI according to claim 13, wherein said main memory LSI is a synchronous dynamic memory.
21. The processor LSI according to claim 13, further comprising:
a first address register corresponding to the first bank; and
a second address register corresponding to the second bank,
wherein said first address register holds the value of the row address in the first access,
wherein when the value of the row address held in said first address register is valid and identical with the value of the row address in second access, then said processor LSI outputs the column address from said first terminals without outputting the row address, and
wherein when the value of the row address held in said first address register is invalid, then said processor LSI outputs the row address and the column address from said first terminals.
22. The processor LSI according to claim 21, further comprising:
a first valid bit provided corresponding to said first address register; and
a second valid bit provided corresponding to said second address register,
wherein each of said first and second valid bits has an information indicating whether the value of an address held therein is valid or invalid.
23. A data processing system, comprising:
a data processing unit;
a memory having a plurality of banks;
a row address strobe signal line coupling said data processing unit and said memory and provided commonly to said plurality of banks;
a column address strobe signal line coupling said data processing unit and said memory and provided commonly to said plurality of banks;
a bank signal line coupling said data processing unit and said memory and provided commonly to said plurality of banks;
a plurality of address registers holding a row address of a recently accessed address; and
selection means for selecting one of said plurality of address registers using information of a specified bit among an access address issued by said data processing unit,
wherein said bank signal line sends a bank control signal for selecting one from said plurality of banks,
wherein when a row address of the access address issued by said data processing unit coincides with the content of the address register selected by said selection means, the row address in the access address issued by said data processing unit is allowed to prevent from being outputted to said memory.
24. A data processing system according to claim 23, wherein said specified bit information a bank bit designating one of said plurality of banks.
25. A data processing system according to claim 23, wherein said data processing unit, said plurality of address registers and said selection means are included internally of a processor LSI (Large Scale Integration).
26. A data processing system according to claim 23, wherein when the address register selected by said selection means does not hold a row address of a valid access address, said data processing system sends the row address and column address of said access address to said memory.
27. A data processing system according to claim 26, wherein said data processing system has a plurality of valid bits representing that said plurality of address registers respectively hold row addresses of valid access addresses, and judges whether or not said plurality of address registers hold row addresses of valid access addresses, respectively.
28. A data processing system according to claim 23, wherein the plurality of banks included in said memory include a first bank and a second bank, and when said data processing unit performs a first access to a first bank of said memory, performs a second access to a second bank of said memory and performs a third access to a third bank of said memory, a row address is not outputted at the third access if the row address for said first access coincides with that for said third access.
29. A data processing system according to claim 23, wherein said memory is a dynamic memory LSI, and said plurality of banks are included in one dynamic LSI.
30. A data processing system according to claim 23, wherein said memory includes a plurality of memories, said row address strobe signal line, said column address strobe signal line and said bank signal line are provided in common to said plurality of memories.
31. A processor, comprising:
an address terminal coupled to a synchronous memory having a plurality of banks including a first bank and a second bank;
a first external terminal for outputting a row address strobe signal commonly to said plurality of banks;
a second external terminal for outputting a column address strobe signal commonly to said plurality of banks; and
a third external terminal for outputting a bank signal commonly to said plurality of banks,
wherein said bank signal is a signal for selecting either one of said plurality of banks of said synchronous memory,
wherein when said processor performs a first access to a first bank of said synchronous memory, performs a second access to a second bank of said synchronous memory and performs a third access to said first bank of said synchronous memory, said processor does not output a row address at the third access if the row address for said first access coincides with that for said third access.
32. A processor according to claim 31, wherein when following said third access, a fourth access is performed to said second bank, said processor does not output a row address for the fourth access to said address terminal if the row address for said second access coincides with that for said fourth access.
33. A processor according to claim 31, wherein said synchronous memory is a dynamic memory.
34. A processor according to claim 31, wherein said processor further comprises address registers provided corresponding to said plurality of banks, the address register holding a row address used when the processor has accessed, and when the row address for said first access held in said address register is invalid, said processor outputs said access address to said synchronous memory without omitting.
35. A processor according to claim 34, wherein said processor further comprises valid bits provided corresponding to said plurality of address registers, the valid bit representing whether a row address held in a relevant one of said address registers is valid or not, and in said third access, said processor judges whether the row address for said first access is valid or not by examining a corresponding one of said valid bits.
36. A processor according to claim 31, wherein said synchronous memory is one LSI, and said plurality of banks are included in said one synchronous memory.
37. A processor according to claim 31, wherein said processor outputs said row address strobe signal, said column address strobe signal and said bank signal in common to a plurality of said synchronous memories.
38. An external circuit control LSI comprising:
a plurality of external terminals coupled to a processor LSI; and
a plurality of second external terminals coupled to a main memory LSI having a plurality of banks including a first bank and a second bank,
wherein said plurality of second external terminals comprises:
an address terminal for outputting an access address;
a terminal for outputting a row address strobe signal commonly to said plurality of banks;
a terminal for outputting a column address strobe signal commonly to said plurality of banks; and
a terminal for outputting a bank signal commonly to said plurality of banks,
wherein said bank signal is a signal for selecting one to be accessed from said plurality of banks, and said external circuit control LSI is enabled to perform a first access to said first bank and thereafter perform a second access to said first bank,
wherein when in said second access, a row address in said second access is different from a row address in said first access, said external circuit control LSI outputs a row address and a column address from said second external terminals,
wherein when in said second access, the row address in said second access is the same as the row address in said first access, said external circuit control LSI outputs the column address from said second external terminals without outputting the row address therefrom.
39. An external circuit control LSI according to claim 38, wherein said second access follows immediately after said first access.
40. An external circuit control LSI according to claim 38, wherein said external circuit control LSI is enabled to perform a third access to said second bank between said first access and said second access.
41. An external circuit control LSI according to claim 38, wherein said external circuit control LSI is enabled to perform a third access to said first bank, and thereafter perform a fourth access to said second bank, and then perform a fifth access to said first bank,
wherein when a row address in said fifth access is different from a row address in said third access, said external circuit control LSI, in said fifth access, outputs a row address and a column address from said second external terminals, and
wherein when the row address in said fifth access is the same as the row address in said third access, said external circuit control LSI, in said fifth access, outputs the column address from said second external terminals without outputting the row address therefrom.
42. An external circuit control LSI according to claim 38, wherein said external circuit control LSI receives a row address and a column address which constitute a current access address through said first external terminals in parallel, and splits the received access address into the row address and the column address and outputs the same to said main memory LSI from said first external terminals.
43. An external circuit control LSI according to claim 38, wherein said main memory LSI is a synchronous dynamic memory.
44. An external circuit control LSI according to claim 38, wherein said external circuit control LSI further comprises address registers provided corresponding to said plurality of banks, the address register holding a row address used when accessed to any of said plurality of banks, and when the row address for said first access held in said address register is invalid, said external circuit control LSI outputs a row address and a column address to said main memory LSI.
45. An external circuit control LSI according to claim 44, wherein said external circuit control LSI further comprises a valid bit representing whether a row address of said first access is valid or not, and judges whether the row address for said first access is valid or not by examining said valid bit.
46. An external circuit control LSI according to claim 38, wherein said plurality of banks are included in one main memory LSI.
47. An external circuit control LSI according to claim 38, wherein said external circuit control LSI outputs said row address strobe signal, said column address strobe signal and said bank signal to a plurality of said main memory LSI's.
Description

This is a continuation of application Ser. No. 08/815,600, filed Mar. 12, 1997 now U.S. Pat. No. 5,873,122; which is a continuation of Ser. No. 08/301,887, filed Sep. 7, 1994, now abandoned.

BACKGROUND OF THE INVENTION

The present invention generally relates to a data processing system, and more particularly to a circuit for controlling memories in a microprocessor LSI (Large Scale Integration) and microprocessor peripheral circuits.

Dynamic memories are generally referred to as DRAM (Dynamic Random Access Memory). Specifications of typical dynamic memory LSI's are described in, for example, “Hitachi IC Memory Data Book 3 (DRAM, DRAM Modules)”, pp. 445-464. A dynamic memory described in this document has ten address input terminals indicated by A0-A9 which are shared to receive row and column addresses (see page 448). Also according to this literature, a read/write access requires a row address and a column address to be provided to the dynamic memory LSI in this order (see page 454), wherein read access time is 70 nanoseconds after the establishment of the externally provided address (1 nanosecond=1Χ10−9 second). Alternative to this read/write access, if a fast page mode (page 461) is used, after the first row and column addresses have been transferred, as long as second and subsequent accesses are made to the same row, transfer of the row address can be omitted, with the result that read access time required for the second and subsequent read accesses is reduced to 20 nanoseconds from the establishment of the external address.

An example of a DRAM control function designed for a conventional microprocessor (hereinafter simply called the “processor”) is described in “Hot Chips IV”, pp. 4.2.2-4.2.12, August. 1992, held in Stanford University. On page 4.2.3 of this document, a drawing is illustrated in which a processor LSI is directly connected to two banks of DRAM chips. Also, timing charts on pages 4.2.7 and 4.2.8 of this document respectively include descriptions “Check fast page cache-hit” and “Check fast page cache-miss”, from which it can be predicted that the fast page mode of the dynamic memory is used under certain hit conditions within the processor. This operation would be enabled, for example, by storing a row address with which a dynamic memory has been accessed at the previous time. The above-mentioned document, however, does not at all refer to how to use two-bank DRAM's or the relation between the cache-hit of the high speed mode and the two-bank DRAM's.

Assume now a conventional processor LSI which includes, among its terminals, dynamic memory address terminals which are used for both row and column addresses. FIG. 2 shows an example of accesses performed by this processor. It should be noted that in FIG. 2, the horizontal direction represents the time axis, and reference numeral 201 designates an access request from the processor; 202 dynamic address terminals A0-A9 of the processor; 203 a row address strobe (RAS−n) signal of a dynamic memory; and 204 a column address strobe (CAS−n) signal of the dynamic memory. A suffix “−n” to a signal line indicates that a signal on that line is of negative polarity.

This exemplary access occurs, for example, when a block of data, i.e., the contents of a memory in a certain region is copied to another region of the memory. In FIG. 2, a region from address A000 is copied to a region from address 7040. It should be noted that in this specification memory addresses are indicated in hexadecimal number. An explanatory diagram 205 shows how to use 32 bits of a physical address. Specifically, bits (30-31) of the physical address are assigned to an in-word address; bits (21-29) to a column address of a dynamic memory; and bits (11-20) to a row address of the dynamic memory. Here, bit (i) indicates the position of the i-th bit from the leftmost bit which is designated as bit 0 position. The copy is carried out by the following time-sequential operations.

Operation 1: The contents at address A000 are read. A row address and column address are transferred to the dynamic memory. The row address given by the bit positions (11-20) of the physical address is “14”, then the column address given by the bit positions (21-29) is zero.

Operation 2: The contents of address A004 are read. Since the row address at this time is the same as that at the previous time, transfer of the row address to the dynamic memory is omitted. Thus, the column address only is transferred to the dynamic memory.

Operation 3: The contents of address A000 are written into address 7040. A row address and column address for this location are transferred to the dynamic memory.

Operation 4: The contents of address A004 are written into address 7044. Since the row address at this time is the same as that at the previous time, the transfer of the row address to the dynamic memory is omitted. Thus, the column address only is transferred to the dynamic memory.

Since the subsequent four accesses perform similar operations to the above, explanation thereon will be omitted.

As shown by the example of FIG. 2, the prior art example has a problem that if a row address of a memory location (source) from which data is read and a row address of a memory location (destination) to which the data is written are different in the block copy processing, the fast mode for omitting the transfer of a row address of the dynamic memory is prohibited each time the access source and destination are switched.

SUMMARY OF THE INVENTION

It is an object of the present invention to solve the problem of the fast mode for omitting the transfer of a row address of a dynamic memory, which is prohibited depending on conditions.

It is another object of the present invention to provide a signal line interface for a processor LSI which allows the fast mode for omitting the transfer of a row address of a dynamic memory to be used in a processor which does not have address terminals for both row and column addresses, and which thereby simultaneously minimizes the amount of logic mounted external to the processor LSI.

It is another object of the present invention to provide a signal line interface for a processor LSI which allows information on a fast page operation mode corresponding to plural banks of synchronous dynamic memories to be set from the processor LSI, and which minimizes the amount of logic mounted external to the processor LSI.

According to one feature of the present invention, there is provided a data processing system comprising: a data processing unit; a memory; a plurality of address registers for holding recently accessed addresses; selector means for selecting one of the plurality of address registers by using particular bit information in a currently accessed address; comparator means for comparing, when the data processing unit issues a bus access to the outside, an access address for the bus access with the contents of the address register selected by the selector means in accordance with the particular bit information; and control means for performing an operation for omitting transfer of the access address to the memory when the result of a comparison made by the comparator means shows coincidence.

A concept of the above-mentioned feature will be explained below with reference to FIG. 1, an internal configuration of a processor which employs the present invention (the configuration in FIG. 1 will be further explained in detail hereinunder; In the present invention, a processor LSI 100 includes a plurality of row address registers (storage units) 101, 102. One or a plurality of particular bits are specified within a plurality of address bits. The specified bits are hereinafter referred to as “DRAM bank bits”. The plurality of row address registers 101, 102 hold row addresses of the respective banks which have been accessed at the previous time.

The dynamic memory is divided into a plurality of banks such that one bank in the dynamic memory specified by the bank bit is accessed at one time.

When the processor LSI issues a bus access to the outside, a coincidence comparator 107 compares an output value 101 of the row address register 101 or 102 selected by the bank bit of an access address with a row address portion of the access address. If the result 113 of comparison is true, the processor LSI performs an operation for omitting transfer of the row address to the dynamic memory LSI.

FIG. 3 shows an access pattern of an information processing system employing the present invention. Components 301, 114, 117, 116, 305 in FIG. 3 correspond to components 201-205 in FIG. 2, respectively, so that explanation thereon will be omitted. As shown in 205, a 32-bit physical address of this example consists of the less significant two bits of bit positions 30-31 assigned to an in-word address; bits of bit positions 21-29 to a column address; bit (20) to a bank address of a dynamic memory; and bits of bit positions 9-19 to a row address of the dynamic memory. In one access, a dynamic memory LSI specified by the bank bit of bit position 20 only is accessed.

In one access, one bank in a dynamic memory specified by the bank bit only is accessed. During a period for accesses to addresses A000, A004, a dynamic memory LSI corresponding to bank 0 is accessed. On the other hand, during a period for access to addresses 7040, 7044, a dynamic memory LSI corresponding to bank 1 is accessed, but the dynamic memory LSI corresponding to bank 0 is not accessed. Subsequently, when address A008 is to be read, the row address at this time is the same as the row address when the dynamic memory LSI corresponding to bank 0 was accessed at the previous time, so that transfer of the row address may be omitted. During this period, the dynamic memory LSI corresponding to bank 1 is not accessed. Subsequently, when data is written into address 7048, the row address is the same as that when the dynamic memory LSI corresponding to the bank 1 was accessed at the previous time, so that transfer of the row address may be omitted.

Comparing FIG. 3 with FIG. 2, it is understood that transfer of the row address may be omitted when the addresses A008, 7048 are accessed. Since access time required in this event is shorter as mentioned above, the omission of the transfer of the row address contributes to a higher processing speed.

When the foregoing block copy processing further continues, the fast page mode may be used to omit the transfer of the row address as long as the same rows are continuously accessed in both source and destination.

Other objects, configurations and effects of the present invention will become apparent from the following description of preferred embodiments taken in conjunction with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram showing the configuration of a processor LSI in an information processing system which employs the present invention;

FIG. 2 shows an access pattern (changes in operation with the lapse of time) in a prior art information processing system;

FIG. 3 shows an access pattern (changes in operation with the lapse of time) in the information processing system employing the present invention;

FIG. 4 is a block diagram showing the configuration of a system which includes a synchronous dynamic memory divided into two banks;

FIG. 5 is a block diagram showing dynamic memory banks and its control circuit in the information processing system employing the present invention;

FIG. 6 is a block diagram showing the configuration of a processor LSI in another information processing system employing the present invention;

FIG. 7 shows a bank selector circuit and a bank bit selector circuit associated with the processor shown in FIG. 6; and

FIG. 8 is a block diagram showing the configuration of an information processing system employing the processor illustrated in FIG. 6.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

FIG. 1 shows an example of a processor, generally indicated by 100, for use in an information processing system according to an embodiment of the present invention. The processor 100 includes row address registers (register=storage unit) 101, 102; bits 103, 104 which indicate the validity of the row address registers 101, 102, respectively; and selector circuits 105, 106. The selector 105 selects one of outputs of the valid bits 103, 104 in accordance with a bank address, and delivers the selected one onto a signal line 112. Reference numeral 107 designates a coincidence comparator for comparing a row address in a access requested address 110 with a row address stored in one of the row address registers 101, 102 selected by the selector circuit 106; 108 a control circuit; and 109 a selector circuit.

The processor 100, when accessing to an external dynamic memory, selects either a row address portion or a column address portion in the access requested address 110 by using the selector circuit 109 and outputs the selected address portion to an address terminal 114 having bits A(0:10) for dynamic memories. The control circuit 108 is applied with the output of a valid bit selected by the selector circuit 105 and with an output signal 113 of the coincidence comparator 107 indicative of the result of comparison between a row address in the access requested address 110 and a row address stored in one of the row address registers 101, 102 selected by the selector circuit 106. The control circuit 108 is also applied with a part of an address bus 110 corresponding to the bank bit. Then, the control circuit 108 outputs signal values of three external terminals BANK (115), RAS−n (117) (“−n” indicates a negative polarity signal), and CAS−n (118).

A 32-bit physical address, as illustrated in FIG. 3, includes fields respectively assigned to a row address, a column address, and a bank address.

A flow of processing executed by the processor 100 when generating an access request proceeds as follows.

First, an access request signal generated from a data processing unit (not described or shown since this is not so deeply related to the present invention) composed of a command processing unit and an operand processing unit, arranged inside the processor 100, is transferred to the control circuit 108 through a signal line 111. Simultaneously, an access requested address 110 is transferred through the address bus PA (0-31). The control circuit 108 selects one of the row address registers 101, 102 by the selector circuit 106 in accordance with a bank address (bit 20) in the access requested address. The control circuit 108 also selects one of the valid bits 103, 104 by the selector circuit 105 in accordance with the bank address. When the row address selected by the selector circuit 106 is equal to a row address portion of the access requested address to cause the associated valid bit to be set to “1”, this is referred to as “hit”.

If a hit occurs, the dynamic memory will be accessed with the same row address as that in the previous access, so that the access at the present time will be performed in an operation mode for omitting transfer of the row address to the dynamic memory.

If a miss occurs, the dynamic memory is first accessed in an operation mode for transferring both the row and column address portions of an access requested address of the dynamic memory. Next, the row address portion is registered in one of the row address registers 101, 102 selected by the bank address, and the valid bit 103 or 104 associated with the selected row address register is written to “1”. It should be noted that the valid bits 103, 104 are set to “0” in an initial state immediately after power-on in order to prevent an erroneous hit from occurring when the row addresses accidentally match with each other in the first access after power-on.

FIG. 4 shows a block diagram of a system including the processor 100. The system also has synchronous dynamic memory LSI's 401, 402. The control signals A(0:10) (114), Bank (115), RAS−n (117), CAS−n (116) are connected to the memory LSI's 401, 402. The synchronous dynamic memory LSI 401 is divided into two banks 403, 404 such that the memory bank 404 is accessed when the control signal BANK (115) is at “0”, and the memory bank 403 is accessed when BANK (115) is at “1”. As is well known, logic “0” on a signal line means “LOW” potential, and logic “1” means “HIGH” potential.

When the signal BANK (115) is at “0”, logic “1” is generated at the output of an invertor 407 and is transferred to a bank 0 memory control circuit 406. This is an indication of an access to the bank 0 memory. Conversely, when the signal BANK (115) is at “1”, logic “0” is transferred to the bank 0 memory control circuit 405. This is an indication of an access to a bank 1 memory.

In addition, each of the dynamic memory LSI's 401, 402 has the following terminals: I/O0-7 (409, 410) representing an 8-bit data input/output signal; WE−n (411, 412) representing a write command signal which remains at logic “0” during a write operation; CLK (413, 414) representing a clock input terminal; CKE (415, 416) representing a clock enable signal for controlling whether or not a clock is transferred to the inside; and DQM (408) representing an access mask signal which functions as an output enable signal in a read access for outputting the data input/output signal I/O0-7, and as a write enable signal in a write access for enabling the data input/output signal I/O0-7 to be written each time the clock is applied thereto.

The dynamic memory LSI's 401, 402 have several forms of operation mode information for their synchronous operation which are RAS delay (the number of clock cycles from RAS to a data access), CAS delay (the number of clock cycles from CAS to a data access), and burst length (a period in which the address is counted up from initial value to an end value and returned to the initial value). These mode information signals are written through the address bits A0-10 when all of RAS−n (116), CAS−n (116), and WE−n (411) are at potential “L”.

FIG. 5 shows dynamic memories and its control circuit in another system including the processor 100. Blocks 501, 503 each include a plurality of invertors. Another block 502 includes a plurality of two-input AND circuits, each of which generates an output value “1” only when both input values are “1”. A block 504 includes dynamic memories of bank 0, and a block 505 includes dynamic memories of bank 1. Lines 506, 507 respectively carry a negative-polarity row address strobe signal and column address strobe signal for the dynamic memories 504 of bank 0. Lines 508, 509 respectively carry a negative-polarity row address strobe signal and column address strobe signal for the dynamic memories 505 of bank 1.

When the signal BANK (115) is at “0”, negative pulses appearing on RAS−n (117), CAS−n (116) (see 303, 304 in FIG. 3) are transferred to the signal lines 506, 507, respectively, but not to the signal lines 508, 509. As a result, a dynamic memory 504 in bank 0 is accessed. Conversely, when the signal BANK (115) is at “1”, the negative pulses appearing on RAS−n (117), CAS−n (116) are transferred to the signal lines 508, 509, respectively, but not to the signal lines 506, 507. As a result, a dynamic memory 505 in bank 1 is accessed.

FIG. 6 shows an example of another processor, generally indicated by 600, which employs the present invention. Since components 601-608 and signal lines 610, 611-613 are arranged similarly to their correspondents 101-108, 110, 111-113 in FIG. 1, explanation thereon will be omitted. The processor 600 does not have a shared address terminal through which row and column addresses are specified, but a 32-bit address terminal A(0-31) (614) including a separate row address and column address.

A two-input selector 615 selects one of an access requested address 610 and an address stored in a register 617, and delivers the selected one to the address terminal A(0:31) (614). An output buffer 616 for the processor 600 is arranged at the output of the selector 615, and delivers an output with a logic value identical to that of an input.

When the two-input selector 615 selects the access requested address 610, the resulting operation is similar to that explained in connection with FIG. 1. This operation to be executed when the access requested address 610 is selected will be explained below.

When the processor 600 generates an access request, the following processing flow is executed. First, an access request signal from a processing unit including a command processing unit and an operand processing unit arranged within the processor 600 is transferred to a control circuit 608 through a signal line 611. Simultaneously, an access address is transferred through an address bus PA(0:31) (610). The control circuit 608 selects one of outputs from row address registers 601, 602 in accordance with a bank address in the access requested address. Also, a selector 605 selects one of valid bits 603, 604 in accordance with the bank address.

If a hit occurs, the control circuit 608 sets an output terminal SAR (609) to “1”. The terminal SAR delivers an output signal which indicates an access to the same row region. The definition of “hit” is the same as that made in connection with the explanation of FIG. 1.

If a miss occurs, the control circuit 608 sets SAR to “0”. Also, information is registered in the registers 601, 602 and valid bits 603, 604 as is done in the example of the processor 100.

A circuit external to the processor LSI 600 detects that SAR (609) is at “1” to know that a fast operation mode may be used for omitting transfer of a row address of an associated dynamic memory.

The processor LSI 600, unlike the processor LSI 100, does not have bank bits fixed at predetermined positions. FIG. 7 shows part of the configuration related to a bank bit selecting method. Since components 601, 602, 606, 608, 612 in FIG. 7 have already been explained, repetitive explanation thereon will be omitted. A two-input selector 606 in FIG. 7 is controlled by a bank bit control signal 703. A 21-input selector 702 selects one from input signals 704, 705, . . . , 706, 707 in accordance with a control signal 701 from the control circuit 608, and delivers the selected signal onto a signal line 703. Input signals 704, 705, 706, 707 are individual address signals included in a requested address on the address bus PA(0:31) (610). The processor 600 can arbitrarily set the control signal 701 by using a particular command. In summary, an arbitrary bit within bit positions 0-20 in a requested address can be used as a bank address.

Next, explanation will be given of how the two-input selector 615 in FIG. 6 selects the register 617. Set in the register 617 is information on an operation mode (RAS delay, CAS delay, burst length) of a synchronous dynamic memory. By the processor 600 executing the particular command, the two-input selector 615 selects the register 617 and outputs information registered therein onto the address terminal A(0:31) (614). An operation mode setting operation is achieved for the synchronous dynamic memory connected external to the processor 600 by a combination of appropriate external circuits.

FIG. 8 shows the configuration of an exemplary information processing system which employs the processor LSI 600. The configuration in FIG. 8 is mainly composed of the processor LSI 600, an external circuit control LSI 801, and main memory LSI's 402 using synchronous dynamic memories.

Explanation will first given of how an address signal 614 is transferred from the processor LSI 600. The address signal 614 is applied to the external circuit control LSI 801 and stored in an address register 802 provided therein. Signal lines 805, 806 from the address register 802 carry a row address and a column address to the main memory LSI's 402, respectively. Either of the row address 805 and the column address 806 is selected by a two-input selector 803 and sent to a system address bus 811. The address signal on the system address bus 811 is further transferred to an address terminal of the respective main memory LSI's 402.

Within an address registered in the address register 802, upper address bits 807 are decoded by an address decoder 814, and the decoded result is transferred to a chip select terminal 813 of the respective main memory LSI's 402.

A 32-bit system data bus 812 is used to communicate data between the processor LSI 600 and the main memory LSI's 402. It should be noted that since the memory LSI 402 has a data terminal of eight bit width, the system of this embodiment should include at least four memory LSI's 402 for communicating 32-bit data.

The external circuit control LSI 801 includes an access request managing logic 804 for managing conditions related to the access. From the processor LSI 600, an access request signal 808, an identical address indicating signal 609, and operation mode setting request signal 815 for dynamic memories are transferred to the access request managing logic 804.

When the signal 815 is at logic “0”, the access request managing logic 804 operates as follows. First, when an access request exists in the signal 808 and the identical address indicating signal is at logic “0”, RAS−n (809) is issued to the main memory LSI's 402, and simultaneously a row address 805 is delivered onto the system address bus 811. Subsequently, CAS−n (810) is issued to the main memory LSI's 402, and simultaneously a column address 805 is delivered onto the system address bus 811.

Second, when an access request exists in the signal 808 and the identical address indicating signal is at logic “1”, issue of the RAS−n (809) and row address 805 is omitted, unlike the first case.

When the processor 600 executes the aforementioned particular command (the command mentioned when the register 617 was explained), the operation mode setting request signal 815 changes to logic “1”. When the signal 815 is at logic “0”, the access request managing logic 804 sets all of RAS−n (809), CAS−n (810), and WE−n (816) to potential “L”. Simultaneously with this, the value stored in the operation mode register 617 is transferred to the main memory LSI's 402 through the system address bus 811. In this manner, the operation mode setting processing is carried out for the main memory LSI's 402. This processing is performed during an initial operation after the power is turned on, or when the system is reset. Since the processor LSI 600 provides the operation mode setting request signal 815, logics so far required for producing a signal used to generate an operation mode setting processing starting signal for the main memory LSI's 402, for example, an address decoding logic, are made unnecessary.

The system of the present invention is implemented in HITACHI HM 5241605 series, 131072-wordΧ16-bitΧ2-bank Synchronous Dynamic RAM which is incorporated herein by reference.

It will be understood that the present invention is not limited to the above described specific embodiments, but may be modified in various manner within the scope of its technical ideas. For example, while the number of row address registers and the number of banks in a dynamic memory are set to two, these numbers may be increased to four, eight, and so forth. Also, the row address registers and coincidence comparator are not necessarily arranged in the processor LSI, and in alternative, processing similar to that of the above embodiments may be performed external to the processor LSI, for example, in the external circuit control LSI 801.

According to the embodiments of the present invention, row addresses of source and destination are held in row address registers for processing such as a block copy, so that a fast operation mode for dynamic memories can be used to omit transfer of the row address.

Also, since the row address hit information 690 is provided as an output signal of the processor LSI 600, the embodiments of the present invention allow processors, which do not have a shared address terminal for both row and column addresses of a dynamic memory, to use the fast operation mode for dynamic memories, whereby the amount of logics external to the processor LSI is minimized, and transfer of a row address to the dynamic memories is omitted.

Further, according to the embodiments of the present invention, since the processor LSI 600 provides the operation mode setting request signal 815, logics so far required for producing a signal used to generate an operation mode setting processing starting signal for the main memory LSI's 402, for example, an address decoding logic and so on, are made unnecessary.

Patent Citations
Cited PatentFiling datePublication dateApplicantTitle
US4792891Nov 20, 1985Dec 20, 1988Hitachi, Ltd.Data processor
US4825358 *Jun 18, 1987Apr 25, 1989Microsoft CorporationMethod and operating system for executing programs in a multi-mode microprocessor
US5187792 *May 9, 1990Feb 16, 1993International Business Machines CorporationMethod and apparatus for selectively reclaiming a portion of RAM in a personal computer system
US5193161 *May 18, 1990Mar 9, 1993International Business Machines Corp.Computer system having mode independent addressing
US5329629 *Jul 3, 1989Jul 12, 1994Tandem Computers IncorporatedApparatus and method for reading, writing, and refreshing memory with direct virtual or physical access
US5430686 *Oct 15, 1992Jul 4, 1995Mitsubishi Denki Kabushiki KaishaSemiconductor memory device and operating method thereof
US5461718 *Apr 24, 1992Oct 24, 1995Digital Equipment CorporationSystem for sequential read of memory stream buffer detecting page mode cycles availability fetching data into a selected FIFO, and sending data without aceessing memory
US5491808 *Sep 30, 1992Feb 13, 1996Conner Peripherals, Inc.Method for tracking memory allocation in network file server
US5652860 *Mar 31, 1995Jul 29, 1997Mitsubishi Denki Kabushiki KaishaMemory control device
US5666494 *Mar 31, 1995Sep 9, 1997Samsung Electronics Co., Ltd.Queue management mechanism which allows entries to be processed in any order
US5873122 *Mar 12, 1997Feb 16, 1999Hitachi, Ltd.Memory system performing fast access to a memory location by omitting transfer of a redundant address
JPH0465739A Title not available
JPH03232186A Title not available
JPH04299752A Title not available
JPH05165713A Title not available
JPS4866745A Title not available
JPS61127056A Title not available
JPS61231640A Title not available
Non-Patent Citations
Reference
1 *"A 50-ns 16 Mb DRAm with a 10-ns Data Rate and On-chip ECC" by H. Kalter et al, IEEE Journal of Solid-State Circuits vol. 25, No. 5 pp. 1113-1128, Oct. 1990.
2 *"Hot Chips IV", by G. Langdon, Stanford University. pp. 4.2.2-4.2.12, Aug. 1992.
3 *"Implementation of a sub-10 ns series access mode to a standard" by Y. Watanabe et al, Custom Integrated Circuits, 1993 Conference IEEE, 1991.
4 *"Performance analysis for a cache system with different DRAM" by N. Mekhiel et al, Electrical and Computer Engineering, 1993 Conference IEEE, 1993.
5 *Hitachi IC Memory Data Book 3, Sep. 1992.
6 *HM 5241605 131072-Word X 16-bit X2-bank Synchronous Dynamic RAM, Hitachi Semiconductor. pp. 1-6, 36-39, Feb. 20, 1994.
Classifications
U.S. Classification711/5, 711/105
International ClassificationG06F12/02, G06F12/00, G06F12/06
Cooperative ClassificationG06F13/1668, G06F12/0215
European ClassificationG06F12/02C
Legal Events
DateCodeEventDescription
May 2, 2012FPAYFee payment
Year of fee payment: 12
May 17, 2011ASAssignment
Free format text: MERGER AND CHANGE OF NAME;ASSIGNOR:RENESAS TECHNOLOGY CORP.;REEL/FRAME:026287/0075
Owner name: RENESAS ELECTRONICS CORPORATION, JAPAN
Effective date: 20100401
Mar 1, 2011CCCertificate of correction
Sep 26, 2003ASAssignment
Owner name: RENESAS TECHNOLOGY CORPORATION, JAPAN
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:HITACHI, LTD.;REEL/FRAME:014569/0585
Effective date: 20030912