A chip test data processing system

Abstract

The present invention relates to a chip test data processing system, which includes an automated calculation subsystem based on established rules and an automatic prompt subsystem based on cell colors; the automated calculation subsystem based on established rules includes: a data import module configured to import the reference values of the reference chip and the original test data of the comparison chip, and an automatic calculation module configured to calculate the scores of the corresponding indicators based on the original test data of the reference chip and the comparison chip according to a pre-configured formula; the automatic prompt subsystem based on cell colors includes: a missing filling prompt module configured to detect whether the value of a cell is empty, and if so, display the background color of the cell as a first color, and a filling error prompt module configured to detect whether the value of the cell meets the pre-configured conditions, and if not, display the background color of the cell as a second color. Compared with the prior art, the present invention can save labor costs and improve efficiency.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and more particularly to a chip test data processing system. Background Art

[0002] Chip evaluation is an essential part in the process of chip research and development and selection. Chip evaluation mainly tests various performances of the chip, summarizes the test results, and obtains some results by using statistics.

[0003] Currently, when comparing two chips, the method adopted is to collect the data obtained from various tests in the form of manual input, and then perform a formula calculation on each data. The process is very cumbersome and manual operations are inevitably prone to mistakes. On the one hand, the process is cumbersome; on the other hand, in order to solve the possible mistakes in manual calculation, a review is required, which further reduces the efficiency. Summary of the Invention

[0004] The purpose of the present invention is to provide a chip test data processing system, which optimizes the test process of artificial intelligence chips. Compared with the manual method, the present invention can make better use of computer resources, directly analyze and calculate data by the pre-written function without manual calculation, greatly increasing the calculation accuracy rate, and can save labor costs and improve efficiency.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] 1. A chip test data processing system, characterized in that the system includes an automated calculation subsystem based on established rules and an automatic prompt subsystem based on cell colors;

[0007] The automated calculation subsystem based on established rules includes:

[0008] A data import module, configured to import the reference values of the reference chip and the original test data of the comparison chip,

[0009] An automatic calculation module, configured to calculate the scores of the corresponding indicators based on the reference chip and the original test data of the comparison chip according to the pre-configured formula;

[0010] The automatic prompt subsystem based on cell colors includes:

[0011] An unfilled prompt module, configured to detect whether the value of the cell is empty. If so, display the background color of the cell as the first color,

[0012] A filling error prompt module, configured to detect whether the value of the cell meets the pre-configured conditions. If not, display the background color of the cell as the second color.

[0013] The types of the pre-configured formulas include:

[0014] Arithmetic mean formula: For test items where the higher the score, the better the performance, divide the test value of the comparison chip by the reference value of the reference chip to obtain a ratio; for test items where the lower the score, the better the performance, divide the reference value of the reference chip by the test value of the test chip to obtain a ratio, and calculate the arithmetic mean of all the obtained ratios;

[0015] Weighted average formula: For test items where the higher the score, the better the performance, divide the test value of the comparison chip by the reference value of the reference chip to obtain a ratio; for test items where the lower the score, the better the performance, divide the reference value of the reference chip by the test value of the test chip to obtain a ratio, and calculate the weighted average by combining each ratio with its corresponding weight;

[0016] Piecewise function formula: When the performance difference between the reference value and the test value is too large, in order to facilitate comparison between chips, the score of each chip is represented by a normalized mapping. The number of segments and the value range of the piecewise function can be flexibly adjusted according to the actual situation or requirements. The score will no longer be calculated using a formula with the reference value, but directly take the score of the segment where the test value is located;

[0017] Multi-function formula: When multiple conditions occur, different judgment schemes will be adopted: When there are multiple test environments, it is necessary to judge the actual scoring rules. If the maximum score is required, for each environment, divide the test value of the comparison chip by the reference value of the reference chip to obtain a ratio as the environment score for the corresponding environment. Then, through a judgment function, compare all the environment scores pairwise to finally find the maximum score as the score for this test item. If the average value is required, it is necessary to judge whether each environment is tested. If not tested, skip it. Finally, calculate the average value based on all the test scores divided by the number of test environments; When the number of test items needs to be restricted, it is necessary to use a counting function to judge whether the required number of tests is reached. If reached, calculate the result by dividing the normal test value by the reference value. If not reached, the result needs to be obtained according to a certain proportion of the normal situation or directly output the result as 0.

[0018] Specifically, calculating the mean value based on the scores of all test items means: calculating the arithmetic mean value based on the scores of all test items.

[0019] Specifically, calculating the mean value based on the scores of all test items means: calculating the weighted average value based on the scores of all test items.

[0020] The missing filling prompt module is further configured to perform the following steps: If the value of the cell is not allowed to be empty, display the background color of the cell as the third color.

[0021] The third color is blue.

[0022] The system further includes a multi-level linkage subsystem, which includes:

[0023] A link module, configured to implement jumps between the sub-tables and the summary table corresponding to each test item;

[0024] A summary module, configured to record the data in the sub-tables corresponding to each test item into the summary table, and calculate the total score through a function with preset values.

[0025] The system further includes a multi-table linkage subsystem, which includes:

[0026] A comparison table generation module, configured to read the summary tables of each comparison chip and generate a horizontal comparison table based on the summary tables of each comparison chip;

[0027] A visualization module, configured to generate statistical charts based on the data in the horizontal comparison table.

[0028] The system further includes a protection subsystem, which is configured to hide some rows and columns and restrict the editing of some cells.

[0029] The system further includes a grading and scoring logic subsystem, which is configured to:

[0030] In a simple environment, directly normalize the highest and lowest scores of all test chips participating in the test to 1 and 0, and divide them into several segments within the range;

[0031] In a complex environment, use calculation logics including data binning and linear regression for segmentation to ensure that the value range of the scores is between 0 and 1.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] 1. Optimize the test process of artificial intelligence chips. Compared with the manual method, the present invention can better utilize computer resources, directly analyze and calculate data by a pre-written function without manual calculation, greatly increasing the calculation accuracy rate, and can save labor costs and improve efficiency.

[0034] 2. Use the conditional formatting function to remind the filler in a visual form marked with colors, improving the convenience of self-review.

[0035] 3. Simultaneously use functions such as freezing cells and protecting worksheets to prevent the table format and benchmark indicators from being tampered with; Brief Description of the Drawings

[0036] Figure 1 It is a schematic diagram of the components of the system of the present invention;

[0037] Figure 2 It is a schematic diagram for horizontal comparison. Specific implementation mode

[0038] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented on the premise of the technical solution of the present invention, and gives detailed implementation manners and specific operation processes, but the protection scope of the present invention is not limited to the following embodiments.

[0039] This application is mainly aimed at using a spreadsheet to collect manufacturer verification data. Based on the requirements of the chip evaluation implementation plan, each test score is automatically calculated through a function in the spreadsheet and entered into the summary table, and finally the total score is automatically obtained. Using this method can greatly reduce the probability of manual errors and save a large amount of manual time. A data set that originally took about 1 hour to process can now be completed in only 5 minutes. And if there are adjustments in the score calculation later, only the calculation formula needs to be rewritten.

[0040] Specifically, a chip test data processing system, as Figure 1 shown. In this embodiment, the system is implemented based on spreadsheet processing software. Of course, in other embodiments, other solutions can also be adopted, such as an independently developed program. When making the solution, for each test item, a worksheet is created respectively, and for each comparison chip, a workbook is created. The system includes an automated scoring subsystem based on established rules and an automatic prompt subsystem based on cell colors;

[0041] The automated scoring subsystem based on established rules includes:

[0042] A data import module configured to import the reference values of the reference chip and the original test data of the comparison chip,

[0043] An automatic calculation module configured to calculate the scores of the corresponding indicators based on the original test data of the reference chip and the comparison chip according to the pre-configured formula;

[0044] The types of the pre-configured formula include:

[0045] Arithmetic mean formula: For most test items, the scoring rules are relatively simple. For test items where the higher the score, the better the performance, divide the test value of the comparison chip by the reference value of the reference chip to obtain a ratio; for test items where the lower the score, the better the performance, divide the reference value of the reference chip by the test value of the test chip to obtain a ratio. And calculate the arithmetic mean of all the obtained ratios;

[0046] Weighted average formula: A small part will add weights in the small-item tests. For the test items with higher scores and better performance, divide the test value of the comparison chip by the reference value of the reference chip to obtain a ratio; for the test items with lower scores and better performance, divide the reference value of the reference chip by the test value of the test chip to obtain a ratio. Then calculate the weighted average by combining each ratio with its corresponding weight;

[0047] Piecewise function formula: When the performance difference between the reference value and the test value is too large, in order to facilitate the comparison between chips, a piecewise function will be used to represent each chip score in a normalized mapping. The number of segments and the value range of the piecewise function can be flexibly adjusted according to the actual situation or requirements. The score will no longer be calculated with the reference value through a formula, but directly obtain the score according to the segment range where the test value is located;

[0048] Multi-function formula [judgment function formula]: The most complex part of the test will use multiple functions to calculate the final score. For such test items, it is necessary to first determine which scoring logic it applies to, and then adopt different judgment schemes: When there are multiple test environments, it is necessary to determine the actual scoring rules. If the maximum score of each environment is required, for each environment, divide the test value of the comparison chip by the reference value of the reference chip to obtain a ratio as the environment score for the corresponding environment. Then, through the judgment function, compare the environment scores of all environments pairwise to finally find the maximum score as the score for this item. If the average value of all environment scores is required, divide the sum of all environment scores obtained by the number of existing environments to obtain the score for a single test item; When the test item score needs to exclude untested items from the score, it is necessary to determine whether there is a number filled in the test value column in each environment. If not, it means that the test has not been performed, and it will be excluded from the score formula; When restricting the number of test items, it is necessary to use a counting function to determine whether the required number of tests is reached. If it is reached, calculate the result by dividing the normal test value by the reference value. If not reached or exceeded the limit, the result needs to be obtained according to a certain proportion of the normal situation or directly output the result as 0; When a score requires multiple test items and weights, it is necessary to capture the scores calculated in each test item involved in the score and multiply them by the corresponding weights.

[0049] The purpose of setting up the automatic prompt subsystem based on cell color is to clearly see the items that need to be filled. The automatic prompt subsystem based on cell color includes:

[0050] The unfilled prompt module is configured to detect whether the value of the cell is empty. If so, display the background color of the cell as the first color, otherwise, display the background color of the cell as the third color. In this embodiment, the third color is blue and the first color is red.

[0051] The filling error prompt module is configured to detect whether the value of the cell meets the pre-configured conditions. If not, the background color of the cell is displayed as the second color. In this embodiment, the second color is red.

[0052] To clearly see the items that need to be filled, conditional formatting is set for the areas that should be filled. When the wrong format is filled in the area that needs to be filled or it is forgotten to be filled, the system will automatically identify and mark such areas with a red background, and use a blue background to mark the filled areas, increasing the convenience of self-examination for the fillers. Moreover, for some test items with special requirements, conditional formatting alone cannot explain well, so functions are also set to automatically judge whether the test items meet the requirements, and corresponding text prompts will be generated for the items that do not meet the requirements.

[0053] The system also includes a multi-level linkage subsystem, and this multi-level linkage subsystem includes:

[0054] The link module is configured to achieve the jump between the sub-tables corresponding to each test item and the summary table;

[0055] The summary module is configured to record the data in the sub-tables corresponding to each test item into the summary table and calculate the total score through a function with preset values.

[0056] Because each test item has a corresponding worksheet, too many worksheets make the in-table switching operation complicated. For this reason, a table of contents is also made, and links are used to link the table of contents with the worksheets of each test item, achieving the function of embedding an index in the table of contents, and a button to directly access the table of contents is set in each table, which can not only improve the aesthetics but also optimize the usage experience.

[0057] Considering that too many test items may cause confusion in the total score calculation, easily lead to omissions, and also hinder the subsequent report production and horizontal comparison, a summary table is first made in the spreadsheet, integrating all the test items included in the total score formula weight into the summary table, and through the data table linkage scheme, the scores of each test item are automatically recorded from their respective worksheets into the summary table, and the total score is automatically calculated through a function, without the need to manually grab the score of each item in each test item table and calculate, saving labor costs and reducing the error probability.

[0058] The system also includes a multi-table linkage subsystem, and this multi-table linkage subsystem includes:

[0059] The comparison table generation module is configured to read the summary tables of each comparison chip and generate a horizontal comparison table based on the summary tables of each comparison chip;

[0060] The visualization module is configured to generate statistical charts based on the data in the horizontal comparison table.

[0061] For convenient horizontal comparison, multiple spreadsheets are also linked. After all the data of the evaluated chips is entered, all the data scores and the total scores will first be recorded in the summary table, and then linked with another spreadsheet to automatically read the summary table of each test chip and import it into a horizontal comparison table. As Figure 2 shown, a series of charts have been made in the table based on requirements. Once the data is read, the charts can be fully automatically generated, and no manual operation is required throughout the process, eliminating human errors and saving the time cost of making charts.

[0062] The system also includes a protection subsystem, which is configured to hide some rows and columns and restrict the editing of some cells. Specifically, to avoid duplicate labor time, the spreadsheet is directly sent to the test manufacturer for filling. To ensure that some calculation logics and table forms are not damaged by misoperations and to hide some calculation logics, an editable area is set, some rows and columns are hidden, and the worksheet and workbook are protected, and the operable permissions are set to achieve the unity of the worksheet.

[0063] The system also includes a grading and scoring logic subsystem, which is configured to: in a simple environment, directly normalize the highest and lowest scores of all the test chips participating in the test to 1 and 0, and divide them into several segments within the range; in a complex situation, use calculation logics including data binning and linear regression for segmentation to ensure that the value range of the scores is between 0 and 1.

[0064] Specifically, in addition, grading and scoring are performed on the scores of the test items within the basic technical specification test indicators, and the grading and scoring algorithm can be adjusted according to the actual situation: in a simple environment, the highest and lowest scores of all the test chips participating in the test can be directly normalized to 1 and 0, and divided into several segments within the range; in a complex situation (such as when the data volume is large and the score calculation is complex), calculation logics such as data binning and linear regression can be used for segmentation. In this way, it can be ensured that the value range of the scores is between 0 and 1, which is convenient for subsequent statistics.

[0065] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

Claims

1. A chip test data processing system, characterized in that, The system includes an automatic counting subsystem based on established rules and an automatic prompt subsystem based on cell color; The automatic calculation molecular system based on established rules includes: The data import module is configured to import the reference value of the reference chip and the original test data of the comparison chip, The automatic calculation module is configured to calculate the scores of the corresponding indicators based on the original test data of the reference chip and the comparison chip according to a preconfigured formula; The automatic prompt subsystem based on cell color includes: The missing value prompt module is configured to detect whether the value of the cell is empty, and if so, display the background color of the cell as the first color. The filling error prompt module is configured to detect whether the value of the cell meets the pre-configured condition, and if not, the background color of the cell is displayed as the second color; The types of preconfigured formulas include: Arithmetic mean formula: For test items with higher scores and better performance, the test value of the comparison chip is divided by the benchmark value of the benchmark chip to obtain a ratio; for test items with lower scores and better performance, the benchmark value of the comparison benchmark chip is divided by the test value of the test chip to obtain a ratio, and the arithmetic mean of all the obtained ratios is calculated; Weighted average formula: For test items with higher scores and better performance, the test value of the comparison chip is divided by the benchmark value of the benchmark chip to obtain a ratio; for test items with lower scores and better performance, the benchmark value of the comparison benchmark chip is divided by the test value of the test chip to obtain a ratio, and the weighted average value is calculated by combining each ratio with its corresponding weight; Piecewise function formula: When the performance difference between the benchmark value and the test value is too large, in order to facilitate the comparison between chips, the score of each chip is expressed as a normalized mapping. The number of segments and value range of the piecewise function are flexibly adjusted according to the actual situation or needs. The score will no longer be calculated with the benchmark value, but directly take the score of the segment where the test value is located; Multi-function formula: When multiple conditions appear, different judgment schemes will be adopted: When there are multiple test environments, it is necessary to judge the actual scoring rules. If the maximum value score is required, for each environment, the test value of the comparison chip is divided by the benchmark value of the benchmark chip to obtain the ratio as the environmental score of the corresponding environment. Then, all environmental scores are compared pairwise through the judgment function to finally find the maximum score as the score of the test item. If the average value is required, it is necessary to determine whether each environment is tested. If not, it is skipped. Finally, the average value is calculated based on all test scores divided by the number of test environments. When the number of test items needs to be limited, the counting function is needed to determine whether the required number of tests is reached. If reached, the result is obtained by dividing the normal test value by the benchmark value. If not reached, the result needs to be obtained based on a certain proportion of the normal situation according to the actual situation or the result is directly output as 0.

2. The chip test data processing system according to claim 1, characterized in that, The missing value prompt module is configured to further perform the following steps: if the value of the cell is not left blank, the background color of the cell is displayed as a third color.

3. A chip test data processing system according to claim 2, characterized in that, The third color is blue.

4. A chip test data processing system according to claim 1, characterized in that, The system also includes a multi-level linkage subsystem, which includes: A link module, configured to implement a jump between the sub-table and the summary table corresponding to each test item; The summary module is configured to record the data in the sub-tables corresponding to each test item into the summary table and calculate the total score through a function with preset values.

5. A chip test data processing system according to claim 4, wherein The system further includes a multi-table linkage subsystem, and the multi-table linkage subsystem includes: A comparison table generation module, configured to read the summary tables of each comparison chip and generate a horizontal comparison table based on the summary tables of each comparison chip; A visualization module, configured to generate statistical charts based on the data in the horizontal comparison table.

6. The chip test data processing system according to claim 1, wherein The system further includes a protection subsystem, and the protection subsystem is configured to hide some rows and columns and restrict the editing of some cells.

7. A chip test data processing system according to claim 1, wherein, The system further includes a grading and scoring logic subsystem, and the grading and scoring logic subsystem is configured to: In a simple environment, directly normalize the highest score and the lowest score of all test chips participating in the test to 1 and 0, and divide them into several segments within the range; In a complex situation, use calculation logic including data binning and linear regression for segmentation to ensure that the value range of the score is between 0 and 1.

Images