JPS585480B2 - Magnetic bubble memory control method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a memory control method for reducing access time when reading or writing a large number of words in a magnetic bubble memory device.

In conventional memory control methods, in order to reduce access time, while magnetic bubble information is being read on a major loop or major line, other magnetic bubble information in the same magnetic bubble memory element is transferred to the major loop or major line. However, at the moment of transfer-out, noise occurs in the magnetic bubble information, making it impossible to read the magnetic bubble information correctly.

FIG. 1 is for explaining a conventional magnetic bubble memory control method. In order to clarify the present invention, which will be described later, FIG. A control method for shortening access time in the case of writing will be explained below.

FIG. 1a shows a magnetic bubble memory module (hereinafter simply referred to as a module) 1 that includes a plurality of magnetic bubble memory elements (hereinafter simply referred to as a chip), and the magnetic bubble memory device includes a plurality of modules 1. Contains.

It is assumed that one chip in the module 1 is, for example, of the well-known major liner loop type shown in FIG.

One chip has multiple memory areas (64 in this case).
) liner loop 10 and read/write transfer area 1
Measure loops 11 are connected by a transfer gate section 12, and a detector 13 for reading out magnetic bubble information from the chip and a generator 14 for writing magnetic bubble information into the chip are provided on the measure loop 11.

The addressing within one module 1 is such that the magnetic bubble located at one bit position of the first Mylar loop 10 in each chip is at address O (the address is indicated by AD).

The same applies to Figure 3 below), the next liner loop 10
The magnetic bubble at the same bit position is address 1,...
..., the magnetic bubble at the same bit position in the last liner loop 10 is set to address 63, and so on, and 64
The address is given to the magnetic bubble of the first liner loop 10 at the next bit position whose distance from address 0 is more than half the length of one liner loop 10, and the address 12727 is O
Other liner loops 1 in the same way as from address to address 63
It is given on top of 0.

It is attached to the magnetic bubble of the first liner loop 10 at the next bit position of the next address 12828, and 19191
Addresses are sequentially assigned to other liner loops 10 in the same way as addresses O to 63, and thereafter addresses are assigned in the same manner.

Once the addressing within one module 1 is completed, subsequent addresses are assigned to other modules 1 in the same manner.

Incidentally, a rotating magnetic field that rotates at a constant period T within the plane of the drawing is applied to each module 1 from the outside.

Normally, the magnetic bubble in each chip is transferred to the next bit position on the same liner loop 10 every period T by this rotating magnetic field, and it takes m times the period T to move on the same liner loop 10 in the direction of the arrow. It has come full circle.

Here, m is the number of bit positions within one liner loop 10, and here m=256.

Although the phase of the rotating magnetic field may be different for each chip in the module 1, here, for the sake of simplicity, it is assumed that the phase of the rotating magnetic field applied to each chip in the module 1 is equal.

Now, let us consider a case where a read or write request for a large number of words occurs in a conventional magnetic bubble memory device including the module 1 shown in FIG. 1A.

As an example, a control method for shortening the access time in a conventional magnetic bubble memory device when a request to read from address 0 to address 19191 occurs will be briefly described.

When a read request arrives at the magnetic bubble memory device, a rotating magnetic field is applied to module 1 containing the first read address, that is, address O, and this module 1 starts transferring the magnetic bubbles in each chip. The position of the magnetic bubble of each chip in module 1 immediately before starting is
(Set as shown in Figure a).

The magnetic bubbles on all the liner loops in module 1 are transferred to the next bit position in synchronization with the rotation of the rotating magnetic field, so the magnetic bubbles on all the liner loops in module 1 are transferred to the next bit position from the start of transfer. It is sufficient to monitor whether the data has been transferred to the specified position using a counter (referred to as a liner counter) provided for each module.

According to the rotation of the rotating magnetic field, as shown in Figure 1,
When the gate signal is applied to module 1 when the liner counter indicates that the intervening magnetic bubble at address 63 has just arrived at the transfer gate section 12,
The magnetic bubbles at addresses 0 to 63 are all simultaneously transferred from the liner loop 10 onto the middle loop 11 via the transfer gate section 12.

This is called transfer out.

The magnetic bubbles transferred out onto the mesh loop 11 are normally arranged every other bit on the mesh loop 11.

Addresses 0 to 63 transferred out in this way
Each magnetic bubble at the address is then transferred to the next bit position on the meji child loop 11 every period T of the rotating magnetic field, but at this time, each magnetic bubble in the module 1 other than addresses 0 to 63 is Since it is not transferred out, it is still being transferred to the next bit position on the same liner loop 10 every cycle T of the rotating magnetic field.

After each magnetic bubble at addresses 0 to 63 is transferred out, the bit position on the meji child loop 11 to which it has been transferred is monitored by a counter (referred to as a measure counter) provided for each module 1. That's enough.

As shown in FIG. 1C, if a read instruction signal is applied to module 1 just before the measure counter indicates that the magnetic bubble at address 0 has just arrived at the position of detector 13, the magnetic bubble information at address 0 will be read. be led.

Thereafter, the read instruction signal continues to be applied to module 1,
1st address, 2nd address, every twice the period T of the rotating magnetic field.
..., when each magnetic bubble passes through the detector 13 in the order of address 63, each magnetic bubble information is read one after another.

As shown in FIG.
When a gate signal is applied to module 1 when the liner counter indicates that position 2 has just been reached, each magnetic bubble at addresses 64 to 127 is transferred out from liner loop 10 onto mejiko loop 11. .

In this case, on the liner loop 10, addresses O to 63 and 6
Since the numbering of addresses 4 to 127 was selected in accordance with the positional relationship described above, on the mejiko loop 11, the magnetic bubble at address 64 is located at the bit position two bits after the bit position occupied at that time by the magnetic bubble at address 63. The bubble is transferred out.

Therefore, the magnetic bubble information at addresses 64 to 127 is read immediately after the magnetic bubble information at address 63 at intervals of twice the period T of the rotating magnetic field, resulting in the shortest access time.

The magnetic bubble information at addresses 128 to 191 that will be read subsequently is read in the same manner.

Even after each magnetic bubble information is read, Mejiko loop 11
However, when it returns to the original position of the transfer gate section 12, the original liner loop 10
are simultaneously returned to the inside.

This is called transfine.

However, this transfer may not be necessary depending on the chip structure within the module 1.

The above is the control method for shortening the access time of the conventional magnetic bubble memory device in response to a read request from address 0 to address 19191, but the same method as described above is repeated for requests to read a larger number of words. It is better to do this.

On the other hand, in the case of a write request, new magnetic bubble information may be written at the timing when each magnetic bubble passes through the generator 14 instead of being read by the detector 13.

In the case of this write, new magnetic bubble information must be stored in the liner loop 10, so when the new magnetic bubble information reaches the position of the corresponding transfer gate section 12 on the mejiko loop 11, transfer is performed. It is necessary.

However, the control method for shortening the access time when reading or writing a large number of words in the conventional magnetic bubble memory device described above has the following drawbacks.

That is, as shown in FIG. 1d, the magnetic bubble information (0
When another magnetic bubble information (addresses 64 to 127) is transferred out on the same chip while it is being read, this transfer out occurs. Noise is likely to occur in the detection signal of the detector 13 due to the influence of the gate signal instantaneously applied to the transfer gate section 12, and it is difficult to read the magnetic bubble information correctly at that instant.

Therefore, among the plurality of liner loops, one liner loop 10 that stores an address corresponding to this position (for example, address 61) cannot be used as a normal address.

In other words, address 61 is actually the next liner loop 10.
It is numbered.

However, doing so would instead address and use one of several spare liner loops 10 that are normally provided within the chip to save some defective liner loops 10 that occur during chip manufacturing. This has the drawback that the number of spare liner loops 10 allocated to defective liner loops 10 that occur during chip manufacturing is reduced by one, which reduces the yield during chip manufacturing and increases the price of the chip.

Furthermore, when the read or write address does not stay in one module 1 but extends to the next module 1, the rotation of the next module 1 is started as soon as it is known that the address has been straddled. There is also the disadvantage that it takes longer.

An object of the present invention is to provide a memory control method for a magnetic bubble memory device that eliminates the above-mentioned drawbacks of the prior art and can shorten access time when reading or writing a large number of words without reducing chip yield. It is to provide.

For this purpose, the present invention provides that while magnetic bubble information is being transferred and read or written on a read/write transfer area, another magnetic bubble information is transferred from a storage area to a read/write transfer area within the same chip. This feature is characterized in that it is not transferred upwards.

In order to make the access time equivalent to that of the conventional memory control method, if addresses corresponding to the number of fixed storage areas in one chip are assigned to one module, the next address will be assigned to another module. When a read or write request occurs after a certain address, a rotating magnetic field is simultaneously applied not only to the tr module including the first address but also to several other modules, and one module is This control system uses a control method in which when reading or writing of a target address within a module is completed, the next address is immediately read or written from another module.

That is, this is a control method in which consecutive addresses are sequentially read or written across a plurality of modules.

The present invention will be explained below with reference to FIGS. 2, 3, and 4.

FIG. 2 is a block diagram of memory control of a magnetic bubble memory device to which the present invention is applied.

This magnetic bubble memory device includes two modules 1, 2, and the addressing within each module 1, 2 is not as shown in FIG. 1, but as shown in FIG. 3.

That is, multiple pieces (64 pieces here) in module 1
First, addresses 0 to 63 are assigned to each magnetic bubble located at the same bit position of the liner loop 10, and then addresses 64 to 12 are assigned to each magnetic bubble at the same bit position of the liner loop 10.
Address 7 is assigned to each magnetic bubble in module 2 located at the next bit position whose distance from addresses 0 to 63 in module 1 is more than half the length of one liner loop 10. be.

In addition, addresses 128 to 191 are assigned to each magnetic bubble located at the next bit position of addresses 0 to 63 in module 1, and in the same way, addresses are assigned to module 1 and module 2 alternately for every 64 address. It will be done.

In FIG. 2, a rotating magnetic field generating circuit 140, 145 for applying a rotating magnetic field to modules 1 and 2, and module address registers 180, 280, . Major address registers 181 and 281 and liner address registers 182 and 282 are provided for modules 1 and 2, respectively.

A liner counter 160 is provided in common for modules 1 and 2, and is used to determine which bit position the rotating magnetic field is transferred to from when the rotating magnetic field begins to be applied to each magnetic bubble on all liner loops 10 and 20 in modules 1 and 2. This is to count how many people have done it.

Liner matchers 183 and 283 are module 1, respectively.
, 2, and liner address registers 182, 282 and liner counter 160 of module 1.2.
Compare the values of liner address registers 182 and 282
When the values of the liner counter 160 match, a match signal is output.

Measure counters 184 and 284 are each module 1
, 2, and counts to which bit position each magnetic bubble transferred out onto the major loop 11.degree. 21 has been transferred after transfer out.

Decoders 187 and 287 each have a measure counter 18
This is for decoding the contents of 4,284.

Major Matsunaya 185 and 285 are module 2 respectively.
It compares the values of the major address registers 181, 281 and the major counters 184, 284 of the modules 1 and 2, and outputs a match signal when the values of the major address registers 181, 281 and the major counters 184, 284 match. Output.

Read/write display flip-flops 186 and 286 are provided for modules 1 and 2, respectively, and are set when they receive match signals from measure matches 185 and 285, and instruct modules 1 and 2 to read or write.

A word counter 190 is provided to count how many words have been read or written to module 1 or 2.

The fixed pattern generation circuit 150 is for inputting fixed patterns to the module address registers 180, 280, media address registers 181, 281, and liner address registers 182, 282 at certain timings.

The buffer register 100 is used to temporarily store commands such as read/write commands exchanged between the magnetic bubble memory device and the data channel device 50, address information in the magnetic bubble memory device, read data, write data, etc. be.

The data register 110 is for storing data read from or written to the modules 1 and 2.

The command register 120 is for storing commands such as read and write commands sent from the data channel device 50 via the buffer register 100.

Decoder 130 is for decoding the contents of command register 120.

Now, let us consider a case where a read or write request for a large number of words occurs in the magnetic bubble memory device shown in FIG.

As an example, a control method when a request for reading from address O to address 191 shown in FIG. 3 is generated will be described.

When a read command arrives at the buffer register 100 from the data channel device 50, the read command is entered into the command register 120 and then decoded by the decoder 130.

If it is determined by this decoding that it is a read command, the decoder 130 activates the rotating magnetic field generating circuits 140 and 145.

This causes a rotating magnetic field to be applied to the modules 1, 2 and initiates the transfer of each magnetic bubble within the liner loops 10, 20 within the modules 1, 2.

Decoder 130 also indicates to modules 1 and 2 that the command is a read command.

Additionally, decoder 130 also activates liner counter 160.

Next, address information indicating which address to read from the data channel device 50 arrives at the buffer register 100 .

This address information is major address information A6 as shown in Figure 4.
It consists of bits (liner loop numbers 0 to 63), module address information B 1 bits (module numbers 0, 1), and minor address information C 7 bits (minor loop addresses 0 to 127). In this case, the first read address is Since the address is 0, all address information is n Ott.

This address information stored in the buffer register 100 is
Since address 0 is included in module 1, in other words, 1 bit of module address information is "0", module address information B is stored in module address register 180 for module 1, and major address information A is stored in module 1. Major address register 181 for
In addition, the liner address information C is transferred to the liner address register 182 for module 1, respectively.

The liner matcher 183 uses the liner address register 182
Since the content power of is “0”, the liner counter 16
A match signal is output when the content of 0 becomes "0".

The contents of the liner counter 160 become "0" when each of the magnetic bubbles at addresses O to 63 exactly reach the position of the transfer gate section 12, and are incremented by "1" every cycle T of the rotating magnetic field. The content is decimal “255”
It is assumed that the content is set in advance so that when it becomes ", the next content becomes "0" (note that during this time, each magnetic bubble on the liner loop 10 is making exactly one round on the liner loop 10).

Therefore, when a match signal is output from the liner matcher 183, the module 1 is instructed to transfer out, and the magnetic bubbles at addresses O to 63 that have just reached the transfer gate section 12 are transferred to the middle loop 11.
transferred out on top.

Furthermore, a match signal from the liner matcher 183 simultaneously activates the measure counter 184.

The content of the measure counter 184 becomes "0" when the magnetic bubble at address 0 exactly reaches the position of the detector 13.
It is assumed that it is set in advance so that when it is stepped every twice the period T of the rotating magnetic field and the content becomes "63" in decimal notation, the next content becomes "0".

In this case, the content of the major address register 181 is “0”.
”, the magnetic bubble at address 0 is detector 1.
3 position is reached, and the content of the measure counter 184 becomes “0”.
When this happens, the measure matcher 185 outputs a match signal and sets the read/write instruction flip-flop 186.

When the read/write instruction flip-flop 186 is set, a read/write instruction is issued to the module 1, and the word counter 190 is activated.

Then, each magnetic bubble information at addresses 0 to 63 is sequentially read out from the module 1 and sent to the data channel device 50 via the data register 110 and the buffer register 100.
sent to.

While the read/write instruction flip-flop 186 is set, the word counter 190 increments by one each time each piece of magnetic bubble information at one address is read from the module 1.

The word number counter 190 outputs an output to the fixed pattern generation circuit 150 when its content reaches a specific value that can be arbitrarily set, for example, "30" in decimal notation.

In this case, reading starts from address O of module 1, so from the time the contents of address 29 are read to the last 191
The output from word number counter 190 is input to fixed pattern generation circuit 150 until address 91 has been extracted.

Therefore, at a certain timing after the magnetic bubbles at addresses 0 to 63 are transferred out onto the meji child loop 11, for example, when the content of the major counter 184 becomes "35" in decimal, the decoder 187 outputs emits,
Its output is applied to fixed pattern generation circuit 150.

At this point, the output from the word number counter 190 has already been input, so the fixed pattern generation circuit 150 is activated.

The fixed pattern generation circuit 150 forcibly inputs “1” into the module address register 280 for module 2, “0” into the major address register 281, and decimal “64” into the liner address register 282.

Therefore, while each magnetic bubble from addresses O to 63 in module 1 is transferred on the mejiko loop 11, the liner matcher 283 for module 2 changes the contents of the liner address register 282 ("64" in decimal number) The contents of the liner counter 160 continue to be compared.

When the value of the liner counter 160 reaches 64'' in decimal notation, the liner matcher 283 outputs a match signal.

The match signal instructs the module 2 to transfer out, and each magnetic bubble at addresses 64 to 127, which has just reached the transfer gate section 22,
1 so that it is transferred out.

In this case, addresses 0 to 63 in the liner loop 10 in module 1 and the liner loop 20 in module 2
The numbering of addresses 64 to 127 in
Since we have chosen the particular positional relationship shown in the figure, in module 1, the bit position 2 bits after the bit position currently occupied by the magnetic bubble at address 63 (however, in module 2) is at address 64. magnetic bubbles are being transferred out.

Furthermore, a match signal from the liner matcher 283 simultaneously activates the measure counter 284.

In module 1, magnetic bubble information from addresses 0 to 62 is sequentially read out, and at the next timing when the magnetic bubble information at address 63 is finally read out, the content of the major counter 284 for module 2 becomes "0". , Mesha Matcha 2
85 outputs a match signal and sets a read/write instruction flip-flop 286.

When the read/write instruction flip-flop 286 is set, a read/write instruction is issued to the module 2.

The magnetic bubble information at addresses 64 and 127 from the module 2 is sequentially read out, and the information on the magnetic bubbles at addresses O to 63 in the module 1 is sequentially read out from the data register 11.
0 is now sent to the data channel device 50 via the buffer register 100.

Each magnetic bubble from address 64 to address 127 is Mejiko loop 1
At a certain timing after being transferred out to 1,
For example, the content of the major counter 284 is “35” in decimal notation.
”, the output from the decoder 287 is input to the fixed pattern generation circuit 150.

At this point, the output from the word number counter 190 has already been input, so the fixed pattern generation circuit 150 is activated.

The fixed pattern generation circuit 150 then forcibly inputs “0” into the module address register 180 for module 1, “0” into the major address register 181, and decimal “1” into the liner address register 182. .

After this, each magnetic bubble information at addresses 128 to 191 in module 1 is transferred to the data register 110 following each magnetic bubble information at addresses 64 to 127 in module 2.
, the control method when data is sent to the data channel device 50 via the buffer register 100 is as follows:
727 address.

In this way, when the last bubble information for address 19191 is sent to the data channel device 50, the end of the read command is sent from the data channel device 50 to the magnetic bubble memory device, and a read request for addresses 0 to 191 is sent to the data channel device 50. finish.

If a request is made to read a larger number of words than this, it is understood that the same method as described above can be repeated.

In addition, in the case of a write request, new magnetic bubble information may be written at the timing when each magnetic bubble passes through the generator 14.24 instead of being read by the detector 13.23.

Although the above is for the case where the number of modules is two, it is clear that the above control can be easily extended to the case where the number of modules is three or more.

In this example, when the read/write request is for 29 words or less, the fixed pattern generation circuit 1
Although the operation when 50 words are not activated is shown, the number of words can be set arbitrarily.

In the present invention, as in the above-mentioned actual sister example, magnetic bubble information is transferred on the read/write transfer area, and while it is being read or written, another magnetic bubble information is transferred from the storage area for read/write within the same chip. Not transported over the area.

However, instead, other magnetic bubble information is transferred from the storage area to the read/write transfer area at the same time in another chip.

Therefore, unlike the conventional magnetic bubble memory device, the yield during chip manufacturing does not decrease, and the access time can be shortened to the same level as the conventional magnetic bubble memory device.

[Brief explanation of drawings]

1A to 1D are schematic diagrams showing the addressing of each magnetic bubble memory element in one magnetic bubble memory module in a conventional magnetic bubble memory device and changes over time of magnetic bubbles at each address; The figure is a block diagram of a magnetic bubble memory device that is a sister example of the present invention, and FIGS. 3a and 3b show each magnetic bubble memory element in each magnetic bubble memory module in the magnetic bubble memory device that is a sister example of the present invention. FIG. 4 is a diagram showing the structure of address information sent from a data channel device to a magnetic bubble memory device in a sister example of the present invention. 1.2... Magnetic bubble memory module, 120... Command register, 140, 145... Rotating magnetic field generation circuit, 150... Fixed pattern generation circuit, 160... Minor counter, 180, 280... Module Address register, 181, 281...Measure address register, 1
82,282...minor address register, 183,2
83...Mainamatsunaya, 184,284...Mejiya counter, 185,285...Mejiya matsunaya, 186°286...Read/write instruction flip-flop, 190
...word counter.

Claims (1)

[Claims] 1. Memory control in a magnetic bubble memory device configured using a magnetic bubble memory element consisting of a plurality of loop-shaped storage areas and one or more loop-shaped or line-shaped read/write transfer areas. In this method, when a magnetic bubble memory device is configured using a plurality of magnetic bubble memory modules each consisting of a plurality of magnetic bubble memory elements, and a memory access request for one or more words is made to the magnetic bubble memory device. is a magnetic bubble characterized in that a rotating magnetic field is simultaneously applied to a magnetic bubble memory module including an address to which a memory access request has been made and an arbitrary number of other magnetic bubble memory modules among the magnetic bubble memory modules. Memory control method. 2. One or more address registers indicating addresses within the magnetic bubble memory module are provided in the magnetic bubble memory device, and when a memory access request is for a specific number of words or more, one or more address registers are specified for an arbitrary number of address registers. 2. The magnetic bubble memory control system according to claim 1, wherein a specific value is set at each time. 3. The magnetic bubble memory control system according to claim 2, wherein an address register is provided for each magnetic bubble memory module. 4. The magnetic bubble memory control system according to claim 3, wherein a word count counter is provided to determine whether the number of words is greater than or equal to a specific number of words.