Instruction issuance method and device for out-of-order processor
By adopting speculative wake-up and replay mechanisms in out-of-order processors, the problem of write-first and then read-related instructions execution delay is solved, and more efficient instruction execution and pipeline throughput is achieved.
Patent Information
- Application Number
- CN202411238627.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-04
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-09-04
AI Technical Summary
In the prior art, there is a three-cycle execution delay between two instructions with write-first and then read correlation, resulting in poor execution efficiency.
When the write instruction is selected to be transmitted from the transmit queue, the destination register address of the write instruction is broadcast, the source register address matching the destination register address of the write instruction is determined, and the read instruction corresponding to the source register address is awakened, and the counter is determined Read instructions wait for the period of the source operand and wait or return to the transmit queue within the preset period, using speculative wake-up and replay mechanisms to improve execution efficiency.
Through the speculative wake-up and replay mechanism, two instructions with write first and read correlation can be executed back-to-back, saving three cycles, greatly improving the execution efficiency of out-of-order processors, and preventing pipelines when cached data is not hit. Stuck, improve the throughput of the pipeline.
Smart Images

Figure CN118760475B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of central processing units, and in particular, to an instruction transmission method, apparatus, computer equipment and storage medium for an out-of-order processor. Background Art
[0002] Emission is the process of selecting instructions that meet certain conditions from the IQ (Issue Queue) and sending them to the FU (Function Unit). Among them, the IQ is the most critical component in the overall emission pipeline.
[0003] For sequential transmission, instructions are written into the transmission queue according to the original order in the program. At this time, the transmission queue is equivalent to a FIFO (First Input First Output). Only when the source operands of the oldest instruction are ready, the instruction will be transmitted. If this instruction is not transmitted, all subsequent instructions should wait and cannot be transmitted.
[0004] For out-of-order transmission, the transmission queue will select instructions whose source operands are ready according to certain rules and send them to the FU for execution. This process is called transmission. After the instructions arrive in the transmission queue, they will not flow in the processor in the order specified in the program. Only when the operands of an instruction in the transmission queue are ready and the transmission conditions are met can they be sent to the corresponding FU for execution. Therefore, for an out-of-order processor, transmission is the dividing line from sequential execution to out-of-order execution. All instructions before the transmission stage flow sequentially, and after the transmission stage, they are all executed out of order until the submission stage of the pipeline uses reordering to pull the instructions back to the original order specified in the program.
[0005] There are many ways to implement the launch phase, and the central component of this phase is the launch queue. Its implementation method directly determines the implementation method of the entire launch phase, centralized or distributed, compressed or non-compressed, data capture or non-data capture.
[0006] For instructions that have been dispatched to IQ, the following conditions must be met before they can be executed by EU (Execution Unit):
[0007] The source operands of the instruction are ready or can be obtained from the bypass network, and the instruction is selected by the selection logic of the IQ.
[0008] Only instructions that meet the above conditions can be ejected from the IQ and executed in the EU. For instructions that have been dispatched to the IQ but whose source operands are not yet ready, the bypass network needs to be monitored until the physical register ID (Index) matches at a certain point in time, indicating that the source operand of the instruction can be obtained from the bypass network. At this time, the process of the instruction jumping from the unprepared state to the ready state is called wake-up. The wake-up process of the existing technical solution can be used Figure 1 express.
[0009] The existing technical solution is that the instruction source operands dispatched to the IQ must be ready before they can be selected from the IQ to the EU for execution. Figure 1 As can be seen, for two RAW (Read After Write) instructions, the current IQ dispatch mechanism cannot complete these two instructions back-to-back. There is an execution delay of three cycles between them, so the execution efficiency is obviously not the best. Summary of the invention
[0010] Embodiments of the present application provide an instruction transmission method, apparatus, computer device, and storage medium for an out-of-order processor.
[0011] According to a first aspect of an embodiment of the present application, there is provided an instruction issuing method for an out-of-order processor, comprising:
[0012] For two instructions with a write-before-read dependency, when a write instruction is selected to be emitted from the emission queue, the destination register address of the write instruction is broadcast;
[0013] Determine a source register address that matches the destination register address of the write instruction, and wake up a read instruction corresponding to the source register address, wherein the read instruction is not an instruction returned from the operand cache stage;
[0014] After the read instruction enters the operand cache stage, a cycle of the read instruction waiting for a source operand is determined by a counter, wherein the source operand is an operation result of the write instruction;
[0015] When the read instruction waits for the source operand within a preset cycle, the source operand of the read instruction is sent to the execution unit to execute the read instruction;
[0016] When the read instruction does not wait for the source operand within a preset cycle, it is determined whether to return the read instruction to the issue queue or continue to wait for the source operand according to the type of the issue queue where the read instruction is located.
[0017] In an optional embodiment of the present application, the transmit queue includes a memory access instruction transmit queue and a non-memory access instruction transmit queue, and when the read instruction does not wait for the source operand within a preset cycle, determining whether to return the read instruction to the transmit queue or continue to wait for the source operand according to the type of the transmit queue where the read instruction is located includes:
[0018] In the case where the read instruction does not wait for the source operand within the preset cycle, if the emission queue where the read instruction is located is a non-memory access instruction emission queue, the read instruction is returned to the emission queue, and the source operand of the read instruction in the emission queue is prepared as a condition for the read instruction to be re-awakened;
[0019] In the case that the read instruction does not wait for the source operand within the preset cycle, if the issue queue where the read instruction is located is the memory access instruction issue queue, it continues to wait for the source operand.
[0020] In an optional embodiment of the present application, after the read instruction is returned to the transmit queue, the source operand of the read instruction is obtained by reading from a physical register file.
[0021] In an optional embodiment of the present application, the write instruction is selected to be transmitted from the transmission queue, including:
[0022] The oldest instruction among the instructions with ready source operands is selected from the age matrix maintained by the issue queue and issued.
[0023] In an optional embodiment of the present application, the operation result of the write instruction is obtained by the following steps:
[0024] The source operand of the write instruction is read from the physical register stack according to the destination register address, and the write instruction is executed in the execution unit according to the source operand of the write instruction to obtain the operation result of the write instruction.
[0025] In an optional embodiment of the present application, the read instruction waits for the source operand within a preset cycle, including:
[0026] In the case where the write instruction and the read instruction belong to the same transmission queue, the read instruction waits within a preset period for the operation result obtained by the execution module in the same network executing the write instruction;
[0027] In the case that the write instruction and the read instruction do not belong to the same transmission queue, the read instruction waits for the operation result obtained by the execution module in the bypass network executing the write instruction within a preset cycle.
[0028] In an optional embodiment of the present application, the number of the transmission queues is more than one, and the destination register address of the broadcast write instruction includes:
[0029] The destination register address of the write instruction is broadcasted to a multiplex network including all the transmit queues, and instructions having a dependency relationship with the write instruction are searched from the instructions of all the transmit queues.
[0030] According to a second aspect of an embodiment of the present application, there is provided an instruction issuing device for an out-of-order processor, comprising:
[0031] A broadcast module, for broadcasting a destination register address of a write instruction when the write instruction is selected to be transmitted from the transmission queue for two instructions with a write-before-read dependency;
[0032] A wake-up module, used to determine a source register address matching a destination register address of a write instruction, and wake up a read instruction corresponding to the source register address, wherein the read instruction is not an instruction returned from an operand cache stage;
[0033] A first determination module is used to determine, through a counter, a cycle of the read instruction waiting for a source operand after the read instruction enters the operand cache stage, wherein the source operand is an operation result of the write instruction;
[0034] An execution module, configured to send the source operand of the read instruction to the execution unit to execute the read instruction when the read instruction waits for the source operand within a preset cycle;
[0035] The second determination module is used to determine whether to return the read instruction to the transmit queue or continue to wait for the source operand according to the type of the transmit queue where the read instruction is located when the read instruction does not wait for the source operand within a preset cycle.
[0036] According to a third aspect of an embodiment of the present application, a computer device is provided, comprising: a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of any of the above instruction issuance methods for an out-of-order processor are implemented.
[0037] A fourth aspect of an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of the instruction issuance method for an out-of-order processor as described above are implemented. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:
[0039] Figure 1 A flow chart of an IQ dispatch mechanism for two instructions with write-before-read dependency in the prior art;
[0040] Figure 2 A flowchart of an instruction issuing method for an out-of-order processor provided by an embodiment of the present application;
[0041] Figure 3 A flowchart of speculative wakeup provided for one embodiment of the present application;
[0042] Figure 4 A flowchart of a replay mechanism provided by an embodiment of the present application;
[0043] Figure 5 A schematic diagram of a memory access instruction issuance queue and a non-memory access instruction issuance queue provided for one embodiment of the present application;
[0044] Figure 6 A schematic diagram of the structure of an instruction issuing device for an out-of-order processor provided by one embodiment of the present application;
[0045] Figure 7 A schematic diagram of the computer device structure provided for one embodiment of the present application. DETAILED DESCRIPTION
[0046] In the process of implementing the present application, the inventors found that currently, for two instructions with a write-before-read dependency, there is an execution delay of three cycles between them, which results in poor execution efficiency.
[0047] In response to the above problems, an embodiment of the present application provides an instruction issuance method for an out-of-order processor, for two instructions with a write-before-read dependency, when the write instruction is selected to be emitted from the emission queue, the destination register address of the write instruction is broadcast; a source register address matching the destination register address of the write instruction is determined, and a read instruction corresponding to the source register address is awakened, wherein the read instruction is not an instruction returned from the operand cache stage; after the read instruction enters the operand cache stage, a cycle for the read instruction to wait for a source operand is determined by a counter, wherein the source operand is an operation result of the write instruction; when the read instruction waits for the source operand within a preset cycle, the source operand of the read instruction is sent to an execution unit to execute the read instruction; when the read instruction does not wait for the source operand within a preset cycle, it is determined whether the read instruction is returned to the emission queue or continues to wait for the source operand according to the type of the emission queue where the read instruction is located, and the problem of low execution efficiency of two instructions with a write-before-read dependency during out-of-order dispatch can be solved through a speculative wake-up and replay mechanism.
[0048] The solutions in the embodiments of the present application can be implemented using various computer languages, for example, the object-oriented programming language Java and the interpreted scripting language JavaScript.
[0049] In order to make the technical solutions and advantages in the embodiments of the present application more clearly understood, the exemplary embodiments of the present application are further described in detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than an exhaustive list of all the embodiments. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.
[0050] See also Figure 2 The instruction issuing method for an out-of-order processor provided in the embodiment of the present application includes the following steps S1 to S5:
[0051] Step S1, for two instructions with a write-before-read dependency, when a write instruction is selected to be transmitted from a transmission queue, a destination register address of the write instruction is broadcast.
[0052] In an optional embodiment of the present application, the write instruction is selected to be transmitted from the transmission queue, including:
[0053] The oldest instruction among the instructions with ready source operands is selected from the age matrix maintained by the issue queue and issued.
[0054] In an optional embodiment of the present application, the number of the transmission queues is more than one, and the destination register address of the broadcast write instruction includes:
[0055] The destination register address of the write instruction is broadcasted to a multiplex network including all the transmit queues, and instructions having a dependency relationship with the write instruction are searched from the instructions of all the transmit queues.
[0056] Step S2, determining a source register address that matches the destination register address of the write instruction, and waking up a read instruction corresponding to the source register address, wherein the read instruction is not an instruction returned from the operand cache stage.
[0057] See also Figure 3 For two instructions A and B with a write-before-read dependency, when the dependent instruction A is selected to pop up the IQ, the physical register ID of the destination register of the current instruction A is broadcast, and then matched with the physical register ID of the source register of the dependent instruction B, and instruction B is awakened in the IQ. This process is a speculative awakening process. In the stage of reading the physical register stack, instruction B does not obtain operands from the physical register stack, but obtains the operation result of instruction A from the bypass network as the source operand, and then executes in the EU. Compared with the prior art, the speculative awakening adopted by this application allows two instructions with a write-before-read dependency to be executed back-to-back, which can save three cycles and greatly improve the execution efficiency of the out-of-order processor.
[0058] exist Figure 3In the example, IQBUF is the transmit queue buffer and OPBUF is the operand buffer.
[0059] Step S3, after the read instruction enters the operand cache stage, the cycle of the read instruction waiting for the source operand is determined by a counter, wherein the source operand is the operation result of the write instruction.
[0060] In an optional embodiment of the present application, the operation result of the write instruction is obtained by the following steps:
[0061] The source operand of the write instruction is read from the physical register stack according to the destination register address, and the write instruction is executed in the execution unit according to the source operand of the write instruction to obtain the operation result of the write instruction.
[0062] See also Figure 4 ,Speculative wakeup can only be guaranteed if instruction A can be executed in a timely and correct manner, such as Figure 4 As shown in the figure, if instruction A is a load instruction, the operation result of instruction A can be obtained through the bypass network in the DCACHE E0 stage, and instruction B can be awakened in the DCACHE E2 stage. Instruction B can obtain the operation result of instruction A through the bypass network in the physical register stack reading stage, which can save three cycles and improve the execution efficiency of instruction B. However, in the case of a DCACHE (cache data) hit, that is, the operation result of instruction A can be obtained in time in the bypass network, and in the case of a DCACHE miss, the operation result of instruction A cannot be obtained in time in the bypass network. At this time, instruction B is stuck in the pipeline in the OPBUF (operand cache) stage because the source operand cannot be obtained. Until the operation result of instruction A can be obtained, instruction B can be executed in the EU and the pipeline can be restarted. In this case, waiting for instruction A to return the operation result in the case of a DCACHE miss may take dozens or even hundreds of cycles, which seriously affects the execution efficiency of the pipeline. To address this problem, a replay mechanism is proposed. That is, when instruction B reaches the OPBUF stage, OPBUF starts a counter to calculate the cycle of instruction B waiting for the source operand. When the counter reaches the preset cycle value, but instruction B has not yet obtained the source operand, instruction B returns to IQ to wait for reawakening. IQ will record such instructions that need to be reawakened, that is, it cannot be awakened speculatively, and can only be awakened truly when all source operands are ready. In this way, during the speculative awakening process, for situations such as DCACHE misses, the pipeline can execute instructions with other source operands ready in IQ through the replay mechanism, instead of being stuck and affecting the pipeline throughput.
[0063] Step S4, when the read instruction waits for the source operand within a preset cycle, the source operand of the read instruction is sent to the execution unit to execute the read instruction.
[0064] In an optional embodiment of the present application, the read instruction waits for the source operand within a preset cycle, including:
[0065] In the case where the write instruction and the read instruction belong to the same transmission queue, the read instruction waits within a preset period for the operation result obtained by the execution module in the same network executing the write instruction;
[0066] In the case that the write instruction and the read instruction do not belong to the same transmission queue, the read instruction waits for the operation result obtained by the execution module in the bypass network executing the write instruction within a preset cycle.
[0067] Step S5, when the read instruction does not wait for the source operand within the preset cycle, determine whether to return the read instruction to the transmit queue or continue to wait for the source operand according to the type of the transmit queue where the read instruction is located.
[0068] In an optional embodiment of the present application, the transmit queue includes a memory access instruction transmit queue and a non-memory access instruction transmit queue, and when the read instruction does not wait for the source operand within a preset cycle, determining whether to return the read instruction to the transmit queue or continue to wait for the source operand according to the type of the transmit queue where the read instruction is located includes:
[0069] In the case where the read instruction does not wait for the source operand within the preset cycle, if the emission queue where the read instruction is located is a non-memory access instruction emission queue, the read instruction is returned to the emission queue, and the source operand of the read instruction in the emission queue is prepared as a condition for the read instruction to be re-awakened;
[0070] In the case that the read instruction does not wait for the source operand within the preset cycle, if the issue queue where the read instruction is located is the memory access instruction issue queue, it continues to wait for the source operand.
[0071] In an optional embodiment of the present application, after the read instruction is returned to the transmit queue, the source operand of the read instruction is obtained by reading from a physical register file.
[0072] See also Figure 5 The emission queue includes the memory access instruction emission queue and the non-memory access instruction emission queue. The memory access instruction emission queue is a sequential FIFO. It receives the dispatched memory access instruction and pops it out from the IQ unconditionally. It reads the operands of the physical register stack at the IQBUF stage. If there are two instructions with a read-before-write dependency, then both instructions are flushed to the OPBUF along the pipeline to wait for the accurate operands to be delivered. Figure 5 In the example, RF-Rport is used to read the physical register stack.
[0073] The emission queue of the present application is a non-memory access instruction emission queue. Each instruction enters sequentially and exits randomly. Each entry of the IQ maintains three flags for replay, authenticity wakeup, and speculative wakeup. The first appearance of the dep signal indicates that the current instruction is speculatively awakened without replay and can be dispatched from the IQ for execution; the depwb signal indicates that the current instruction is authenticity awakened in the case of replay, and the source operand is ready, so it can be dispatched from the IQ. In actual application scenarios, the IQ will select the oldest instruction from the ready instructions in the age matrix maintained by the IQ and pop it out. If the depwb signal is cleared (depwb=0), then the instruction will read the operand of the physical register stack in the IQBUF. If the depwb signal is not cleared for the first time (depwb=1), then the instruction follows the pipeline and rushes to the OPBUF to wait for the accurate number to be delivered. When the instruction rushes to OPBUF, OPBUF maintains a counter to monitor whether the depwb signal is cleared. If the depwb signal is cleared, the instruction will get the data and enter the execution unit for execution; if the counter counts to the preset cycle value and the depwb signal has not been cleared, then the instruction returns to the IQ and waits for the data to be ready. At this time, the instruction cannot be sent to the IQ for dispatch through the dep signal until the depwb signal is cleared.
[0074] The instruction issuance method for an out-of-order processor of the present application can greatly improve the execution efficiency of two instructions with write-before-read dependency through IQ speculative wakeup, and can prevent the pipeline from being stuck due to IQ speculative wakeup in situations such as cache data misses through a replay mechanism, thereby improving the pipeline throughput.
[0075] It should be understood that, although the various steps in the flow chart are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the figure may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.
[0076] See also Figure 6 An embodiment of the present application provides an instruction issuing device for an out-of-order processor, comprising:
[0077] A broadcast module 11 is used for broadcasting the destination register address of the write instruction when the write instruction is selected to be transmitted from the transmission queue for two instructions with a write-before-read dependency;
[0078] A wake-up module 12 is used to determine a source register address that matches a destination register address of a write instruction, and wake up a read instruction corresponding to the source register address, wherein the read instruction is not an instruction returned from an operand cache stage;
[0079] A first determination module 13, configured to determine, through a counter, a cycle of a read instruction waiting for a source operand after the read instruction enters an operand cache phase, wherein the source operand is an operation result of a write instruction;
[0080] An execution module 14, configured to send the source operand of the read instruction to the execution unit to execute the read instruction when the read instruction waits for the source operand within a preset cycle;
[0081] The second determination module 15 is used to determine whether to return the read instruction to the transmit queue or continue to wait for the source operand according to the type of the transmit queue where the read instruction is located when the read instruction does not wait for the source operand within a preset cycle.
[0082] For the specific limitations of the above-mentioned instruction emission device for the out-of-order processor, please refer to the limitations of the instruction emission method for the out-of-order processor above, which will not be repeated here. Each module in the above-mentioned instruction emission device for the out-of-order processor can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.
[0083] In one embodiment, a computer device is provided, the internal structure diagram of which can be as follows: Figure 7 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, an instruction transmission method for an out-of-order processor as described above is implemented. It includes: a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, any step in the instruction transmission method for an out-of-order processor as described above is implemented.
[0084] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, any step in the above instruction issuance method for an out-of-order processor can be implemented.
[0085] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.
[0086] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0087] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0088] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.
[0089] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.
[0090] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.
Claims
1. A method for issuing instructions for an out-of-order processor, characterized in that: include: For two instructions with a write-before-read dependency, when a write instruction is selected to be emitted from the emission queue, the destination register address of the write instruction is broadcast; Determine a source register address that matches the destination register address of the write instruction, and wake up a read instruction corresponding to the source register address, wherein the read instruction is not an instruction returned from the operand cache stage; After the read instruction enters the operand cache stage, a cycle of the read instruction waiting for a source operand is determined by a counter, wherein the source operand is an operation result of the write instruction; When the read instruction waits for the source operand within a preset cycle, the source operand of the read instruction is sent to the execution unit to execute the read instruction; In the case where the read instruction does not wait for the source operand within the preset cycle, it is determined whether to return the read instruction to the transmit queue or continue to wait for the source operand according to the type of the transmit queue where the read instruction is located. The transmitting queue includes a memory access instruction transmitting queue and a non-memory access instruction transmitting queue, and when the read instruction does not wait for the source operand within a preset cycle, determining whether to return the read instruction to the transmitting queue or continue to wait for the source operand according to the type of the transmitting queue where the read instruction is located includes: In the case where the read instruction does not wait for the source operand within the preset cycle, if the emission queue where the read instruction is located is a non-memory access instruction emission queue, the read instruction is returned to the emission queue, and the source operand of the read instruction in the emission queue is prepared as a condition for the read instruction to be re-awakened; In the case that the read instruction does not wait for the source operand within the preset cycle, if the issue queue where the read instruction is located is the memory access instruction issue queue, it continues to wait for the source operand.
2. The method according to claim 1, characterized in that After the read instruction is returned to the issue queue, the source operand of the read instruction is obtained by reading from the physical register file.
3. The method according to claim 1, characterized in that The write instruction is selected to be transmitted from the transmission queue, including: The oldest instruction among the instructions with ready source operands is selected from the age matrix maintained by the issue queue and issued.
4. The method according to claim 1, characterized in that: The operation result of the write instruction is obtained by the following steps: The source operand of the write instruction is read from the physical register stack according to the destination register address, and the write instruction is executed in the execution unit according to the source operand of the write instruction to obtain the operation result of the write instruction.
5. The method according to claim 1, characterized in that The read instruction waits for the source operand within a preset cycle, including: In the case where the write instruction and the read instruction belong to the same transmission queue, the read instruction waits within a preset period for the operation result obtained by the execution module in the same network executing the write instruction; In the case where the write instruction and the read instruction do not belong to the same transmission queue, the read instruction waits within a preset cycle for the operation result obtained by the execution module in the bypass network executing the write instruction.
6. The method according to claim 1, characterized in that The number of the transmitting queues is more than one, and the destination register address of the broadcast write instruction includes: The destination register address of the write instruction is broadcasted to a multiplex network including all the transmit queues, and instructions having a dependency relationship with the write instruction are searched from the instructions of all the transmit queues.
7. An instruction issuing device for an out-of-order processor, characterized in that: include: A broadcast module, for broadcasting a destination register address of a write instruction when the write instruction is selected to be transmitted from the transmission queue for two instructions with a write-before-read dependency; A wake-up module, used to determine a source register address matching a destination register address of a write instruction, and wake up a read instruction corresponding to the source register address, wherein the read instruction is not an instruction returned from an operand cache stage; A first determination module is used to determine, through a counter, a cycle of the read instruction waiting for a source operand after the read instruction enters the operand cache stage, wherein the source operand is an operation result of the write instruction; An execution module, configured to send the source operand of the read instruction to the execution unit to execute the read instruction when the read instruction waits for the source operand within a preset cycle; The second determination module is used to determine whether to return the read instruction to the transmit queue or continue to wait for the source operand according to the type of the transmit queue where the read instruction is located when the read instruction does not wait for the source operand within a preset cycle. The transmitting queue includes a memory access instruction transmitting queue and a non-memory access instruction transmitting queue, and when the read instruction does not wait for the source operand within a preset cycle, determining whether to return the read instruction to the transmitting queue or continue to wait for the source operand according to the type of the transmitting queue where the read instruction is located includes: In the case where the read instruction does not wait for the source operand within the preset cycle, if the emission queue where the read instruction is located is a non-memory access instruction emission queue, the read instruction is returned to the emission queue, and the source operand of the read instruction in the emission queue is prepared as a condition for the read instruction to be re-awakened; In the case that the read instruction does not wait for the source operand within the preset cycle, if the issue queue where the read instruction is located is the memory access instruction issue queue, it continues to wait for the source operand.
8. A computer device comprising: A memory and a processor, wherein the memory stores a computer program, and wherein the processor implements the steps of the instruction issuance method for an out-of-order processor as described in any one of claims 1 to 6 when executing the computer program.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the instruction issuing method for an out-of-order processor according to any one of claims 1 to 6 are implemented.
