Cache managing method and system based on session type

This application claims the benefit of Taiwan application Serial No. 109139486, filed Nov. 12, 2020, the disclosure of which is incorporated by reference herein in its entirety.

The disclosure relates in general to a cache managing method and system, and more particularly to a cache managing method and system based on session type.

In the manufacturing industry, a manufacturing process normally includes several stages. To increase the production capacity and the product yield, each stage of the manufacturing process is adjusted through measurement, calculation and feedback control. Along with the popularity of manufacturing digitization and the evolution of the manufacturing process, the volume of measurement data becomes huge and the control rules are getting more and more complicated. That is, a large volume of data computation needs to be performed in the manufacturing process. Furthermore, the efficiency of data computation may directly affect the production capacity and the product yield.

To increase the efficiency of data computation, the existing technology normally performs data handshaking and data computation using a cache memory and a data computation acceleration unit. Since the cache memory is a limited and expensive resource, the cache memory needs to be managed more efficiently to improve the efficiency of data computation.

According to one embodiment of the present disclosure, a cache managing method based on session type is provided. The cache memory includes several cache blocks. A cache manager has a trace table recording a priority of each cache block. The cache managing method includes the following steps. A request for a first data block is received from a first session, wherein the first session includes a first session type corresponding to a first priority. A cache block whose priority is lower than the first priority is assigned to the first session according to the trace table. The first data block is loaded to the assigned cache block. The priority of the assigned cache block of the trace table is updated to the first priority.

According to another embodiment of the present disclosure, a cache managing system based on session type is provided. The cache managing system based on session type can create a first session with an application, wherein the first session includes a first session type corresponding to a first priority. The cache managing system includes a data source, a cache memory and a cache manager. The data source stores several data blocks. The cache memory includes several cache blocks. The cache manager assigns the cache blocks and includes a trace table recording a priority of each cache block. The cache manager is configured to: assign a cache block whose priority is lower than the first priority to the first session according to the trace table when receiving a request for a first data block from the first session; load the first data block to the assigned cache block from the data source; and update the priority of the assigned cache block of the trace table to the first priority.

The above and other aspects of the disclosure will become better understood with regard to the following detailed description of the preferred but non-limiting embodiment(s). The following description is made with reference to the accompanying drawings.

FIG. 1 is a schematic diagram of a manufacturing process of a manufacturing industry according to an embodiment of the disclosure;

FIG. 2 is a schematic diagram of a cache managing system based on session type according to an embodiment of the disclosure;

FIG. 3 is a flowchart of a cache managing method based on session type according to an embodiment of the disclosure; and

FIG. 4 is a flowchart of sub-steps of step S 180 according to an embodiment of the disclosure.

FIG. 5 is a schematic diagram of a cache managing system based on session type according to another embodiment of the disclosure.

FIG. 6 is a flowchart of a cache managing method based on session type according to another embodiment of the disclosure.

Referring to FIG. 1 , a schematic diagram of a manufacturing process of a manufacturing industry according to an embodiment of the disclosure is shown. A factory F includes several production lines LN 1 , LN 2 , . . . , etc. Each production line includes several workstations. For example, the production line LN 1 includes workstations C 1 , C 2 , C 3 , . . . , etc. Each production line includes a buffer zone, at which the products manufactured in the production line are stored. For example, the production line LN 1 includes a buffer zone BF. Each of the workstations can be composed of an equipment, a measuring machine and a buffer zone. For example, the workstation C 1 is composed of an equipment EP 1 , a measuring machine INS 1 , and a buffer zone BF 1 . The buffer zone is used to store the products of the workstation.

In the manufacturing process, data computation in different zones of the production line has different timeliness requirements, and can be divided into four types according to the timeliness requirements. The four types are in-line type, on-line type, at-line and off-line. The data computation relevant to the interior of the equipment, the workstation, the production line, and the factory are in-line type, on-line type, at-line type, and off-line type, respectively. The immediacy requirements sorted in a descending order are in-line type, on-line type, at-line type and off-line type. Let an actual manufacturing field be taken for example. The in-line type relates to the inspection information obtained when a processing machine is performing a machining task. For example, the run-to-run (R2R) control in a chemical mechanical planarization (CMP) semiconductor process, the processing machine determines whether the processing parameters of the processing machine need to be adjusted according to the R2R inspection information. The in-line type has a high immediacy requirement. The on-line type relates to the inspection information of the output of a processing machine. For example, after a photographic processing is completed, an overlay measurement machine is used to inspect the overlay quality to determine whether to proceed to the next stage or to halt the conveyance of the front open unit pod (FOUP) for further inspection. The immediacy requirement of the on-line type is lower than that of the in-line type. The at-line type relates to the inspection information of phased products. For example, in the production line for metal mobile phone cases, critical dimensions of the material having completed the initial carving process are inspected using three-dimensional measurement to determine whether this batch of material can proceed to the fine carving process. The immediacy requirement of the at-line type is lower than that of the on-line type. The off-line type relates to the inspection information of finished products. For example, a defect inspection is performed on a finished wafer to determine whether the finished wafer passes or not. The immediacy requirement of the off-line type is lower than that of the at-line type.

Referring to FIG. 2 , a schematic diagram of a cache managing system 100 based on session type according to an embodiment of the disclosure is shown. The cache managing system 100 , which creates at least one session with an application 110 , includes a cache manager 120 , a cache memory 130 , a data source 140 and a data computing unit 150 . The cache memory 130 includes several cache blocks L 1 , L 2 , L 3 , L 4 , L 5 , . . . , etc. The data source 140 stores several data blocks D 1 , D 2 , D 3 , D 4 , D 5 , D 6 , D 7 , . . . , etc. The data blocks D 1 , D 2 , D 3 , D 4 , D 5 , D 6 , D 7 . . . can be loaded to the cache blocks L 1 , L 2 , L 3 , L 4 , L 5 , . . . , etc. of the cache memory 130 . The cache manager 120 includes a trace table URT and a session request record table RRT. In the present embodiment, the cache memory 130 can be realized by a static random-access memory (SRAM), and the data source 140 can be realized by a hard disc, a file or a database. Moreover, the application 110 can be executed by a factory management platform; the data computing unit 150 can be realized by a chip, a circuit, a graphics processing unit (GPU), a field programmable gate array (FPGA) or a GPU database; the cache manager 120 can be realized by a driver program or an application used to manage the memory.

The application 110 can create a session with a session type and can send a request for loading a data block. In the present embodiment, the application 110 creates a first session S 1 and a second session S 2 . The session type is relevant with the immediacy requirement in data computation. In the present embodiment, the session type can be one of the in-line type, the on-line type, the at-line type and the off-line type. The session types have different priorities, and the corresponding priorities of the session types sorted in a descending order are the in-line type, the on-line type, the at-line type and the off-line type. The cache manager 120 can assign one of the cache blocks to a session according to the session type of the session and the trace table URT. For example, the cache manager 120 assigns one of the cache blocks to the first session S 1 according to the session type of the first session S 1 and the trace table URT. The data computing unit 150 accesses the assigned cache block to obtain and compute the requested data block, and further transmit the computing result to the session, such as the first session S 1 .

Referring to Table 1, a trace table URT according to an embodiment of the disclosure is shown. The trace table URT includes several columns recording relevant information of each of the cache blocks L 1 , L 2 , L 3 , L 4 , L 5 , . . . , etc. The “block” column of the trace table URT records and indicates the data block stored in the cache block. As indicated in the trace table URT, if the “block” column of the cache block L 1 is N/A, this implies that the cache block L 1 does not store any data block; if the “block” column of the cache block L 2 is “D 1 ”, this implies that the cache block L 1 stores the data block D 1 . The “locked” column of the trace table URT records the number of sessions, to which a cache block is assigned, and indicates that the data computing unit 150 is using the cache block for computation. For example, the trace table URT as indicated in Table 1 shows that the cache block L 3 is assigned to 2 sessions and is being used by the data computing unit 150 for computation. The “priority” column of the trace table URT records the session type corresponding to the highest priority among the sessions having used the data stored in the cache block. For example, the session type corresponding to the highest priority among the sessions having used the data stored in the cache block L 4 is the off-line type. The “session quantity” column of the trace table URT records the quantity of the sessions being using the cache block. For example, 1 session is using the cache block L 5 . The “expiry” column of the trace table URT records the final time at which the cache block is lastly locked. For example, the cache block L 4 is lastly locked at T 2 . In the present specification, the timing sequence is represented by T 1 , T 2 , T 3 , . . . , etc., and the rest can be obtained by the same analogy.

FIG. 3 is a flowchart of a cache managing method based on session type according to an embodiment of the disclosure. Refer to FIGS. 2 and 3 and Table 1. Detailed descriptions of the steps of the cache managing method of the present embodiment are disclosed below. Firstly, the method begins at step S 110 , a first session S 1 is created by the application 110 , and a request for accessing a data block is transmitted by the first session S 1 . The first session S 1 includes a first session type (on-line) corresponding to a first priority.

Next, the method proceeds to step S 120 , a request for loading a data block from the first session S 1 is received by the cache manager 120 .

Then, the method proceeds to step S 130 , whether the data block requested by the first session S 1 is already stored in the cache memory 130 is determined by the cache manager 120 according to the trace table URT. When the data block requested by the first session S 1 already exists in the cache memory 130 , the method proceeds to step S 140 ; when the data block requested by the first session S 1 does not exist in the cache memory 130 , the method proceeds to step S 180 .

For example, if the first session S 1 requests for the data block D 4 , the cache manager 120 determines that the data block D 4 is stored in the cache block L 3 of the cache memory 130 according to the “block” column of the trace table URT as indicated in Table 1. That is, since the “block” column of the cache block L 3 of the trace table URT is “D 4 ”, the cache manager 120 determines that that the data block requested by the first session S 1 exists in the cache memory 130 , and the method proceeds to step S 140 . In step S 140 , the cache block L 3 storing the data block D 4 requested by the first session S 1 is assigned to the first session S 1 by the cache manager 120 .

Then, the method proceeds to step S 150 , the data block D 4 is obtained from the assigned cache block L 3 and used by the data computing unit 150 for computation, and the “locked” column, “expiry” column, “session quantity” column and “priority” column of the trace table URT are updated by the cache manager 120 . For example, the trace table URT updated by the cache manager 120 is listed in Table 2. In the trace table URT as indicated in Table 2, the cache manager 120 updates the “locked” column corresponding to the assigned cache block L 3 to 3 from 2, updates the “priority” column from the at-line type to the on-line type corresponding to the first session S 1 with a higher priority, updates the “session quantity” column to 4 from 3, and updates the “expiry” column to the lastly locked time T 4 from T 3 . In other words, when the cache manager 120 assigns the cache block L 3 to the first session S 1 , the first session S 1 notifies the data computing unit 150 to access data from the cache block L 3 , and the cache manager 120 further locks the cache block L 3 and adds the “locked” column corresponding to the assigned cache block L 3 by 1, updates the “expiry” column to the lastly locked time, and adds the value of the “session quantity” column by 1. Moreover, if the priority of the first session type (on-line) of the first session S 1 is higher than the priority of the session type (at-line) recorded in Table 1, the cache manager 120 updates the “priority” column to the first session type (on-line). In another embodiment, if the first session type of the first session S 1 is the at-line type or the off-line type and the first priority of the first session type is not higher than the priority (at-line) of the cache block L 3 recorded in Table 1, then the “priority” column of the trace table URT will not be updated in the step S 150 .

Then, the method proceeds to step S 160 , the computing result is transmitted to the first session S 1 by the data computing unit 150 , and the “locked” column of the trace table URT is updated by the cache manager 120 . For example, the cache manager 120 updates the “locked” column of the cache block L 3 of the trace table URT as indicated in Table 2, and the updated trace table URT is listed in Table 3. In Table 3, the value of the “locked” column of the cache block L 3 is updated to 2 from 3. In other words, after the data computing unit 150 transmits the computing result to the first session S 1 and finishes the computation using the cache block L 3 , the cache manager 120 deducts the value of the “locked” column corresponding to the cache block L 3 of the trace table URT by 1.

Then, the method proceeds to step S 170 , the first session S 1 is terminated by the application 110 , and the “session quantity” column of the trace table URT is updated by the cache manager 120 . For example, after the application 110 terminates the first session S 1 , the cache manager 120 updates the “session quantity” column of the cache block L 3 of the trace table URT as indicated in Table 3, and the updated trace table URT is listed in Table 4. In the trace table URT as indicated in Table 4, the session quantity of the cache block L 3 is updated to 3 from 4. In other words, after the first session S 1 is terminated, the cache manager 120 deducts the session quantity of the cache block L 3 of the trace table URT by 1. In an embodiment, after the first session S 1 is terminated, the cache manager 120 deducts the session quantity of each cache block corresponding to the first session S 1 of the trace table URT by 1. In the present embodiment, through the session request record table RRT of Table 5, the cache manager 120 records the cache block requested by the first session S 1 . In the session request record table RRT of Table 5, the data block requested by the first session S 1 includes data blocks D 1 and D 4 . Therefore, after the first session S 1 is terminated, the cache manager 120 deducts the session quantity of each of the cache blocks L 2 and L 3 corresponding to the first session S 1 of the trace table URT of Table 3 by 1 according to the session request record table RRT of Table 5, and the updated trace table URT is listed in Table 6. In the present embodiment, the session request record table RRT records the data block requested by the session. However, the session request record table RRT can also record the cache block, to which the data block requested by the session is assigned.

In step S 110 , when the first session S 1 requests for accessing the data block D 3 , if the cache manager 120 determines that the data block D 3 does not exist in the cache memory 130 according to the trace table URT of Table 1, the method proceeds to step S 180 . Since the requested data block D 3 is not stored in the cache memory 130 , the cache manager 120 needs to assign a cache block and load the data block D 3 to the assigned cache block from the data source 140 for the first session S 1 to use. When all cache blocks are used, the cache manager 120 needs to release the assigned cache block beforehand so that data can be loaded to the assigned cache block.

In step S 180 , a cache block is assigned to the first session S 1 by the cache manager 120 according to the trace table URT. Referring to FIG. 4 , a flowchart of sub-steps of step S 180 according to an embodiment of the disclosure is shown. In the present embodiment, step S 180 includes sub-steps S 181 to S 187 , wherein the cache manager 120 assigns one of the several cache blocks L 1 , L 2 , L 3 , L 4 , L 5 to the first session S 1 according to the trace table URT.

Then, the method proceeds to step S 181 , whether there exists an empty cache block is determined by the cache manager 120 according to the “block” column of the trace table URT. If there exists an empty cache block, the method proceeds to step S 186 . If there does not exist any empty cache blocks, the method proceeds to step S 182 . Refer to the trace table URT as indicated in Table 1. If the cache manager 120 determines that there exists an empty cache block (that is, the cache block L 1 ) according to the “block” column of the trace table URT, the method proceeds to step S 186 , the cache block L 1 is assigned to the first session S 1 by the cache manager 120 . In an embodiment, if the cache manager 120 determines that the trace table URT has more than one empty cache block, the cache manager 120 assigns one of the empty cache blocks to the first session S 1 in step S 186 .

In step S 181 , if the cache manager 120 determines that the trace table URT as indicated in Table 7 does not have any empty cache blocks according to the “block” column of the trace table URT, the method proceeds to step S 182 . In step S 182 , whether there exists an unlocked cache block is determined by the cache manager 120 according to the “locked” column of the trace table URT. If it is determined that there exists an unlocked cache block, the method proceeds to step S 183 . If it is determined that there does not exist any unlocked cache blocks, the method proceeds to step S 187 , a “fail” message is transmitted to the first session S 1 by the cache manager 120 . Refer to the trace table URT as indicated in Table 7. If the cache manager 120 determines that there exists unlocked cache blocks L 1 , L 2 , and L 5 according to the “block” column of the trace table URT, the method proceeds to step S 183 .

In step S 183 , whether all unlocked cache blocks are used in the session is determined by the cache manager 120 according to the “session quantity” column of the trace table URT. If it is determined that all unlocked cache blocks are used in the session, the method proceeds to step S 185 . If it is determined not all unlocked cache blocks are used in the session, the method proceeds to step S 184 . For example, the cache manager 120 determines that the values of the “session quantity” column corresponding to the unlocked cache blocks L 1 , L 2 , and L 5 are all greater than 0 (that is, the unlocked cache blocks L 1 , L 2 , and L 5 are all used in the session) according to the “session quantity” column of the trace table URT as indicated in Table 7, the method proceeds to step S 185 .

In step S 185 , the cache block whose priority is lower than the first priority of the first session type is selected by the cache manager 120 according to the “priority” column of the trace table URT. For example, referring to the trace table URT as Table 7, the priority of the in-line type of the cache block L 1 is not lower than the first priority of the first session type (on-line) of the first session S 1 , the priority of the in-line of the cache block L 2 is not lower than the first priority of the first session type (on-line) of the first session S 1 , and the priority of the at-line type of the cache block L 5 is lower than the first priority of the first session type (on-line) of the first session S 1 , therefore the cache manager 120 selects the cache block L 5 . Then, the method proceeds to step S 186 , the cache block L 5 selected in step S 185 is assigned to the first session S 1 by the cache manager 120 . In an embodiment, when step S 185 is performed, if the quantity of the cache blocks whose priority is lower than the priority of the first session type is greater than 1, the cache manager 120 selects any one of the cache blocks whose priority is lower than the priority of the first session type, and assigns the selected cache block to the first session S 1 in step S 186 .

Refer to the trace table URT as indicated in Table 8. In step S 183 , if it is determined by the cache manager 120 according to the “session quantity” column of the trace table URT that not all unlocked cache blocks L 1 , L 2 , and L 5 are used in the session, the method proceeds to step S 184 . In step S 184 , the cache block whose lastly locked time is most remote from the current time is selected from the cache blocks not being used in the session by the cache manager 120 according to the “expiry” column of trace table URT. For example, the cache manager 120 selects the cache block L 1 from the cache blocks L 1 and L 2 according to the “expiry” column of the trace table URT, wherein the cache blocks L 1 and L 2 are not being used in the session, and the lastly locked time of the cache block L 1 is most remote from the current time. Then, the method proceeds to step S 186 , the cache block L 1 is assigned to the first session S 1 by the cache manager 120 .

Then, the method proceeds to step S 190 , the cache block assigned in step S 180 is released by the cache manager 120 , and the data block is loaded to the assigned cache block from the data source 140 . Following step S 190 , steps S 150 to S 170 are sequentially performed. Steps S 150 to S 170 are already disclosed above and are not repeated here.

In the above disclosure, it is exemplified that the application 110 creates a first session S 1 and the first session S 1 requests for a data block. However, it should be noted that the first session S 1 can transmit request for several times to access different data blocks. That is, the session can perform the cache managing method for several times. In another embodiment, after the application 110 creates the first session S 1 , the application 110 can further create a second session S 2 , which request for a data block. The second session S 2 includes a second session type corresponding to a second priority. The data block requested by the second session S 2 can be identical to or different from the data block requested by the first session S 1 , and the second priority of the second session type of the second session S 2 can be identical to or different from the first priority of the first session type of the first session S 1 . Moreover, the application 110 can create the second session S 2 before or after the first session S 1 is terminated. Also, the second session S 2 can be created by another application, and not limited to the disclosed embodiments. In the present embodiment, like the first session type, the second session type can be one of the in-line type, the on-line type, the at-line type and the off-line type.

Refer to FIGS. 5 and 6 and Table 9. FIG. 5 is a schematic diagram of a cache managing system 100 based on session type according to another embodiment of the disclosure. FIG. 6 is a flowchart of a cache managing method based on session type according to another embodiment of the disclosure. The cache managing system 100 based on session type creates a first session with the application 110 , wherein the first session S 1 includes a first session type corresponding to a first priority. The cache managing system 100 includes a cache manager 120 , a cache memory 130 and a data source 140 . The data source 140 stores several data blocks D 1 , D 2 , D 3 , D 4 , D 5 , D 6 , D 7 , . . . , etc. The cache memory 130 includes several cache blocks L 1 , L 2 , L 3 , L 4 , L 5 , . . . , etc. The cache manager 120 , which assign the cache blocks L 1 , L 2 , L 3 , L 4 , L 5 , . . . , etc. to sessions, further includes a trace table URT as indicated in Table 9. The trace table URT records a priority of each of the cache blocks L 1 , L 2 , L 3 , L 4 , L 5 , . . . , etc. The cache manager 120 is configured to perform the following steps.

In step S 210 , a request for loading the data block D 3 is received from a first session S 1 by the cache manager 120 , wherein the first session S 1 includes a first session type (at-line) corresponding to a first priority.

In step S 220 , a cache block L 4 whose priority is lower than the first priority is assigned to the first session S 1 by the cache manager 120 according to the trace table URT. Referring to Table 9, since the priority (off-line) of the cache block L 4 is lower than the first priority of the first session type (at-line), the cache manager 120 assigns the cache block L 4 to the first session S 1 .

In step S 230 , the data block D 3 is loaded to the assigned cache block L 4 by the cache manager 120 .

In step S 240 , the priority of the assigned cache block L 4 of the trace table URT is updated to the first priority by the cache manager 120 . In the trace table URT as indicated in Table 10, the priority of the assigned cache block L 4 is updated to the at-line type by the cache manager 120 . Besides, in the trace table URT as indicated in Table 10, the cache manager 120 can also update the value of the “locked” column corresponding to the assigned cache block L 4 to 1, update the session quantity to 1, update the expiry time to the lastly locked time T 4 , and update the “block” column to the data block D 3 .

The cache managing method and cache managing system based on session type of the present disclosure can manage the cache memory according the priorities of the session types and the trace table for enabling data computation with higher priority to preferentially obtain computing resources and stay in the cache memory longer, such that the efficiency of data access and computation can be increased. Particularly, when used in the manufacturing process, the cache managing method and cache managing system based on session type of the present disclosure can manage the cache memory according to the immediacy requirement of data and data computation, such that the data computation required for measurement, calculation and feedback control can meet the operation requirements of the production line, such that the production line will not be idled due to data access, and the production capacity can be increased.

It will be apparent to those skilled in the art that various modifications and variations can be made to the disclosed embodiments. It is intended that the specification and examples be considered as exemplary only, with a true scope of the disclosure being indicated by the following claims and their equivalents.