JPH02123432A - Hardware expression/description system and its simulation system - Google Patents

Abstract

PURPOSE:To more efficiently analyze an error by expressing hardware in accordance with the order of the use of an operated object by a operating object, the start of another operating object from one operating object, the transition of operation, and the end of operation of the operating object. CONSTITUTION:In the case of starting another operating object (b) from a certain operating object (a), the process ID (process ID:1) of the starting operating object (a) is added to the starting history information of the object (a) to use the added result as the starting history information of the started operating object (b) and a new number (process ID:2) is allocated as the process ID of the object (b). In case of operation transition from the object (b) to an operating object (c), the starting history information (process ID:X) of the transmitting object (a) is used as the starting history information of the transmitted object (c) and the process ID (process ID:1) of the object (a) is used as the process ID of the object (c). Thus, complex simultaneous operation can be controlled and sorted by the process ID directly connected to each function.

Description

[Detailed Description of the Invention] "Industrial Application Field" This invention relates to the design and development of hardware that performs highly parallel operations.
It concerns hardware representation/description methods with high descriptive properties suitable for description and verification, and simulation methods with high analytical performance.

``Conventional technology'' Conventional hardware description methods have a clear distinction between action subjects and action objects, and also describe static relationships such as connections between components included in the description target, 11
Mixing of descriptions of dynamic relationships such as 10 ordination is allowed [2], so 5. J, l At the time of nosion, it is possible to specify in what situation and which action a certain action object was used by two objects.
In the design of sequential motion, it has been difficult to analyze the causes of design errors such as unfair competition between a certain motion object (f2).

1] The purpose of this invention is the conventional hardware description method, L
This solves the low verifiability of the Simicollation method for highly parallel operating hardware, and provides a hardware description method that can provide the Simicollation method with sufficient information to perform various verifications. The purpose of this research is to provide a simulation method that allows for accurate verification.

``Solving the Problem 1 - Means of Solving - 1 This invention solves the problem by defining the hardware object to be described.1)
The function, structure, and operation of the hardware are divided into multiple motion bodies that control the motion, and multiple motion objects that are the objects of control, and the functions, structures, and operations of the hardware are explained from the motion body 1 to the motion 8: body use and motion objects. Activation of other moving bodies, transition from the moving body to other moving bodies, and the order in which the movements of the moving bodies end (Expression 17 from 2) This is used in the simulation. This is the most important feature, and it differs greatly from the conventional technology in that it eliminates the static 4[description].

[゛Example-1 The first part 1 is a diagram illustrating an example of the content related to the claims (]) of this invention.
A certain Bird Core description format based on 2 (2 shows the entire hardware description example described above.
is the action object definition part, 2 is the action object definition part, 3 is the description format of the order of use of action objects within the action object definition part, and 4 is the activation format of another action object from the action object between action objects. , 5 is the transition form of the action from the action upper body to another action upper body, and 6 is the termination form of the action upper body action. Activation, action - The transition of an action from one action object to another action object, and the end of the action of an action object are called generate, relay, and finish, respectively. o uSeA means that action object A uses action object A. However, generate b represents that the action body a activates the action body A, and relay c represents the transition of the action from the action body a to the action body C.

As an object of operation, we usually choose something such as a register or an adder that directly corresponds to a simple function (a register is a function that holds one value, an adder is a function that holds one value), and an arithmetic function).

Operation 4 The three bodies correspond to the control circuit. In a pipeline computer, each stage has its own control circuit.

The function of each action object is expressed by describing the order of use of action objects (expressed from the second point).

- On the other hand, parallel control by multiple operating bodies C is the activation of other operating subjects from the operating bodies between the operating subjects.
te ), the transition of movement from one working body to another working body (
relay ), the end of the action of the action subject (finish
)−2.

In generate, the operation of the activation source behavioral body continues (1'), and the activation source behavioral body shares part of the processing of the activation source behavioral body. In relay, the motion of the transition destination behavioral body ends, and the remaining processing is performed by the transition destination behavioral body. Finish indicates that the execution of the task has been completed.

For example, generate and fini are used for leave-behind control or parallel control that keeps the parallelism of hardware the same.
sh, re lay and fini for pipeline control
Apply sh.

Next, Figure 2 shows an example of the content related to claim (2) c1 of the present invention, and it shows stains, -4 ratios (work (2) numbered )
7 shows the process ID and 8 is the startup history information. Only the operating body that is in operation retains the process ID and activation history information.

The number of startup source process IDs that can be retained in the startup history information (there is no limit to 2).The process ID can be saved, for example, every time a new process ID is required from the simulation start time (
1, 2, 3 in order.・Granted.

When starting another operating entity from operating entity a (gen
erase ) adds the process ID (process ID = 1) of the startup entity to the startup history information of the startup entity a to obtain the startup history information of the destination entity a, and also creates the process of the destination entity a. A new number (process ID: 2) is assigned as the ID.

In the case of a transition of an action from an action entity a to another action entity C (
relay ) is the activation history information (
Let process ID:

At the end of the operation of the operator, the process ID and activation history information are deleted.

With this mechanism, the process ID does not change in pipeline control, where one process is executed while transitioning between multiple operating entities, and in left-behind control and parallel control, a new process ID is assigned and the Whether a part of the processing is shared is left in the process ID in the activation history information.

As a result, the cooperative work of many operating entities (control units): thus, the complicated simultaneous operations of the hardware,
They can be managed and classified using process IDs that are directly linked to functions.

``Effects of the Invention'' As explained above (2. According to this invention, for every designed and written operation, the action entity and process ID that cause it correspond. When a design error is detected, such as a value being written or a conflict in the use of the same combinational circuit, the simulator can immediately output the operating entity and process ID that caused the error, making it easier to analyze the error. This has the advantage that it can be done more efficiently.

[Brief explanation of drawings]

Figure 1 is a diagram showing an entire hardware description example written using a certain hardware description format (2) according to the present invention, and Figure 2 is a diagram showing a method of writing process ID and startup history information during simulation. 1: Action object definition section, 2: Action object definition section, 3: Description format of action object usage order within the action object definition section, 4: Invocation format from one action object to another action object among action objects. 5:
Transition format of an action from an action entity to another action entity, 6: Termination format of an action of an action entity, 7: Process ID, 8: Activation history information.

Claims (2)

[Claims] (1) Divide the hardware object to be described into a plurality of action subjects that control control and a plurality of action objects that are the objects of control, and explain the function, structure, and operation of the hardware as described by the action objects by the action living body. use, memory of an action subject to another action subject, transition of action from an action organism to another action organism,
A hardware representation/description method characterized by representing the order of the end of the action of the action subject. (2) A mechanism is provided on the simulator to store the process ID and the history of which process ID has been started for each operating entity, and when an operating entity starts another operating organism, the starting operating entity Adding the process ID of the starting action entity to the startup history information of the startup action entity to create the startup history information of the startup action entity, and assigning a new number as the process ID of the startup action entity, and instructing the action from the action entity to another action entity. In the case of a transition, the startup history information of the transition source action entity is set as the startup history information of the transition destination action entity, and the process ID of the transition source action entity is set as the process ID of the transition destination action entity.
By doing so, any use of an action object identifies the action object that directly executed it, the process ID that caused the action by operating that action object, and the action object with a specific process ID. A simulation method characterized in that it is possible to specify the use of all action objects carried out by all action subjects whose activation history information includes living organisms and their process IDs.