JPH0355862B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a memory access control device in an information processing device, and more particularly to a memory access control device that controls continuous access to data consisting of one or more elements.

[Conventional technology]

Conventionally, in December 1982, the present inventor developed a memory access control device that performs continuous access control of data consisting of multiple elements such as vector data.
The patent application No. 58-243323, filed on the 23rd, proposes the name of the invention ``Memory Access Control Device.'' The proposed memory access control device obtains and holds the address information of the memory unit of the last element of the data to be accessed in advance from the address information of the memory unit of the first element, for example, the bank address and the number of elements. Compare the difference between the memory unit address information of the first element of the subsequent data accessed by the memory unit and the memory unit address information of the last element of the preceding data held, and the width of the memory unit that is in use, Even if the comparison unit starts accessing the subsequent data, a signal is sent indicating that the subsequent data will not be accessed to the memory unit that is in use due to the access of the preceding data. It is controlled so that it is possible to start sending out access elements.

[Problem that the invention seeks to solve]

However, the conventional memory access control device as described above is a device that has a plurality of arithmetic units and transfers data from memory to all the arithmetic units at the same time. In an information processing device where the number of elements that can access memory at the same time is variable, such as one or more arithmetic units, including one or more arithmetic units, that operate in a degenerate manner, the number of elements that can access memory simultaneously is Depending on the number, the relationship between the difference in address information of memory units and the width of the memory unit in use changes, making it impossible to compare them correctly. Therefore, the signal indicating that it is possible to start accessing subsequent data may be erroneously enabled, that is, logic level "1".
This method has the disadvantage that the access start enable signal must be disabled, which reduces the efficiency of hardware use. Or, conversely, even though it is actually possible to start accessing subsequent data, the access start enable signal may remain invalid, that is, remain at logic level "0", resulting in degraded performance. There is a drawback.

[Means for solving problems]

The memory access control device of the present invention is composed of a plurality of memory units that can be accessed independently of each other, and each element is arranged consecutively on the memory and has one or more memory units arranged in memory unit order. A memory access control device for controlling access to data consisting of elements, comprising: element number supply means for supplying the number of data elements; address information supply means for supplying address information of a memory unit of a leading element; address information calculation means for calculating memory unit address information of a final element from the number of elements and memory unit address information of the first element; address information holding means for retaining address information of the final element; and memory of the final element. a time information calculation means for calculating time information until the unit is no longer in use; a time information holding means for holding the time information; and a memory for the final element of the first data held in the address information holding means. a difference calculation means for calculating a difference between address information of a unit and address information of a memory unit of a first element of second data that is accessed subsequent to the first data; and a number of elements that can be accessed simultaneously. a correction means for correcting the difference in address information calculated by the difference calculation means according to a mode signal; and a correction means for correcting the difference in address information corrected by the correction means and the time information held in the time information holding means. It is characterized in that it includes a comparison means for making a comparison, and a means for making an access request for the second data based on the result of the comparison means.

〔Example〕

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an embodiment of a memory access control device according to the present invention. In the figure, the memory access control device of this embodiment is composed of an addition circuit 1, a register 2, a subtraction circuit 3, a gate circuit 4, a shift circuit 5, a register 6, a subtraction circuit 7, a subtraction circuit 8, and an inversion circuit 9. There is.

The adder circuit 1 receives the bank address of the leading element of the data supplied via the connection 101 and the connection 102.
and the number of elements of data supplied via
The value "1" is subtracted from this addition result to calculate the bank address of the final element of data, which is supplied to the register 2 via the connection 103.

Register 2 holds the bank address of the final element of data, which is the output of the adder circuit 1, supplied via connection 103;
4 to the subtraction circuit 3.

The subtraction circuit 3 extracts the data from the bank address of the final element of the data supplied via the connection 104.
1, and calculate their difference by subtracting the bank address of the first element of the data supplied through connection 10.
5 to the gate circuit 4. here,
The bank address of the last element and the bank address of the first element of the data supplied to the subtraction circuit 3 are not the bank addresses of the last element and the first element of the same data, but the bank address of the last element is the bank address of the data accessed in advance. Note that the bank address of the last element of is the bank address of the last element, and the bank address of the first element is the bank address of the first element of the data that is accessed subsequent to the preceding data including the last element.

The gate circuit 4 calculates the difference between the bank address of the last element of the preceding data supplied via the connection 105 and the bank address of the first element of the subsequent data, based on the memory bank number information supplied via the connection 106. The signal is masked and supplied to the shift circuit 5 via the connection 107. In this embodiment, there are three modes for the number of memory banks: 256 banks, 128 banks, and 64 banks. In the case of 256 banks, all bits given by connection 105 are directly transferred to connection 1.
07, in the case of 128 banks, the most significant bit given by connection 105 is set to "0" and connection 107
In addition, in the case of 64 banks, the gate circuit 4 is controlled so that the upper two bits given by the connection 105 are set to "0" and output to the connection 107.

The shift circuit 5 transfers the masked bank address difference supplied via the connection 107 to the connection 10.
The signal is shifted to the right according to information on the number of arithmetic pipelines supplied via line 109, and is supplied to subtraction circuit 8 via connection 109. In this example, there are three modes for the number of calculation pipelines: 4, 2, and 1. In the case of 4 pipelines, 4 elements can be accessed simultaneously, and in the case of 2 pipelines, 2 elements can be accessed simultaneously. The shift circuit 5 performs control such that if there are two, it shifts to the right by two bits, if there are two, it shifts by one bit to the right, and if there is one, it does not shift.

Register 6 has memory bank cycle time information provided via connection 110 and connection 111.
The output of the subtraction circuit 7, which is supplied via a connection 112, is selectively supplied by an access signal supplied via a connection 112, and the output is supplied via a connection 113 to the subtraction circuits 7 and 8. This register 6
Here, when the logical value of the access signal is "0", the output of the subtraction circuit 7 is selected, and when the logical value of the access signal is "1", the memory bank cycle time information is selected. Furthermore, the register 6 is controlled so that when the data it holds becomes "0", it remains at "0" until the access signal becomes "1".

The subtraction circuit 7 is a circuit that subtracts by "1" the contents held in the register 6, which is supplied via the connection 113, and is supplied to the register 6 via the connection 111.

The subtraction circuit 8 calculates a difference between the shifted bank addresses supplied via a connection 109 to a connection 113.
These are compared by subtracting the contents held in the register 6, which is supplied via a connection 114, and the sign of the subtraction result is supplied to the inverting circuit 9 via a connection 114.

The inversion circuit 9 inverts the polarity of the code supplied via the connection 114, and an access start enable signal is sent out via the connection 115.

In this embodiment, the maximum number of memory banks is 256, and the maximum possible number of data elements is:
If the number of calculation pipelines is 4, it is 256, if it is 2, it is 128, and if it is 1, it is 64, so the connections 101 to 105, 107, and 109 are 8 bits wide.
Furthermore, the memory bank cycle time is assumed to be 16 clock cycles, and the connections 110, 111, and 113 are also expressed in 8-bit width.

Next, the operation of one embodiment of the present invention will be described in detail with reference to the time charts of FIGS. 2, 3, and 4. Here, Figure 2 shows an example where the number of calculation pipelines is 4 and the number of memory banks is 256.
The figure shows an example where the number of calculation pipelines is 2 and the number of memory banks is 256. Figure 4 is an example where the number of calculation pipelines is 4 and the number of memory banks is 128. The number of data elements is the preceding data and the following data. Both numbers are 256 in Figures 2 and 4, and 128 in Figure 3.

Referring to FIG. 2, the bank address of the leading element of the preceding data is "0", and the bank address of the leading element of the succeeding data is "193". First, when an access request for preceding data is issued, the bank address "0" of the leading element supplied via the connection 101 and the number of elements "256" supplied via the connection 102 are supplied to the adder circuit 1. Bank address “255” of the final element of the preceding data is taken into register 2.

The bank address "255" of the last element of the preceding data is held in the register 2 until the sending of an access request for subsequent data to the memory is started. When a subsequent data access request is issued, the bank address “255” of the last element of the preceding data held in the register 2 by the subtraction circuit 3 and the bank address of the first element of the subsequent data supplied by the connection 101 are The difference from address "193" is "62", that is, "00111110" is calculated in binary representation.

When the number of memory banks is 256, no operation is performed in the gate circuit 4, and when the number of arithmetic pipelines is 4, the shift circuit 5 is shifted 2 bits to the right. is supplied with "00001111" in binary representation, that is, the value "15". This value means that an access request for an element of subsequent data will be sent to the bank of the last element of previous data 15 clock cycles after the start of sending a request to access memory for subsequent data.

At timing T ₆₃ when the access request for the final element of the first data is sent to the memory, connection 1
The logic value of 12 becomes "1" and register 6 is supplied with the memory bank cycle time "16" via connection 110. At timing T ₆₄ , no access request to the memory is sent due to the preceding data, and the arithmetic circuit 8 receives a difference of "15" between the shifted bank addresses supplied via the connection 109.
The content "16" held by the register 6, which is supplied via the connection 113, is subtracted from . Since the difference is negative, "1" is output from the connection 114, and the access start valid signal is set to "0" by the inverting circuit 9. As a result, no access request to the memory is sent at timing _T64 , and connection 1
A logical value “0” is supplied from register 6.
The content held in is subtracted by the value "1" by the subtraction circuit 7, and becomes "15".

At the next timing T ₆₅ , the value held in register 6 is "15", and the value held in register 2 is "255" because the second data access request has not yet been sent to the memory. ”, the result of the subtraction circuit 8 becomes “0”, and “0” is sent from the connection 114, becomes “1” in the inversion circuit 9, and is sent as the access start valid signal via the connection 115. A second data access request begins to be sent to the memory.

Next, referring to FIG. 3, it is assumed that the bank address of the leading element of the preceding data is "128" and the bank address of the leading element of the succeeding data is "193". Second
Similarly to the example shown in the figure, when an access request for preceding data is issued, in adder circuit 1, connection 1
The bank address "255" of the final element is calculated from the bank address "128" of the first element supplied via the line 101 and the number of elements "128" supplied via the connection 102, and is taken into the register 2.

When a subsequent data access request is subsequently issued, the subtraction circuit 3 selects the bank address “255” of the last element of the preceding data held in the register 2 and the beginning of the subsequent data supplied via the connection 101. The difference from the element bank address "193" is "62", that is, "00111110" in binary representation is calculated.

Since the number of memory banks is 256 and the number of arithmetic pipelines is 2, the bank address difference "62" is shifted to the right by 1 bit in the shift circuit 5, and is sent to the subtraction circuit 8 via the connection 109 as "00011111" in binary representation. ”, that is, the value “31” is supplied.

At timing T ₆₃ when the access request for the final element of the first data is sent to the memory, connection 1
The logical value of 12 is "1" and register 6 is supplied with the memory bank cycle time "16" via connection 110. At timing _T64 , the access request to the memory based on the preceding data has already been completed, and the subtraction circuit 8 subtracts the value "16" held in the register 6 from the shifted bank address difference "31". Since the difference becomes positive, "0" is output from the connection 114, and the inverting circuit 9 outputs the accelerator start enable signal for the subsequent data from the connection 115 as "1".
Sending of subsequent data access requests to the memory is started.

Finally, referring to FIG. 4, the bank address of the leading element of the preceding data is set to "0", and the bank address of the leading element of the succeeding data is set to "65". When an access request for preceding data is issued, in the adder circuit 1, the bank address "0" of the leading element supplied via the connection 101 and the connection 102 are processed.
The value “255” from the number of elements “256” supplied via
is obtained and taken into register 2. Here, since the number of memory banks is 128, the bank address of the final element of the preceding data is the value “127” modulo 128.
However, since adder circuit 1 and register 2 have a width of 8 bits, the value "255" is taken into register 2.

When an access request for subsequent data is subsequently issued, the subtraction circuit 3 subtracts the bank address "65" of the subsequent data supplied via the connection 101 from the value "255" held in the register 2. The difference "190", ie "10111110" in binary representation, is supplied to the gate circuit 4.

Since the number of memory banks is 128, the gate circuit 4 sets the most significant bit of the 8-bit data supplied via the connection 105 to "0". In this case, the output “190” of the subtraction circuit 3 is transferred to the gate circuit 4.
is supplied to the shift circuit 5 as "00111110", that is, "62" in binary representation. Also, since the number of arithmetic pipelines is four, the shift circuit 5 receives the value "62" supplied via the connection 107. ” is on the right 2
The bit shifted value "15" is supplied to the subtraction circuit 6 via connection 109.

On the other hand, at timing T ₆₃ , the last access request for the preceding data is sent to the memory, and register 6
The memory bank success time “16” is taken in, and at timing T ₆₃ , the output of the subtraction circuit 8 becomes negative, so it is not possible to start accessing the subsequent data, and the value held in the register 6 becomes “ 1” is subtracted and becomes “15”. Therefore, the timing
At T ₆₅ , the calculation result of the subtraction circuit 8 becomes "0",
“0” is sent from the connection 114 and the inverting circuit 9
As a result, the signal becomes "1" and is sent as a subsequent data access start valid signal via the connection 115, and the sending of a subsequent data access request to the memory is started.

〔Effect of the invention〕

As explained above, according to the present invention, the address information of the last element of the first data that is accessed in advance and the address of the first element of the second data that is accessed subsequent to the first data By providing a correction means that corrects the difference in information using a mode signal that specifies the number of elements that can be accessed simultaneously, even if the number of elements that can be accessed simultaneously is variable, after the first data access request has been sent, , it is possible to correctly generate the access start enable signal for the second data, which has the effect of improving hardware usage efficiency and effective performance.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of a memory access control device according to the present invention, and FIG.
FIGS. 3 and 4 are time charts showing an example of the operation of the present invention, respectively. 1... Addition circuit, 2... Register, 3... Subtraction circuit, 4... Gate circuit, 5... Shift circuit, 6
...Register, 7, 8...Subtraction circuit, 9...Inversion circuit, 101-115...Wiring connection.

Claims (1)

[Claims] 1. For a memory that is composed of a plurality of memory units that can be accessed independently of each other and that are addressed in the order of the memory units, each element is arranged consecutively on the memory and is composed of one or more elements. A memory access control device for controlling access to data, comprising: element number supply means for supplying the number of data elements; address information supply means for supplying address information of a memory unit of a leading element; and the number of elements of the data. address information calculation means for calculating memory unit address information of the last element from the memory unit address information of the first element; and address information holding means for holding the memory unit address information of the last element; a time information calculation means for calculating time information until a memory unit is no longer in use; a time information holding means for holding the time information; and a final element of the first data held in the address information holding means. A difference calculating means for calculating a difference between address information in a memory unit and address information in a memory unit of a first element of second data that is accessed subsequent to the first data, and specifying the number of elements that can be accessed simultaneously. a correction means for correcting the difference in address information calculated by the difference calculation means according to a mode signal, and a difference between the address information corrected by the correction means and the time information held in the time information holding means; A memory access control device comprising: a comparison means for comparing the data; and a means for making an access request for the second data based on the result of the comparison means.