JPH01195538A - Data flow type information processor - Google Patents

Abstract

PURPOSE:To avoid the joint of an internal process carried out by an internal processing means and a process moving toward outside from an external processing means by adding the 1st and 2nd joint means and a branch means. CONSTITUTION:In case an internal process is carried out by an information processor A via an internal data buffer 5, a data packet is sent successively to a program memory part 1, a pair data detecting part 2, an arithmetic processing part 3, a branch part 4, the buffer 5 and a joint part 7 respectively. In case the data packet is processed by an external data memory 6 and sent to an information processor C, the data packet is sent to the memory 6 from the part 4 via a joint part 21 and a branch part 22 and then to the processor C from the memory 6 via a joint part 23, a branch part 24, the part 21 and the part 22.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a data flow type information processing apparatus, and more particularly to a data flow type information processing apparatus equipped with an external storage device.

[Prior Art] FIG. 6 is a schematic block diagram of a conventional data flow type information processing device, and FIG. 7 is a diagram showing the structure of a data packet circulating through each part of the information processing device of FIG. 6. .

First, the configuration and general operation of a conventional data flow type information processing apparatus will be explained with reference to FIGS. 6 and 7. Destination information is stored in the destination field of the data packet in FIG. 7, command information is stored in the command field, and operand data is stored in the data 1 field or data 2 field.

In FIG. 6, the program storage unit 1 includes a program memory (not shown), and the program memory includes:
As shown in FIG. 8, a data flow program consisting of destination information and command information is stored. The program storage unit 1 reads destination information and command information by addressing based on the destination information of the data packet, stores each information in the destination field and the command field of the data packet, and outputs the stored information.

The paired data detection unit 2 waits for data packets input from the program storage unit 1, and converts the operand data of one of the two data packets with matching destination information to predetermined data of the other data packet. Store it in a field and output it. Note that at this time, one of the data packets mentioned above disappears. Arithmetic processing unit 3
decodes the instruction information of the data packet input from the paired data detection unit 2, performs predetermined arithmetic processing on the two operand data, stores the result in the data field of the data packet, and sends it to the branch unit 4. Output.

Branching unit 4 outputs the data packet to internal data buffer 5 or external data memory 6 based on the destination information of the data packet. Data packets output from the internal data buffer 5 or the external data memory 6 are given to the confluence section 7, and the confluence section 7 supplies these data packets to the program storage section 1 on a first-come, first-served basis.

In the data flow type information processing device shown in FIG. 6, data packets are stored in a program storage section 1→pair data detection section 2→arithmetic processing section 3-branching section 4→internal data buffer 5 or external data memory 6→ By continuing to rotate as in the merging section 7..., arithmetic processing based on the program stored in the program storage section 1 progresses.

FIG. 9 is a diagram showing a general configuration of the program storage section 1 shown in FIG. 6. In FIG. will be deleted.

Note that the input data latch unit 11 also latches operand data. The destination information latched by the input data latch section 11 is given to the address calculation section 12, and the address of the program memory 13 is calculated from the destination information. The program memory 13 stores a data flow program consisting of destination information and command information as shown in FIG. 8 described above.

The new destination information and command information read from the program memory 13 at the above address are output to the output data latch section 14.
is given and latched. Note that the output data latch section 14 latches the operand data latched by the input data latch section 1.1 as is.

[Problem to be solved by the invention]

A multiprocessor system is configured by combining a plurality of the above data flow type information processing devices. In this multiprocessor system, when a data packet output from the external data memory 6 of one information processing device is sent to another information processing device, the data packet must pass through an internal processing path. That is, the data packet output from the external data memory 6 is sent from the branching unit 4 to another information processing device via the merging unit 7, program storage unit 1, paired data detection unit 2, and arithmetic processing unit 3. Become. Therefore, the internal processing that goes through the internal data buffer 5, the program memory section 1, etc. and the processing that goes to the outside from the external data memory 6 are merged at the merging section 7. Therefore, there is a problem in that the speed of internal processing and processing going to the outside becomes slow.

Therefore, as shown in FIG. 10, it is conceivable to connect the output of the external data memory 6 directly to another information processing device without passing through the merging section 7.

In this case, internal data buffer 5 - program storage section 1
The internal processing that goes through the external data memory 6 and the processing that goes to the outside from the external data memory 6 do not merge. However, a problem arises in that the data packets processed by the external data memory 6 cannot be directly used in internal processing without passing through another information processing device.

The main object of the present invention is to provide data flow type information processing in which internal processing and processing directed to the outside from external storage means do not merge, and data packets output from external storage means can also be used in internal processing. The purpose is to provide equipment.

[Means for Solving the Problems] A data flow type information processing device according to the present invention includes an internal processing means for performing arithmetic processing on data according to a data flow program, an external storage means for storing data, and a data flow type information processing device for storing data from the external storage means. a first merging means for receiving data, a first branching means for receiving data from the first merging means, a second merging means for receiving data from the internal processing means, and a second merging means for receiving data from the second merging means. It is equipped with branching means. The first merging means selectively supplies data supplied from the external storage means or data supplied from the outside to the first branching means.

The first branching means selectively supplies the data provided from the first merging means to the internal processing means or the second merging means. The second merging means selectively supplies the data supplied from the first branching means or the data supplied from the internal processing means to the second branching means. The second branching means selectively outputs the data provided from the second merging means to external storage means or the outside.

[Operation] In the information processing device according to the present invention, when data from the internal processing means is sent to the external storage means, it passes through the second merging means and the second branching means.

When data from the external storage means is sent to the outside, it passes through the first merging means, the first branching means, the second merging means, and the second branching means. Further, when data from the external storage means is sent to the internal processing means, it passes through the first merging means and the first branching means. Therefore, it is possible to output data from the external storage means to the outside without passing through the internal processing means, thereby avoiding the merging of internal processing performed by the internal processing means with processing directed to the outside from the external processing means. Can be done.

[Example] An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an embodiment of a data flow type information processing apparatus of the present invention.

In FIG. 1, the connections among the program storage section 1, paired data detection section 2, arithmetic processing section 3, branching section 4, internal data buffer 5, and merging section 7 are the same as in the conventional information processing apparatus shown in FIG. be.

In this embodiment, the output of the branching section 4 is connected to one input of the merging section 21, and the output of the merging section 21 is connected to one input of the merging section 21.
is connected to the input of Further, one output of the branching section 22 is connected to an input of the external data memory 6, an output of the external data memory 6 is connected to one input of the merging section 23,
The output of the merging section 23 is connected to the input of the branching section 24 . Further, one output of the branching section 24 is connected to one human power of the merging section 7 , and another output of the branching section 24 is connected to another input of the merging section 21 .

FIG. 2 is a diagram showing the structure of a data packet circulating through each part of the information processing device shown in FIG. 1.

The data packet consists of a destination field 31, an instruction field 32, a data 1 field 33, and a data 2 field 34. Destination field 31 stores destination information, and command field 32 stores command information. The data 1 field 33 stores operand data (left data), and the data 2 field 34 stores operand data (right data). The destination information includes a destination identification code, a node number, etc. The destination identification code is
Processor external/internal selection bit, external data memory/
It consists of an internal arithmetic processing selection bit, a left/right data selection bit, and a two-manpower/one-manpower selection bit. The node number is
Normally, it represents a node number (branching part or merging part number), and if the destination identification code selects outside the processor, it represents the information processing device number and the entrance node number of that information processing device.

FIG. 3 is a block diagram showing the configuration of the branch portions 4, 22 and 24 shown in FIG. 1. The selection device 41 determines a branch destination according to the destination information of the data packet input to the data latch 42 and instructs the branch arbitration device 43 . Branch arbitration device 43 transfers the data packet held in data latch 42 to data latch 44 or data latch 45 based on a command from selection device 41.

FIG. 4 is a block diagram showing the configuration of the merging section 7.21.23 shown in FIG. 1. When data packets are input to the data latches 52 and 53, the confluence selection device 51 transfers the data packets to the data latch 54 in the order of arrival first.

Next, the operation of the information processing apparatus shown in FIG. 1 will be explained.

The data packet processed by the arithmetic processing section 3 is given to the branching section 4. The branching section 4 transfers the data packet to the internal data buffer 5 or the merging section 21 according to the destination identification code. For example, if internal is selected by the processor external/internal select bit and external data memory/
When internal arithmetic processing is selected by the internal arithmetic processing selection bit, the data packet is transferred to the internal data buffer 5.

The merging unit 21 transfers data packets from the branching unit 4 or data packets from the branching unit 24 to the branching unit 22 on a first-come, first-served basis. The branching section 22 stores the data packets from the merging section 21 in an external data memory 6 according to the destination identification code.
Or transfer it to an external information processing device. For example, when internal is selected by the processor external/internal selection bit and external data memory is selected by the external data memory/internal arithmetic processing selection bit, the data packet is transferred to internal data memory 6, and the processor external/internal selection bit selects external data memory. When external is selected by , and internal arithmetic processing is selected by the external data memory/internal arithmetic processing selection bit, the data packet is transferred to the outside.

The merging unit 23 sends data packets from the outside or data packets from the external data memory 6 to the branching unit 2 on a first-come, first-served basis.
Transfer to 4. The branching section 24 transfers the data packet from the merging section 23 to the merging section 7 or the merging section 21 according to its node number.

In this case, the node number represents the number of the information processing device,
If the number matches the number of the information processing device into which the data packet was input, the data packet is transferred to the merging unit 7, and if they do not match, it is transferred to the merging unit 211.

FIG. 5 is a diagram showing a partial configuration of a multiprocessor system using the information processing apparatus of this embodiment.

The data packet output from the information processing device ifB is given to the merging unit 23 of the information processing device A, and the data packet is outputted from the information processing device ifB.
The data processed in is given to the information processing device C from the branching unit 22.

When internal processing via the internal data buffer 5 is performed in the information processing device A, the data packet is stored in the program storage section 1, paired data detection section 2, arithmetic processing section 3, branching section 4, as shown by a. The data is sent to the internal data buffer 5 and the merging unit 7 in that order. Further, after the data packet is processed by the external data memory 6, another information processing device C
When the data packet is sent to the external data memory 6, as shown by b, the data packet is sent from the branching section 4 via the merging section 21 and the branching section 22 to the external data memory 6, and then from the external data memory 6 to the merging section 23, It is sent to another information processing device C via the branching section 24, the merging section 21, and the branching section 22.

Therefore, data packets processed by the external data memory 6 are quickly sent to another information processing bag fc without passing through the internal route of the program storage unit 1 or the like. Also, since the internal processing via the internal data buffer 5 and the processing going to the outside from the external data memory 6 do not merge,
There is no delay in internal processing. Further, when the data packet processed by the external data memory 6 is sent to the merging section 7 via the merging section 23 and the branching section 24, internal processing via the external data memory 6 is executed.

The data flow type information processing device configured as described above is effective when processing results by the external data memory 6 are referred to by another information processing device.

[Effects of the Invention] As described above, according to the present invention, it is possible to avoid the merging of the processing directed to the outside from the external processing means and the internal processing performed by the internal processing means, thereby speeding up data access. It will be planned.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a data flow type information processing apparatus according to an embodiment of the present invention, FIG. 3 is a block diagram showing the configuration of the branching section of the information processing device in FIG. 1, FIG. 4 is a block diagram showing the configuration of the merging section of the information processing device in FIG. 1, and FIG. A diagram showing the configuration of a multiprocessor system using a processing device, FIG. 6 is a block diagram showing the configuration of a conventional data flow type information processing device, FIG. 7 is a diagram showing the configuration of a data packet, and FIG. 8 is a program FIG. 9 is a block diagram showing the configuration of a program storage unit of a data flow type information processing device. FIG. 10 is another example of a conventional data flow type information processing device. FIG. In the figure, 1 is a program storage unit, 2 is a paired data detection unit, 3 is an arithmetic processing unit, 4, 22.24 is a branch unit, 5 is an internal data buffer, 6 is an external data memory, 7.21.
23 indicates a confluence section. 20th 4th element 5: 86 concave 7th ri 8th field

Claims (1)

[Scope of Claims] A data flow type information processing device comprising: an internal processing means for performing arithmetic processing on data according to a data flow program; an external storage means for storing data; and a first storage means for receiving data from the external storage means. a merging means, a first branching means for receiving data from the first merging means, a second merging means for receiving data from the internal processing means, and a second merging means for receiving data from the second merging means. branching means; the first merging means selectively supplies the data given from the external storage means or the data given from the outside to the first branching means; The data provided from the merging means is selectively provided to the internal processing means or the second merging means, and the second merging means receives the data provided from the first branching means or the data provided from the internal processing means. selectively to the second branching means, and the second branching means outputs the data given from the second merging means to the external storage means or the outside.