JPS61175837A - Multiple memory access system - Google Patents

Abstract

PURPOSE:To minimize the access waiting time by setting that which can execute access first to each storage medium, to the first access order, and as for the remainder, setting that which can execute access first, after the preceding access has been ended, to the next access order. CONSTITUTION:Magnetic bubble storage mediums M0-M3 are divided into three areas, respectively, and an information store area of logical addresses 1, 5 and 9 is assigned to the medium M0. In the same way, information store areas of logical addresses 10, 2 and 6; 7, 11 and 3; and 4, 8 and 12 are assigned to the mediums M1, M2 and M3, respectively, in the same order. These information store areas 1-12 rotate counterclockwise at a constant speed in the corresponding mediums M0-M3. Accordingly, when the information store area 1 of the medium M0 has passed completely through an accessible position (DET), the storage area 2 of the medium M1 reaches the DET position, and subsequently, the storage area 3 of the medium M2 reaches the DET position. In this way, the continuity of the logical address is held.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a multiple access method for a storage device used in an information processing system such as a switchboard or an electronic computer.

Conventional Technology In sequential access storage media, such as magnetic drums, magnetic bubbles, and COD, in which the information storage area cannot be randomly accessed, the information storage area to be accessed must wait until the information storage area to be accessed reaches the access position of the read/write head, etc. 1 is inherently associated with access latency.

In a storage device configured with such a sequential access storage medium, a multiple access method is adopted in order to reduce the above-mentioned access waiting time as much as possible.

In this multiple access method, information is distributed and stored in multiple sequentially accessed storage media, and instead of processing access requests that occur one after the other to multiple storage areas in the order in which they occur, information to be accessed is This method processes access requests by sorting them in descending order of access waiting time based on the distance between the storage area and the accessible location.

Problems to be Solved by the Invention In the above-mentioned conventional multiple access system, when assigning access orders to access requests, subsequent access orders are determined without considering the time point at which the previous access ends.

As a result, if the preceding access has not finished when the subsequent access start position reaches the accessible position, even if the preceding access ends, the subsequent access will start after that access ends. There is a problem in that access must wait until the position reaches an accessible position again, resulting in unnecessary access waiting time.

In order to compensate for the above-mentioned shortcomings, it is possible to limit access to single word units, or to improve memory allocation for software processing and access/access while being fully aware of the hardware structure.
Increasing difficulties arise, such as having to set blocks.

Means for Solving the Constituent Problems of the Invention The multiple memory access method of the present invention, which solves the problems of the prior art described above, selects the one that can be accessed first among multiple access requests to each storage medium. The access order is set as the access order, and for the remaining access requests, the next access order is the one that can be accessed first after the end of the previous access, thereby reducing the access waiting time from the end of the previous access to the start of the subsequent access. It is configured to minimize.

Hereinafter, the operation of the present invention will be explained in detail by way of examples.

Embodiment FIG. 1 is a block diagram showing the configuration of a magnetic bubble storage device to which an embodiment of the present invention is applied.

This magnetic bubble storage device is composed of a memory control device (MC) and four magnetic bubble recording media (MO to M3) connected to the memory control device in a time division multiplex manner.

Information on each recording medium is accessed at an information read gate (CDET) and an information write gate (GEN).

Each recording medium is divided into three areas in this example, and an information storage area with a logical address of 1.5.9 is allocated to the recording medium MO, and similarly, the recording medium Ml.

M2. Each of M3 has a logical address of 10°2.6
;7,11.3:4. s, 12 information storage areas are allocated in the same order. These information storage areas 1 to 12 are rotated at a constant speed in the direction of the arrow in FIG. 2 (counterclockwise) within the corresponding recording medium M OM 3. Therefore, when the information storage area 1 of the recording medium MO finishes passing through the accessible position, the information storage area 2 of the recording medium M1 reaches the accessible position, and then the information storage area 3 of the recording medium M2 reaches the accessible position. The continuity of logical addresses is maintained such that .

In the magnetic bubble storage device described above, each recording medium MO
The recording capacity of M3 to M3 is 160 words each, and the recording medium M1 is shown with diagonal lines in FIG. 2(a).
Assume that access requests are made to four information storage areas on M3.

That is, based on the leading position of each recording medium.

20 words starting from the 20th word for the recording medium MO, 40 words starting from the 60th word for the recording medium M1, 40 words starting from the 80th word for the recording medium M2, the recording medium As for M3, it is assumed that an access request is generated for 40 words starting from the 0th word and accepted by the memory control device f (MC).

At the time point in FIG. 2(a) when the above-mentioned access occurs, the recording medium that can be accessed first is M3. Therefore, the first access order is assigned to this M3. At the end of this earliest access, the state of each recording medium is as shown in FIG. 2(b). Therefore, the remaining recording medium (with an undetermined access order) that can be accessed first after the end of the earliest access is M1, and the second access order is assigned to this recording medium M1.

At the end of the second access, the state of each recording medium is as shown in FIG. 2(C). Therefore, 2
The remaining recording medium that can be accessed first after the end of the th access is recording medium M0, so this recording medium M
The third access order is assigned to O, and the last access order is assigned to the remaining recording medium M2, as shown in FIG. 2(d).

All accesses end after the state shown in (e).

The explanation regarding the allocation of the above-mentioned access order will be supplemented with reference to the table of FIG. 3.

To in the table is the standardized access waiting time until access becomes possible in each of the recording media MO to M3, and these correspond to the access start word position of each recording medium. On the other hand, T1 in Table 9 is the recording medium M O
-M These are the standardized access end points when access is started independently in each of the 3, and these correspond to the access end words of each recording medium.

When the memory control unit (MC) determines that the four access requests should be processed in parallel by rearranging the large number of access requests, it creates the table shown in FIG. 3 for each access request.

The memory control unit (MC) searches the table created in 9 to minimize the access latency (in this example, O).
Assign the first access order to the recording medium M3 0th then 1
The end time of the first access (T1=40) is read from the table, and the remaining three recording media MO, M1 . From M2, the one that becomes accessible first after the end of the earliest access is searched based on the access waiting time TO in the table.

Since the recording medium M1 with a waiting time of Tl-60 corresponds to this, the second access order is assigned to it.

Continue from Table 1 to the end point of the second access (T
I=100) is read, and the remaining two recording media MO and M
2, the one that becomes accessible first after the end of the second access is searched based on the access waiting time To in the table. In the remaining recording medium, the maximum waiting time T 1
If there is no corresponding recording medium with a waiting time equal to or less than max (= 160), the maximum waiting time T1max (-160) is added to these waiting times, and the above search is repeated. As a result, the third access order is assigned to the recording medium MO, and the last access order is assigned to the remaining recording medium M2.

The case where the information storage area in each recording medium rotates at a constant speed has been exemplified above, but the multiple access system of the present invention is configured to allocate an access order based on the access waiting time and the access end point.

Each information storage area does not necessarily need to rotate at the same speed.

In addition, although a magnetic bubble storage device is shown as an example of a sequential access recording medium, other sequential access recording media such as magnetic disks, those that involve mechanical rotation of drums, and COD in which information is rotated by transfer within a static storage medium are also available. The access method of the present invention can be applied to a storage device equipped with.

Effects of the Invention As explained in detail above, the multiple memory access method of the present invention gives the first access request among multiple access requests to each storage medium the highest access order, and the remaining access requests is configured so that the next access is the one that becomes accessible after the previous access ends, so the access waiting time from the end of the previous access to the start of the subsequent access is minimized. can do.

That is, according to the conventional method in which only the access waiting time of each recording medium is considered, the case shown in FIG. 2 occurs.

Recording media M3゜MO in descending order of access waiting time TO
, Ml, and M2 are accessed in this order.

However, at the end of the access to the recording medium M3, the next access cannot be started until the next recording medium MO to be accessed is accessed, resulting in unnecessary access waiting time.

In this way, in the case of Fig. 2, according to the conventional method, it takes three rotations to complete all accesses, but with the 9-Ki-Ai method, all accesses are completed in two rotations.
Access time is significantly reduced.

Furthermore, since this method can be applied without being aware of data allocation, software productivity is also greatly improved.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a magnetic bubble storage device to which a multiple access method according to an embodiment of the present invention is applied;
3 and 3 are conceptual diagrams for explaining the operation of the magnetic bubble storage device of FIG. 1 according to the above embodiment. MC...Memory control device, MO, Ml, M2 and M3...
・Storage medium, DET・・Information readout gate, GEN・・
Information write gate, TO... access waiting time, Tl...
- At the end of access.

Claims (1)

[Scope of Claim] A storage medium comprising a plurality of sequentially accessed recording media, each of which cannot be accessed at the same time, and an access order determining means for determining an access order based on a plurality of access requests to these recording media. The access order determining means sets the access request that can be accessed first among the plurality of access requests as the first access order, and the remaining access requests are the ones that can be accessed first after the previous access ends. A multiple memory access method characterized in that the next access order is .