Assembly line simulation method for constructing instruction set into discrete three-dimensional matrix
Through the pipeline simulation method of constructing the instruction set into a discrete three-dimensional matrix, the problem of inefficient simulation of high-complexity instruction set pipeline scenarios in the existing technology is solved, and a more efficient and accurate simulation process is achieved.
Patent Information
- Application Number
- CN202411781459.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-05
- Publication Date
- 2025-04-18
AI Technical Summary
The prior art is difficult to effectively handle high-complexity processor instruction collection pipeline scenarios, resulting in long-term simulation processes and low efficiency.
A pipeline simulation method is adopted to construct the instruction set into a discrete three-dimensional matrix. By obtaining the function, time period and attribute information of the instructions, a discrete three-dimensional model is built, and the index table and the running quadrant model are sorted and simulated.
It realizes a more intuitive and comprehensive instruction set description, reduces development difficulty, ensures the correctness of simulation, improves simulation efficiency, and can effectively handle instruction parallel relationships, multi-cycle instruction entry time and nested relationships.
Smart Images
Figure CN119847607A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a pipeline simulation method for constructing an instruction set as a discrete three-dimensional matrix, and belongs to the technical field of instruction set simulation. Background Art
[0002] In software debugging and testing, the use of instruction set simulators is an important means to ensure software quality. The complexity of the processor instruction set pipeline will affect the difficulty of simulator development and the correctness of the simulation. Therefore, for high-complexity processor instruction set pipeline scenarios, it is very important to be able to correctly execute the instruction set during the simulation process.
[0003] The instruction set architectures of many existing high-performance processors support features such as multi-instruction parallelism, instruction functions taking effect in cycles, delay slots that can nest other instructions, and context influences. For scenarios where the above features exist, it is necessary to describe them from three dimensions: time, function, and attribute. However, the existing instruction set simulation uses a one-by-one analysis method, which is a one-dimensional description method that inserts instructions one by one along the time chain. This method requires continuous identification of the parallel relationship between instructions; it requires continuous judgment of the effective time of multi-cycle instructions; it requires continuous updating of the nesting relationship in the delay slot; and it requires continuous analysis of the influence of instruction contexts on each other. It is difficult for existing simulation methods to handle the three-dimensional characteristics of the above-mentioned highly complex instruction set architecture, which makes the simulation process time-consuming and inefficient. Therefore, a simulation method is needed to effectively handle this highly complex instruction set pipeline scenario. Summary of the invention
[0004] The technical problem solved by the present invention is: to overcome the deficiencies of the prior art, to provide a pipeline simulation method for constructing an instruction set as a discrete three-dimensional matrix, and to solve the simulation problem that the existing simulator is difficult to sort the highly complex instruction set pipeline.
[0005] The technical solution of the present invention is: a pipeline simulation method for constructing an instruction set as a discretized three-dimensional matrix, comprising:
[0006] According to the semantics of the instruction set defined by the processor, three aspects of information about the instruction are obtained, including the function executed by the instruction at a certain moment, the execution time cycle of the instruction, and the execution attributes of the instruction;
[0007] Construct corresponding discretized three-dimensional models for all instructions of the processor respectively, fill the three aspects of information of the instructions into corresponding positions of the discretized three-dimensional models, and complete the construction of the discretized three-dimensional models of each instruction;
[0008] Number all instructions of the processor one by one, and establish an index table of the corresponding relationship between the instruction number and the discretized three-dimensional matrix model;
[0009] Parse the target code to obtain the instruction content that constitutes the code segment in the target code;
[0010] The instruction number corresponding to the instruction content constituting the target code segment is used as the index key value, and the discretized three-dimensional matrix model corresponding to each instruction content in the target code segment is obtained in turn through the index table of the corresponding relationship between the instruction number and the discretized three-dimensional matrix model, and the discretized three-dimensional matrix model is preliminarily sorted according to the order in the code segment to form a serial stream;
[0011] Construct the instruction set operation quadrant model according to the three dimensions of execution function, execution time and execution attribute;
[0012] Fill the contents of the serial stream into the constructed instruction set operation quadrant model to construct a three-dimensional instruction set pipeline matrix;
[0013] Convert the three-dimensional instruction set pipeline matrix into a two-dimensional matrix in series;
[0014] The two-dimensional matrix after serial conversion is simulated and executed.
[0015] Preferably, the three dimensions in the discretized three-dimensional model are specifically:
[0016] The first dimension of the model is recorded as columns, which represent the functions executed by the instructions at a certain moment;
[0017] The second dimension of the model is recorded as rows, which represents the execution time cycle of instructions;
[0018] The third dimension of the model is denoted as layer, which represents the execution properties of instructions.
[0019] Preferably, in the discretized three-dimensional model of the instruction:
[0020] If multiple functions are executed in parallel within the same cycle of the current instruction, the different columns of the same row of the discretized three-dimensional model represent the parallel execution of the multiple functions; if the function of the current instruction takes effect in different cycles, the different rows of the same column of the discretized three-dimensional model represent the different cycles in which a certain instruction function takes effect; if the function point of the current instruction has execution attributes, the execution attribute characteristics of the current function point are recorded through the corresponding row and column positions of the second layer of the current discretized three-dimensional model.
[0021] Preferably, when establishing an index table of correspondence between instruction numbers and discretized three-dimensional matrix models:
[0022] All instruction contents of the processor are numbered one by one, and the numbers are matched with each discrete three-dimensional matrix model. The key value is used as the number to establish an instruction set number and a discrete three-dimensional matrix model index table. The discrete three-dimensional matrix model corresponding to the instruction can be obtained by matching the instruction content to the number, and then the various functions of the instruction and the execution cycle of each function can be obtained through the element values at the corresponding row and column positions in the discrete three-dimensional matrix model.
[0023] Preferably, the instruction set running quadrant model is a three-dimensional array containing three-dimensional information, capable of arranging and combining the discretized three-dimensional matrix models of multiple instructions;
[0024] The three dimensions of information include:
[0025] The first dimension is the columns of the matrix, each column represents a discrete function point of instructions running in the same execution time cycle. Different columns in the same execution time cycle represent different discrete function points executed in parallel;
[0026] The second dimension is the rows of the matrix, each row represents an execution time cycle of the running instruction;
[0027] The third dimension is the layer of the matrix, which represents the execution properties of the current code segment when it is running. There are two layers in total. The first layer is the matrix composed of the above rows and columns, that is, the specific discrete function points executed at the corresponding moment. The second layer represents the execution properties of the specific discrete function points executed at the corresponding moment, and supplements and saves the information required for runtime.
[0028] Preferably, the method for constructing a three-dimensional instruction set pipeline matrix is:
[0029] The element values at corresponding positions in the discretized three-dimensional matrix model in the model serial stream are sequentially filled into corresponding positions of the instruction set running quadrant model.
[0030] Preferably, the method of serially converting the three-dimensional instruction set pipeline matrix into a two-dimensional matrix is:
[0031] The rows of the first layer of the three-dimensional instruction set pipeline matrix are arranged in series as the first row of the two-dimensional matrix; the rows of the second layer of the three-dimensional instruction set pipeline matrix are arranged in series as the second row of the two-dimensional matrix, representing the characteristic attributes of the corresponding instruction function points.
[0032] Preferably, the two-dimensional matrix after serial conversion is simulated and executed, and the specific method is:
[0033] The simulation of the instruction discrete function points is completed one by one in the ascending order of the columns of the first row of the two-dimensional matrix. At the same time, the characteristic attributes of the instruction discrete function points corresponding to the second row of the two-dimensional matrix provide the information required for the simulation execution of the instruction discrete function points.
[0034] Compared with the prior art, the present invention has the following advantages:
[0035] (1) The present invention adopts a technical means of describing the instruction set pipeline in three dimensions, achieving a more intuitive and comprehensive expression effect, thereby reducing the difficulty of development and ensuring the correctness of the simulation.
[0036] (2) The present invention adopts the technical means of abstracting the content of the instruction set into an independent model, thereby achieving a more flexible instruction sorting effect.
[0037] (3) The present invention adopts a three-dimensional matrix to perform pipeline sorting on the instruction set module, thereby achieving the effect of correctly handling the instruction parallel relationship, multi-cycle instruction effective time, and nested relationship through sorting in the decoding stage, thereby avoiding subsequent additional maintenance work and improving efficiency.
[0038] (4) The present invention adopts the technical means of setting instruction attributes during the instruction set sorting process, thereby achieving the effect of effectively processing the characteristics required for instruction execution, including the influence of instruction contexts on each other, without the need for subsequent continuous maintenance of these influences, thereby improving efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1 A flow chart of the steps of a pipeline simulation method for constructing an instruction set as a discretized three-dimensional matrix provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0040] The technical solution of the present invention is: a pipeline simulation method for constructing an instruction set as a discretized three-dimensional matrix, comprising the following steps:
[0041] 1. Construct all instructions of the processor into independent discrete three-dimensional models. Obtain three aspects of information based on the semantics of the instruction set defined by the processor. Construct the three aspects of information into three dimensions of the model:
[0042] (1) The first dimension (column) of the model represents the function executed by the instruction at a certain moment.
[0043] (2) The second dimension (row) of the model represents the execution time cycle of the instruction.
[0044] (3) The third dimension (layer) of the model represents the execution properties of instructions.
[0045] If multiple functions are executed in parallel in the same cycle of the current instruction, the same row and different columns of the model respectively represent the parallel execution of multiple functions. If the functions of the current instruction are effective in different cycles, the different rows of the model represent the different cycles in which the instruction functions are effective. If the function point of the current instruction has characteristic attributes, such as attributes that affect the context, the attribute characteristics of the current function point are recorded through the corresponding position of the second layer of the current model.
[0046] 2. Number all the instructions of the processor one by one. According to the discretized three-dimensional matrix model of all the instructions of the processor established in step 1, an index table of the corresponding relationship between the instruction number and the discretized three-dimensional matrix model is established.
[0047] 3. The target code parsing software parses the target code to obtain the instruction content that constitutes the code segment in the target code.
[0048] 4. The instruction numbers corresponding to the instruction contents that make up the target code segment in step 3 are input into the instruction number and model index table established in step 2 as index key values one by one, thereby obtaining the discretized three-dimensional matrix model corresponding to each instruction in the target code segment. The model is preliminarily sorted according to the time sequence of the execution of each function of the instructions in the code segment to form a serial stream and save it.
[0049] 5. The instruction set operation quadrant is a container model that can arrange and combine the discrete three-dimensional matrix models of multiple instructions. Similarly, step 1 is constructed according to the three dimensions of execution function, execution time, and execution attribute.
[0050] 6. The discretized three-dimensional matrix model corresponding to the instruction content of the target code segment in step 4 fills the execution function and execution attributes into the instruction set operation quadrant model constructed in step 5 according to the execution time, thereby constructing a three-dimensional instruction set pipeline matrix. The construction method should follow: If multiple instructions are executed in parallel in the same cycle, the discretized three-dimensional matrix models corresponding to the multiple instructions in the same cycle are filled into the same row and different columns of the instruction set operation quadrant model. The increase in execution time is represented by the increase in the second dimension (row) of the instruction set operation quadrant model. The third dimension (layer) of the instruction set operation quadrant model records the execution attributes of the instruction set discretized three-dimensional matrix model, and also records the information required when the simulator is running, which is used to handle the cycle-by-cycle effectiveness of instructions, context influence, instruction nesting, etc.
[0051] 7. Convert the three-dimensional instruction set pipeline matrix in step 6 into a two-dimensional matrix. The same row of the first layer of the three-dimensional instruction set pipeline matrix is arranged in series according to the increasing columns and converted into the first row of the two-dimensional matrix, indicating that the functions of the parallel instructions at the same time are executed in sequence. The same row of the second layer of the three-dimensional instruction set pipeline matrix is arranged in series according to the increasing columns and converted into the second row of the two-dimensional matrix, indicating the characteristic attributes of the corresponding instruction function point. In the first layer of the three-dimensional instruction set pipeline matrix, the column tail of the current row and the column head of the next row are arranged in series, and the column tail of the first row of the two-dimensional matrix is filled in, and the arrangement continues backward, indicating that the instruction functions at different times are arranged in series with the execution time increasing. Similarly, in the second layer of the three-dimensional instruction set pipeline matrix, the column tail of the current row and the column head of the next row are arranged in series, and converted into the second row of the two-dimensional matrix, and the arrangement continues backward, indicating that the characteristic attributes of the corresponding instruction functions are arranged in series, and the execution time of the instruction is recorded in the characteristic attributes.
[0052] The serialized instruction set pipeline is a two-dimensional sequence that performs single-point function execution in chronological order. The execution attributes are used to assist in completing the single-point function simulation, thereby completing the simulation of the instruction set pipeline.
[0053] 8. Simulate the two-dimensional matrix after serial conversion in step 7. Simulate the discrete function points of the instruction one by one according to the columns of the first row of the two-dimensional matrix. At the same time, the characteristic attributes of the discrete function points of the instruction corresponding to the second row of the two-dimensional matrix provide the information required for the simulation execution of the discrete function points of the instruction.
[0054] Example:
[0055] STEP.1: Discrete instruction set 3D model construction
[0056] Step 101: Obtain information of three dimensions of all instruction sets according to the instruction set semantics defined by the processor. The information of the three dimensions includes:
[0057] (1) The first dimension (column) represents the discrete function points executed by the instruction at a certain moment.
[0058] (2) The second dimension (row) represents the execution time cycle of the instruction.
[0059] (3) The third dimension (layer) represents the execution properties of instructions.
[0060] Step 102: Establish all instructions of the processor as independent discretized three-dimensional models. The discretized three-dimensional model of the instruction refers to the three-dimensional array Ins[z][y][x]. The three dimensions of the three-dimensional array Ins[z][y][x] correspond to the three-dimensional information in step 101. The specific corresponding relationship is:
[0061] (1) x corresponds to the first dimension (column), which represents the discrete function point executed by the instruction at a certain moment. The value of x is the maximum number of functions executed in parallel in the same cycle.
[0062] (2) y corresponds to the second dimension (row), which represents the execution time of the instruction. The value of y is the number of cycles required to complete the execution of all functions of the instruction.
[0063] (3) z corresponds to the third dimension (layer), which indicates the execution attribute of the instruction, and the value of z is 2. When z = 0, it is the first layer, which indicates the relationship attribute between the execution function and the execution time, that is, the specific discrete function point executed at the corresponding moment. When z = 1, it is the second layer, which indicates the characteristic attribute of the function point at the corresponding moment.
[0064] Step 103: Fill the three-dimensional information of the corresponding instruction in step 101 into the discretized three-dimensional model of the corresponding instruction in step 102 to complete the construction of the discretized three-dimensional model of all instructions. Specifically:
[0065] (1) Fill the discrete function points executed at the corresponding moment of the current instruction into the array elements of the column row (xy) subscript corresponding to the first layer (z=0).
[0066] (2) Fill the characteristic attributes of the current instruction at the corresponding moment of execution into the array element of the column row (xy) subscript corresponding to the second layer (z=1).
[0067] (3) If the current instruction executes multiple functions in parallel in the same cycle, then different columns (x) in the same row (y) of the array are filled with multiple discrete function points executed in parallel in the same cycle.
[0068] (4) If the function of the current instruction takes effect in different cycles, the corresponding effective discrete function points are filled in the different rows (y) corresponding to the different effective cycles of the array.
[0069] (5) If the discrete function point executed by the current instruction at a certain moment is empty, the array element of the corresponding column and row (xy) subscript is empty.
[0070] Step 104: Number all instructions of the processor one by one. The instruction numbers are matched with the discretized three-dimensional matrix models of all instructions established in step 103, so as to establish an instruction number and model index table. The key value of the index table is the instruction number, and the corresponding discretized three-dimensional matrix model can be quickly obtained through the instruction number.
[0071] Step 105: The target code analysis software analyzes the target code to obtain the instruction set that constitutes the code segment in the target code. The instruction set of the code segment is arranged in the order of the code segments to form a serial stream of instructions.
[0072] Step 106: The numbers corresponding to the instructions in the code segment in step 105 are input as index key values one by one into the instruction number and model index table established in step 104, so as to obtain the discretized three-dimensional matrix model corresponding to each instruction in the target code segment. These models are preliminarily sorted according to the order of the instruction serial stream corresponding to the code segment, and the sorting result is saved.
[0073] STEP.2: Construction of three-dimensional instruction set pipeline matrix
[0074] Step 201: The instruction set running quadrant is a container model that can arrange and combine the discrete three-dimensional matrix models of multiple instructions, and is used to represent a code block composed of multiple instructions. Similarly to steps 101 and 102, the instruction set running quadrant is a three-dimensional array Matrix[Z][Y][X] containing three-dimensional information. The three-dimensional information includes:
[0075] (1) X corresponds to the first dimension (columns). Different columns in the same row represent discrete function points of instructions running in parallel. Function points of multiple instructions can be filled in different columns of the same row, indicating that multiple function points are executed in parallel at the same time. The number of columns is the maximum number of function points allowed to be executed in parallel in the same cycle by the processor.
[0076] (2) Y corresponds to the second dimension (rows). Different rows represent the execution time of each functional point of the instruction. The number of rows is the number of cycles required to complete the execution of all functions of the current code block.
[0077] (3) Z corresponds to the third dimension (layer), which represents the execution properties of the current code block when it is running. In the present invention, the third dimension is 2 layers. The first layer (Z=0) represents the relationship between the execution function and the execution time, that is, the specific discrete function points executed at the corresponding moment. The second layer (Z=1) represents the characteristic attributes of the function points at the corresponding moment, and can also save the information required for the runtime.
[0078] Step 202: Fill the discretized three-dimensional matrix model serial stream corresponding to the target code segment saved in step 106 into the instruction set running quadrant model constructed in step 201, thereby constructing a three-dimensional instruction set pipeline matrix. The specific construction method is:
[0079] (1) The discretized three-dimensional matrix model serial stream corresponding to the target code segment saved in step 106 is used as input. The discretized three-dimensional matrix model Ins[z][y][x] in the model serial stream is sequentially filled into the instruction set running quadrant model Matrix[Z][Y][X] array constructed in step 201. The filling method refers to filling the element values of the corresponding positions in Ins[z][y][x] into the corresponding positions of the Matrix array in ascending order of columns (X) and rows (Y) of the Matrix[Z][Y][X] array.
[0080] (2) If there is no parallel execution relationship between the current discretized three-dimensional matrix model and the previous model, then the current discretized three-dimensional matrix model should be filled into the next row of the instruction set running quadrant model to indicate the increase in execution time.
[0081] (3) If the currently filled-in discretized three-dimensional matrix model has a parallel execution relationship with the previous model, the current discretized three-dimensional matrix model is filled into the current row and different columns of the instruction set running quadrant model, indicating that the function points are executed in parallel and the execution time does not increase.
[0082] (4) If the currently filled discretized three-dimensional matrix model has a function that takes effect in a periodic manner, the first row of the current discretized three-dimensional matrix model is used as the starting row, and the effective function points are filled in the row corresponding to the time point at which the periodic effect is taken effect.
[0083] (5) In the process of building a three-dimensional instruction set pipeline matrix, due to the characteristics of the instructions, complex situations such as context influence and instruction nesting may occur. In order to deal with these problems, in the second layer of the instruction set running quadrant model, it is necessary to record the status in the corresponding rows and columns. These records will be used in the subsequent simulation execution process to help deal with these complex instruction characteristics.
[0084] STEP.3: Matrix serialization conversion and simulation execution
[0085] Step 301: Convert the three-dimensional instruction set pipeline matrix Matrix[Z][Y][X] constructed in step 202 into a two-dimensional matrix Array[J][K]. The specific method is:
[0086] (1) The functions of the instructions in the same row (Y) of the first layer (Z=0) of the three-dimensional instruction set pipeline matrix are arranged in series according to the increasing number of columns (X). When converted to the first row (J=0) of the two-dimensional matrix, they are arranged in series according to the increasing number of columns (K), indicating that the functions of the instructions executed in parallel at the same time are executed sequentially.
[0087] (2) The characteristic attributes of the instructions in the same row (Y) of the second layer (Z=1) of the three-dimensional instruction set pipeline matrix are arranged in series in ascending order of columns (X), and are converted into the second row (J=1) of the two-dimensional matrix and arranged in series in ascending order of columns (K), indicating the characteristic attributes of the corresponding instruction functions.
[0088] (3) The increasing rows (Y) of the first layer (Z=0) of the three-dimensional instruction set pipeline matrix represent the increasing instruction execution time. The tail of the column (X) of the current row (Y) is serially arranged with the head of the column (X) of the next row (Y+1), and the tail of the column (K) of the first row (J=0) of the two-dimensional matrix is continuously arranged backward, indicating the serial arrangement of instructions at different times with increasing execution time.
[0089] (4) The second layer (Z=1) of the three-dimensional instruction set pipeline matrix performs a serial arrangement of the tail of the column (X) of the current row (Y) and the head of the column (X) of the next row (Y+1), and fills in the tail of the column (K) of the second row (J=1) of the two-dimensional matrix and continues to arrange backwards, indicating that the characteristic attributes of the corresponding instruction functions are serially arranged and converted, and the execution time of the instruction is recorded in the characteristic attributes.
[0090] Step 302: Simulate and execute the two-dimensional matrix Array[J][K] after serial conversion in step 301. Simulate the discrete function points of the instruction one by one according to the columns (K) of the first row (J=0) of the two-dimensional matrix. At the same time, the characteristic attributes of the discrete function points of the instruction corresponding to the second row (J=1) of the two-dimensional matrix provide the information required for the simulation execution of the discrete function points of the instruction. The information includes: simulation time, cycle marking, context marking, instruction nesting relationship, etc.
[0091] The contents not described in detail in the specification of the present invention belong to the prior art known to the professional and technical personnel in this field.
Claims
1. A pipeline simulation method for constructing an instruction set as a discretized three-dimensional matrix, characterized in that include: According to the semantics of the instruction set defined by the processor, three aspects of the instruction are obtained, including the function executed by the instruction at a certain moment, the execution time cycle of the instruction, and the execution properties of the instruction; Construct corresponding discretized three-dimensional models for all instructions of the processor respectively, fill the three aspects of information of the instructions into corresponding positions of the discretized three-dimensional models, and complete the construction of the discretized three-dimensional models of each instruction; Number all instructions of the processor one by one, and establish an index table of the corresponding relationship between the instruction number and the discretized three-dimensional matrix model; Parse the target code to obtain the instruction content that constitutes the code segment in the target code; The instruction number corresponding to the instruction content constituting the target code segment is used as the index key value, and the discretized three-dimensional matrix model corresponding to each instruction content in the target code segment is obtained in turn through the index table of the corresponding relationship between the instruction number and the discretized three-dimensional matrix model, and the discretized three-dimensional matrix model is preliminarily sorted according to the order in the code segment to form a serial stream; Construct the instruction set operation quadrant model according to the three dimensions of execution function, execution time and execution attribute; Fill the contents of the serial stream into the constructed instruction set operation quadrant model to construct a three-dimensional instruction set pipeline matrix; Convert the three-dimensional instruction set pipeline matrix into a two-dimensional matrix in series; The two-dimensional matrix after serial conversion is simulated and executed.
2. The pipeline simulation method of constructing an instruction set as a discretized three-dimensional matrix according to claim 1, characterized in that: The three dimensions in the discretized three-dimensional model are: The first dimension of the model is recorded as columns, which represent the functions performed by the instructions at a certain moment; The second dimension of the model is recorded as rows, which represents the execution time cycle of instructions; The third dimension of the model is denoted as layer, which represents the execution properties of instructions.
3. The pipeline simulation method of constructing an instruction set as a discretized three-dimensional matrix according to claim 2, characterized in that: In the discretized 3D model of instructions: If multiple functions are executed in parallel within the same cycle of the current instruction, the different columns of the same row of the discretized three-dimensional model represent the parallel execution of the multiple functions; if the function of the current instruction takes effect in different cycles, the different rows of the same column of the discretized three-dimensional model represent the different cycles in which a certain instruction function takes effect; if the function point of the current instruction has execution attributes, the execution attribute characteristics of the current function point are recorded through the corresponding row and column positions of the second layer of the current discretized three-dimensional model.
4. The pipeline simulation method of constructing an instruction set as a discretized three-dimensional matrix according to claim 1, characterized in that: When establishing an index table of the correspondence between instruction numbers and discretized three-dimensional matrix models: All instruction contents of the processor are numbered one by one, and the numbers are matched with each discrete three-dimensional matrix model. The key value is used as the number to establish an instruction set number and a discrete three-dimensional matrix model index table. The discrete three-dimensional matrix model corresponding to the instruction can be obtained by matching the instruction content to the number, and then the various functions of the instruction and the execution cycle of each function can be obtained through the element values at the corresponding row and column positions in the discrete three-dimensional matrix model.
5. The pipeline simulation method of constructing an instruction set as a discretized three-dimensional matrix according to claim 1, characterized in that: The instruction set operation quadrant model is a three-dimensional array containing three-dimensional information, which can arrange and combine the discrete three-dimensional matrix models of multiple instructions; The three dimensions of information include: The first dimension is the columns of the matrix, each column represents a discrete function point of instructions running in the same execution time cycle, and different columns in the same execution time cycle represent different discrete function points executed in parallel; The second dimension is the rows of the matrix, each row represents an execution time cycle of the running instruction; The third dimension is the layer of the matrix, which represents the execution properties of the current code segment when it is running. There are two layers in total. The first layer is the matrix composed of the above rows and columns, that is, the specific discrete function points executed at the corresponding moment. The second layer represents the execution properties of the specific discrete function points executed at the corresponding moment, and supplements and saves the information required for runtime.
6. The pipeline simulation method of constructing an instruction set as a discretized three-dimensional matrix according to claim 1, characterized in that: The method to construct a three-dimensional instruction set pipeline matrix is: The element values at corresponding positions in the discretized three-dimensional matrix model in the model serial stream are sequentially filled into corresponding positions of the instruction set running quadrant model.
7. The pipeline simulation method of constructing an instruction set as a discretized three-dimensional matrix according to claim 1, characterized in that: The method of serially converting a three-dimensional instruction set pipeline matrix into a two-dimensional matrix is: The rows of the first layer of the three-dimensional instruction set pipeline matrix are arranged in series as the first row of the two-dimensional matrix; the rows of the second layer of the three-dimensional instruction set pipeline matrix are arranged in series as the second row of the two-dimensional matrix, representing the characteristic attributes of the corresponding instruction function points.
8. The pipeline simulation method of constructing an instruction set as a discretized three-dimensional matrix according to claim 1, characterized in that: The two-dimensional matrix after serial conversion is simulated and executed. The specific method is: The simulation of the instruction discrete function points is completed one by one in the ascending order of the columns of the first row of the two-dimensional matrix. At the same time, the characteristic attributes of the instruction discrete function points corresponding to the second row of the two-dimensional matrix provide the information required for the simulation execution of the instruction discrete function points.