A single-particle functional error testing method and system for dynamically reconfigurable chips
By designing the test programs P1-P5, the single-particle function errors were tested for different functional modules of the dynamic reconstructible chip, which solved the problem of difficulty in evaluating the chip's radiation resistance in the existing technology, and achieved the single-particle effect on each module of the chip Full coverage testing and system-level reinforcement.
Patent Information
- Application Number
- CN202111015116.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-08-31
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2041-08-31
AI Technical Summary
Dynamic reconstructible chips are prone to ionization of semiconductor materials in a radiation environment, resulting in functional failure. The existing technology lacks effective testing methods for the functional errors of single particles in different functional modules, making it difficult to accurately evaluate their radiation resistance.
A single-particle function error test method for dynamic reconfigurable chips was designed. By writing test programs P1-P5, the single-particle sensitivity of the controller, on-chip interconnect network and PE processing unit array were tested, and the single-particle particles of each functional module were counted. Functional error section, providing system-level reinforcement reference.
It has achieved full coverage testing of single-particle functional errors in each functional module of the dynamic reconfigurable chip, accurately grasping its single-particle effect sensitivity, providing an important reference for radiation-resistant reinforcement design, and improving the radiation-resistant performance of the chip.
Smart Images

Figure CN113868043B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a single-particle functional error testing method and system of a dynamically reconfigurable chip, belonging to the technical field of device testing. Background Art
[0002] In the military field, network information has become the fifth strategic space, which is as important as land, sea, air and space, and is related to national defense security. Dynamically reconfigurable chips are functional flexible structures. Hardware adapts to software to achieve the best match between hardware and software. Reconfigurable technology provides high computing performance and application flexibility. The use of dynamic reconfigurable chips can configure the functions and parameters of each functional module in the satellite integrated information processing system online according to the system task requirements in a very short time, and quickly realize comprehensive tasks such as communication, navigation, measurement and control, target detection and identification, flight control and information support. While realizing the coordinated processing of comprehensive tasks, it effectively saves system resources, makes up for the lack of effectiveness of large satellite platforms, and realizes tactical rapid response to emergency incidents. Therefore, there is a wide demand in aerospace and weapon systems. However, with the application of dynamic reconfigurable chips in aerospace, new radiation effects and reliability problems arise because the working mechanism and device structure of the device are different from traditional chips. When dynamically reconfigurable chips are used in the fields of national defense and aerospace, incident semiconductor materials such as high-energy particles and heavy particles in the radiation environment will cause ionization of semiconductor materials, resulting in various forms of failure. Radiation resistance is one of the most important indicators of dynamically reconfigurable chips for space applications. Accurate and objective test data can intuitively reflect the radiation resistance of radiation-resistant dynamically reconfigurable chips, which is of great significance for the research on radiation effect testing methods.
[0003] At present, most of the literature on radiation effects related to dynamically reconfigurable chips found in China is based on the single-particle upset and reinforcement technology of memory, and there are few reports on the research of single-particle functional errors in different functional modules of dynamically reconfigurable chips. Summary of the invention
[0004] The purpose of the present invention is to overcome the above defects, provide a single particle functional error test method and system for a dynamically reconfigurable chip, provide a single particle functional error test process for different modules, and perform system-level reinforcement for functional modules with a high sensitivity ratio by separately counting the single particle functional error cross-section of each functional module in the dynamically reconfigurable chip, thereby improving the radiation resistance of the chip. The present invention has the feature of comprehensive coverage of single particle functional error testing, and provides a reference for single particle functional testing and single particle radiation performance evaluation of such reconfigurable chips.
[0005] In order to achieve the above-mentioned object of the invention, the present invention provides the following technical solutions:
[0006] The present invention provides a single-particle functional error testing method for a dynamically reconfigurable chip. The dynamically reconfigurable chip includes a PE processing unit array, an on-chip interconnection network, and a controller. The testing method of the present invention includes the following steps:
[0007] (1) Design test procedures P1, P2, P3, P4 and P5; P1 is used to test the single-particle sensitivity σ of the controller module 1 , P2 is used to test the configuration layer sensitivity of the controller and the on-chip interconnect network σ 2 , P3 is used to test the configuration layer and application layer sensitivity of the controller and the on-chip interconnect network σ 3 , P4 is used to test the configuration layer sensitivity of the controller, the on-chip interconnect network configuration layer, and the PE processing unit array σ 4 P5 is used to test the sensitivity of the controller, the configuration layer and application layer of the on-chip interconnect network, and the configuration layer and application layer of the PE processing unit. 5 ;
[0008] (2) Only the controller is irradiated and the single particle sensitivity σ of the controller module is obtained according to test procedure P1 1 ;
[0009] (3) Irradiate the entire chip and obtain the configuration layer sensitivity σ of the controller and on-chip interconnect network according to test procedure P2 2 ;
[0010] (4) Irradiate the entire chip and obtain the configuration layer and application layer sensitivity σ of the controller and on-chip interconnect network according to the test program P3 and the test values required by the chip external input. 3 ;
[0011] (5) Irradiate the entire chip and obtain the configuration layer sensitivity σ of the controller, on-chip interconnect network configuration layer, and PE processing unit array according to test procedure P4 4 ;
[0012] (6) Irradiate the entire chip and obtain the sensitivity σ of the controller, the configuration layer and application layer of the on-chip interconnect network, and the configuration layer and application layer of the PE processing unit according to the test program P5 and the test values required by the chip external input. 5 ;
[0013] (7) According to σ 1 , σ 2 , σ 3 , σ 4 , σ 5 , we get the controller module single particle functional error cross section σ CC , single-event functional error cross section σ of the on-chip interconnect network configuration layer CR , single-event functional error cross section σ of the on-chip interconnect network application layer AR, PE processing unit array configuration layer single particle function error cross section σ CPE , and PE processing unit array application layer single-particle functional error cross section σ APE .
[0014] Furthermore, in step (2), only the controller is irradiated, and the single particle sensitivity σ of the controller module is obtained according to the test procedure P1. 1 The steps are as follows:
[0015] (21) executing the configuration of the test program P1 on the on-chip interconnect network and the PE processing unit array through the controller;
[0016] (22) determining whether the code stream in the PE processing unit array and the on-chip interconnect network is consistent with the golden code stream, and if not consistent, recording a controller single particle function error as occurring, executing step (21) and recording the total number of controller single particle function errors;
[0017] (23) Steps (21) and (22) are executed repeatedly until the total number of controller single-particle function errors reaches 10 or the total ion injection reaches 1.0E+07 particles / cm 2 , then stop irradiation and the test ends;
[0018] (24) The ratio of the total number of single-particle functional errors of the controller to the total ion injection is calculated to obtain the single-particle sensitivity σ of the controller module. 1 .
[0019] Furthermore, in step (3), the whole chip is irradiated, and the configuration layer sensitivity σ of the controller and the on-chip interconnect network is obtained according to the test procedure P2. 2 The steps are:
[0020] (31) executing the configuration of the test program P2 on the on-chip interconnect network through the controller;
[0021] (32) determining whether the flag of the single-particle functional error is consistent with the on-chip interconnect network configuration code stream and the golden code stream, and if they are inconsistent, recording that a single-particle functional error has occurred in the controller and the on-chip interconnect network configuration layer, executing step (31) and recording the total number of single-particle functional errors;
[0022] (33) Repeat steps (31) and (32) until the total number of single-element functional errors reaches 10 or the total ion injection reaches 1.0E+07 particles / cm 2 , then stop irradiation and the test ends;
[0023] (34) The ratio of the total number of single-event functional errors to the total ion injection is calculated to obtain the configuration layer sensitivity σ of the controller and the on-chip interconnect network. 2 .
[0024] Furthermore, in step (4), the whole chip is irradiated, and the configuration layer and application layer sensitivity σ of the controller and the on-chip interconnection network are obtained according to the test program P3 and the test value required by the chip external input. 3 The steps are:
[0025] (41) executing the configuration of the test program P3 corresponding to the on-chip interconnect network through the controller;
[0026] (42) determining whether the configuration code stream of the configuration layer of the controller and the on-chip interconnect network is consistent with the golden code stream, and monitoring whether the application layer function of the on-chip interconnect network is correctly executed. If inconsistent or incorrect, it is recorded as a single-particle functional error of the configuration layer and application layer of the controller and the on-chip interconnect network, and step (41) is executed and the total number of single-particle functional errors is recorded;
[0027] (43) Repeat steps (41) and (42) until the total number of single-element functional errors reaches 10 or the total ion injection reaches 1.0E+07 particles / cm 2 , then stop irradiation and the test ends;
[0028] (44) The ratio of the total number of single-event functional errors to the total ion injection is calculated to obtain the configuration layer and application layer sensitivity σ of the controller and on-chip interconnect network. 3 .
[0029] Furthermore, in step (5), the whole chip is irradiated, and the sensitivity σ of the controller, the on-chip interconnection network configuration layer and the application layer, and the PE processing unit configuration layer is obtained according to the test program P4. 4 The steps are as follows;
[0030] (51) executing the configuration of the test program P4 on the on-chip interconnect network and the PE processing unit array through the controller;
[0031] (52) determining whether the configuration code stream in the controller, the on-chip interconnect network, and the PE processing unit array is consistent with the golden code stream, and if not consistent, recording a single particle functional error in the configuration layer of the controller, the on-chip interconnect network, and the PE processing unit array, executing step (51) and recording the total number of single particle functional errors;
[0032] (53) Repeat steps (51) and (52) until the total number of single-element functional errors reaches 10 or the total ion injection reaches 1.0E+07 particles / cm 2 , then stop irradiation and the test ends;
[0033] (54) The ratio of the total number of single-event functional errors to the total ion injection is calculated to obtain the configuration layer sensitivity σ of the controller, on-chip interconnect network, and PE processing unit. 4 .
[0034] Furthermore, in step (6), the whole chip is irradiated, and the sensitivity σ of the configuration layer and application layer of the controller, the on-chip interconnection network configuration layer and application layer, and the PE processing unit is obtained according to the test program P5 and the test value required by the chip external input. 5 The steps are as follows:
[0035] (61) executing the configuration of the test program P5 on the on-chip interconnect network and the PE processing unit array through the controller;
[0036] (62) determining whether the configuration information in the controller, the on-chip interconnect network and the PE processing unit in the configuration layer is consistent with the golden code stream, and monitoring whether the function execution or data operation of the on-chip interconnect network and the PE processing unit in the application layer is correct. If they are inconsistent or incorrect, it is recorded as a single-particle functional error in the controller, the on-chip interconnect network configuration layer and the application layer, and the configuration layer and the application layer of the PE processing unit, and executing step (61) and recording the total number of single-particle functional errors;
[0037] (63) Steps (61) and (62) are executed repeatedly until the total number of single-element functional errors reaches 10 or the total ion injection reaches 1.0E+07 particles / cm 2 , then stop irradiation and the test ends;
[0038] (64) The ratio of the total number of single-particle functional errors to the total ion injection is calculated to obtain the sensitivity σ of the configuration layer and application layer of the controller, the on-chip interconnect network, and the configuration layer and application layer of the PE processing unit. 5 .
[0039] Furthermore, in step (7), according to σ 1 , σ 2 , σ 3 , σ 4 , σ 5 , we get CC , σ CR , σ AR , σ CPE , and σ APE The method is:
[0040]
[0041] A single-particle functional error test system for a dynamically reconfigurable chip, used to implement the single-particle functional error test method for the dynamically reconfigurable chip, comprising a program-controlled power supply module, a host computer control center module and a control area module;
[0042] The program-controlled power supply module supplies power to the dynamically reconfigurable chip to be tested and the control area module according to the power control instruction input by the host computer control center module, obtains the current state of the dynamically reconfigurable chip to be tested and feeds it back to the host computer control center module;
[0043] The host computer control center module outputs power control instructions to the program-controlled power module, and monitors the current according to the current state input by the program-controlled power module; outputs test programs P1, P2, P3, P4 and P5 and test configuration code streams to the control area module;
[0044] The control area module receives the test programs P1, P2, P3, P4 and P5 and the test configuration code stream input by the upper computer control center module and outputs them to the dynamically reconfigurable chip to be tested, obtains the data obtained by testing the dynamically reconfigurable chip to be tested according to the test programs P1, P2, P3, P4 and P5 and the test configuration code stream, determines whether a single particle functional error occurs in the dynamically reconfigurable chip to be tested, and calculates the single particle functional error cross section of each module.
[0045] Compared with the prior art, the present invention has the following beneficial effects:
[0046] (1) The present invention provides a method for obtaining the single-particle functional error cross-section of different functional modules of a dynamically reconfigurable chip, which can respectively count the single-particle functional error cross-section of each independent functional module in the dynamically reconfigurable chip, intuitively reflect the single-particle sensitivity characteristics of each independent functional module, and has the characteristic of comprehensive coverage of single-particle effects;
[0047] (2) The single-particle functional error test method of the dynamically reconfigurable chip in the present invention accurately grasps the single-particle effect sensitivity ratio of different functional modules of the dynamically reconfigurable chip, providing an important reference for radiation hardening design;
[0048] (3) The single-particle functional error test system of the dynamically reconfigurable chip in the present invention provides a stable and reliable test environment for the single-particle functional error test method of the dynamically reconfigurable chip. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] Figure 1 A schematic diagram of a single-particle functional error test system for a dynamically reconfigurable chip according to the present invention;
[0050] Figure 2 A flowchart for obtaining sensitivity characteristics of different functional modules of a dynamically reconfigurable chip according to the present invention;
[0051] Figure 3 The present invention is a flow chart of single particle functional error testing of each functional module in the dynamically reconfigurable chip. DETAILED DESCRIPTION
[0052] The following detailed description of the present invention will make the features and advantages of the present invention more clear and explicit.
[0053] The word "exemplary" is used exclusively herein to mean "serving as an example, embodiment, or illustration." Any embodiment described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments. Although various aspects of the embodiments are shown in the drawings, the drawings are not necessarily drawn to scale unless otherwise noted.
[0054] In order to solve the problem that it is difficult to obtain the single-particle sensitivity of different functional modules in the existing single-particle functional test of dynamically reconfigurable chips, the present invention discloses a single-particle functional error testing method for a dynamically reconfigurable chip, writes corresponding test programs for different functional modules, and performs radiation tests according to the designed test scheme, so as to obtain the single-particle functional error cross-sections under different test programs, and jointly solve the single-particle functional error cross-sections of each independent functional module, thereby providing a reference for the single-particle radiation performance evaluation of the dynamically reconfigurable chip. The present invention also discloses a single-particle functional error testing system for a dynamically reconfigurable chip, comprising a programmable power supply module, a host computer control center module and a control area module, thereby providing a stable testing environment for the implementation of the single-particle functional error testing method for the dynamically reconfigurable chip.
[0055] The functional test of the chip to be tested is mainly divided into the processing unit array, the on-chip interconnection network, and the controller. The single-particle functional error test method of the dynamically reconfigurable chip of the present invention first designs the single-particle functional error test scheme for different functional modules respectively; then writes the corresponding test program for different functional modules, and performs radiation tests according to the functional error test scheme, and counts the number of single-particle functional errors of different functional modules in the process; finally, calculates the single-particle functional error cross-section of different functional modules. The present invention compares and analyzes the single-particle functional error cross-sections of different functional modules, and can obtain the sensitivity ratio of different functional modules of the dynamically reconfigurable chip, and perform system-level reinforcement for functional modules with high sensitivity ratio.
[0056] The present invention provides a single-particle functional error testing method for a dynamically reconfigurable chip, which specifically comprises the following steps:
[0057] (1) Design test procedures P1, P2, P3, P4 and P5; P1 is used to test the single-particle sensitivity σ of the controller module 1 , P2 is used to test the configuration layer sensitivity of the controller and the on-chip interconnect network σ 2 , P3 is used to test the configuration layer and application layer sensitivity of the controller and the on-chip interconnect network σ 3 , P4 is used to test the configuration layer sensitivity of the controller, the on-chip interconnect network configuration layer, and the PE processing unit array σ 4P5 is used to test the sensitivity of the controller, the configuration layer and application layer of the on-chip interconnect network, and the configuration layer and application layer of the PE processing unit. 5 ;
[0058] (2) The controller configures the on-chip interconnect network and the processing unit array to execute the test program P1, irradiating only the controller to obtain the single-particle sensitivity σ of the controller module. 1 ;
[0059] (3) The controller executes the configuration of the test program P2 on the on-chip interconnect network, and the full chip is irradiated to obtain the configuration layer sensitivity σ of the controller and the on-chip interconnect network 2 ;
[0060] (4) The controller configures the on-chip interconnection network to execute the test program P3 and provides the sampling period, specific values of the execution operation and other data through the external control area FPGA. The operation is performed according to the configuration of P3 and the input test values required. The configuration layer sensitivity and application layer sensitivity σ of the controller and the on-chip interconnection network are obtained by full chip irradiation. 3 ;
[0061] (5) The controller executes the configuration of the test program P4 on the on-chip interconnect network and the PE processing unit array, and the full chip is irradiated to obtain the configuration layer sensitivity of the controller, the on-chip interconnect network, and the configuration layer sensitivity of the PE processing unit σ 4 ;
[0062] (6) The controller configures the on-chip interconnect network and the PE processing unit to execute the test program P5, and provides the sampling period, specific values of the execution operation and other data through the external control area FPGA. The operation is performed according to the configuration of P5 and the input test required values. The sensitivity of the configuration layer and application layer of the controller, the on-chip interconnect network and the configuration layer and application layer of the PE processing unit array are obtained by full chip irradiation. 5 ;
[0063] (7) Analyze the above data and combine σ 1 , σ 2 , σ 3 , σ 4 , σ 5 , the independent single-particle sensitivity characteristics of different functional modules can be obtained.
[0064] Specifically, the dynamically reconfigurable chip can be divided into a configuration layer and an application layer. The configuration layer defines the logical functions of the design, including the configuration of the processing unit array and external data transmission and the configuration information index of the array computing task; the application layer runs the processing unit, including the storage unit used in the user design, which is used to save the running data in the design working state and the content of the predefined function lookup table.
[0065] Specifically, P1, P2, P3, P4, and P5 are test programs written for different functional test requirements. 1 , σ 2 , σ 3 , σ 4 , σ 5 The single particle functional error cross section obtained under different test procedures.
[0066] In steps (2) to (6), the single-particle functional error test process and judgment method are as follows: the single-particle functional error sensitivity is tested by a "configure->read->compare-configure->read->compare" loop execution method.
[0067] Furthermore, the implementation method in step (2) is:
[0068] 2.1 Write the test program P1. The controller configures the test information for the processing unit and the on-chip interconnection network. The control area module reads the control code stream in the processing unit and the on-chip interconnection network and compares it with the Golden code stream stored in the memory before irradiation. The results unanimously indicate that the controller functions correctly.
[0069] 2.2 Irradiate the controller area. During the test, execute step 2.1 repeatedly. The control area module statistically compares the code stream data in the processing unit and the on-chip interconnection network with the golden code stream data. If they are inconsistent, record a controller single particle function error and upload the error number to the host computer.
[0070] 2.3 The total number of controller single-particle function errors reaches 10 or the total ion injection reaches 1.0E+07 particles / cm 2 , then stop irradiation and the test ends.
[0071] 2.4 Statistics of single-particle functional error cross section σ of test function modules under P1 test procedure 1 , the ratio of the total number of single event functional errors to the irradiation dose, σ 1 =σ CC , where σ CC Error cross section for single particle functions of the controller module.
[0072] Furthermore, the implementation method of step (3) is:
[0073] On-chip interconnection realizes the connection between each processing module and interface through a router, and realizes data transmission and control between each module.
[0074] 3.1 Write the test program P2 and configure the P2 test information on the on-chip interconnect network through the controller;
[0075] 3.2 Irradiate the entire chip, monitor the on-chip interconnection network configuration code stream and compare and analyze it with the golden code stream. If they are consistent, execute step 3.3. If they are inconsistent, record a single-particle functional error and reconfigure the on-chip interconnection network with P2 test information through the controller;
[0076] 3.3 The number of single-event functional errors reaches 10 or the total ion injection reaches 1.0E+07 particles / cm 2 , then stop irradiation and the test ends.
[0077] 3.4 Statistics of the single-particle functional error cross section σ of the test function module under the P2 test program 2 , the single-event functional error cross section is the ratio of the total number of single-event functional errors to the irradiation dose. σ 2 =σ CC +σ CR , where σ CR Configuration layer single event functional error cross sections for on-chip interconnect networks;
[0078] Furthermore, the implementation method of step (4) is:
[0079] 4.1 write the test program P3, and configure the test program P3 on the on-chip interconnection network through the controller;
[0080] 4.2 The sampling period of the on-chip interconnect network, the specific numerical value of the execution operation and other data input are performed through FPGA, and the operation is performed according to the configuration of P3 and the numerical value required for the input test. The whole chip is irradiated, and the functional errors of the configuration layer controller, the on-chip interconnect network and the on-chip interconnect network of the application layer are monitored during the irradiation process;
[0081] Functional error: Any inconsistency with normal function is recorded as a functional error.
[0082] 4.3 The number of single-event functional errors reaches 10 or the total ion injection reaches 1.0E+07 particles / cm 2 , then stop irradiation and the test ends.
[0083] 4.4 Statistics of single-particle functional error cross section σ of test function modules under P3 test program 3 , the single-event functional error cross section is the ratio of the total number of single-event functional errors to the irradiation dose. σ 3 =σ CC +σ CR +σ AR , where σ AR Application layer single event functional error cross sections for on-chip interconnect networks;
[0084] Furthermore, the implementation method of step (5) is:
[0085] 5.1 Write the test program P4 and configure the P4 test information for the on-chip interconnect network and processing unit array through the controller;
[0086] 5.2 Irradiate the entire chip, monitor the configuration code stream in the on-chip interconnect network and the processing unit array, and compare and analyze it with the golden code stream. If they are consistent, execute step 5.3. If they are inconsistent, record a single-particle functional error and return to step 5.1.
[0087] 5.3 The number of single-element functional errors reaches 10 or the total ion injection reaches 1.0E+07 particles / cm 2 , then stop irradiation and the test ends.
[0088] 5.4 Statistics of single-particle functional error cross section σ of test function modules under P4 test program 4 , the single-event functional error cross section is the ratio of the total number of single-event functional errors to the irradiation dose. σ 4 =σ CC +σ CR +σ CPE , where σ CPE Configuration layer single particle function error cross section for PE processing unit array;
[0089] Furthermore, the implementation method of step (6) is:
[0090] 6.1 Write the test program P5 and configure the P5 test information on the on-chip interconnect network and the processing unit array through the controller;
[0091] 6.2 The control area module FPGA is used to input the sampling period, specific values of the operation, and other data of the on-chip interconnect network and the processing unit array, and the operation is performed according to the configuration of P5 and the input test values required. The whole chip is irradiated, and the functional errors of the configuration layer controller, the on-chip interconnect network and the processing unit array and the single-particle functional errors of the on-chip interconnect network and the processing unit array module of the application layer are monitored during the irradiation process;
[0092] 6.3 The number of single-event functional errors reaches 10 or the total ion injection reaches 1.0E+07 particles / cm 2 , then stop irradiation and the test ends.
[0093] 6.4 Statistics of single-particle functional error cross section σ of test function modules under P5 test program 5 , the single-event functional error cross section is the ratio of the total number of single-event functional errors to the irradiation dose. σ 5 =σ CC +σ CR +σ CPE +σ APE +σ AR , where σAPE Applying layer single particle function error cross section for processing cell array module;
[0094] Furthermore, the implementation method of step (7) is:
[0095]
[0096] Simultaneous σ 1 , σ 2 , σ 3 , σ 4 , σ 5 The equation system is solved to get σ CC , σ CR , σ AR , σ CPE , σ APE The value size can intuitively reflect the single-particle sensitivity of each functional module.
[0097] Example 1
[0098] like Figure 1 As shown, an embodiment of the present invention provides a single-particle functional error test system for a dynamically reconfigurable chip, including a programmable power supply module, a host computer control center module, and a control area module:
[0099] Programmable power supply module: The power supply and current monitoring of the dynamically reconfigurable chip to be tested are completed by a high-precision programmable power supply. The collected current data is uploaded to the host computer through the network port and stored.
[0100] The host computer control center module is controlled by the host computer software and has functions such as configuration code stream, readback, power supply control, current monitoring, and interrupt monitoring.
[0101] The control area module includes the bit stream storage area, PROM and main control FPGA, etc. It is responsible for configuring the dynamically reconfigurable chip to be tested, SEFI monitoring, reading back, uploading error counts and other operations.
[0102] The dynamically reconfigurable chip to be tested includes a controller module, an on-chip interconnection network and a PE processing unit array. The on-chip interconnection network and the PE processing unit array can be divided into a configuration layer and an application layer. The configuration layer defines the logical functions of the design, including the configuration of the PE processing unit array and the external data transmission and the configuration code stream index of the array computing task, the connection relationship between the modules of the on-chip interconnection network, etc. The application layer operation processing unit includes a storage unit used in the user design, which is used to save the operation data under the design working state and the content of the predefined function lookup table. The on-chip interconnection network of the application layer realizes data transmission between various modules. In the preferred scheme, the dynamically reconfigurable chip to be tested also includes a memory.
[0103] Controller module: The controller of the reconfigurable chip selects different functions of the ALU, selects different inputs of the MUX, and selects to store in different registers through configuration information, thereby realizing the functional reconstruction of a single PE, and further realizing the functional reconstruction of the PEA (processing unit array). In the existing architecture, it is not only responsible for the management of the configuration flow, including the parsing and scheduling of configuration information; it is also responsible for the management of the data path, that is, generating control codes according to the configuration information, and controlling the key system states of the data path, such as the timing function of array operations and resource scheduling.
[0104] PE Processing Element Array (PEA): PE is usually composed of an arithmetic logic unit (ALU), multiple multiplexers (MUX), and input and output registers. The core design parameters mainly include: PE granularity, PEA homogeneous and heterogeneous modes, PEA configuration mode and configuration depth, PEA topology and routing strategy, PEA interface technology, and PEA computing model.
[0105] On-chip interconnect network: The chip has many reconfigurable processing units (PEs) of different granularities, which form a huge "computing pool" through the on-chip interconnect network (i.e., routing network), which can perform various types of logic, arithmetic operations and logic control in parallel; at the same time, the dynamically reconfigurable chip can also dynamically change the interconnection relationship between PEs, control the flow of data, and further improve the data throughput of the operation.
[0106] Before irradiation, the host computer control center module communicates with the controller module of the dynamically reconfigurable chip to be tested through the RS485 interface. The control FPGA in the controller module is responsible for downloading the test configuration code stream to the dynamically reconfigurable chip to be tested. The dynamically reconfigurable chip is first configured according to the test configuration code stream before irradiation, and the configurable resources are initialized. The configurable resources include processing unit functions, interconnection lines and other resources. During the irradiation process, the IO current and core current of the test dynamic reconfigurable chip are monitored in real time through the programmable power supply module, and then the working status of the test dynamic reconfigurable chip is monitored. The sampling period is set through the host computer control center module. The control area module reads back the code stream generated by the dynamically reconfigurable chip to be tested after irradiation in real time according to the set sampling period, and counts the number of functional errors of different functional test modules of the dynamic reconfigurable chip, and transmits these data back to the host computer control center module through the communication interface.
[0107] like Figure 2As shown, the present invention provides a single-particle functional error test method for different functional modules of a dynamically reconfigurable chip. Different test programs are designed according to different computing tasks so as to reconfigure the reconfigurable logic device during the task execution process. The difference between them lies in the resource allocation strategy and reconstruction method of each test program. During the execution process, the device configuration files corresponding to each stage are sequentially configured to the reconfigurable logic device in chronological order, and the computing task is finally completed. The single-particle functional error test of different functional modules of the dynamically reconfigurable circuit is completed under different test programs to obtain the single-particle functional error cross section.
[0108] Figure 3 The single-particle functional error testing method of the dynamically reconfigurable chip of the present invention is as follows:
[0109] Step 1: First, select particles and fluence rate;
[0110] Step 2: Run the test programs P1, P2, P3, P4 and P5 to configure the dynamically reconfigurable chip to be tested and check whether the functions are normal;
[0111] Step 3: Start particle irradiation;
[0112] Step 4: During the irradiation process, the functional status of different functional modules of the chip is monitored. If a single particle functional error occurs in the sample, a single particle functional error is recorded and the process returns to step 2. If not, the process proceeds to step 5.
[0113] Step 5: Continue until the total number of single-event functional errors reaches 10 or the total single-event dose reaches 1.0E+07 / cm 2 , stop irradiation and the test ends;
[0114] This embodiment has formulated single-particle functional error test methods for several main functional modules of this dynamically reconfigurable chip. The test system has a high degree of integration. In terms of software, it can complete the related work of single-particle effect test with one-click operation. This test system can not only flexibly start, pause, and terminate the test, but also display the current state, readback error number, interruption number, and can save data and automatically generate test record files. Compared with the previous test system, this test system saves test time, is more intelligent and complete, and avoids some losses caused by erroneous operations.
[0115] The present invention has been described in detail above in conjunction with specific implementations and exemplary examples, but these descriptions cannot be understood as limiting the present invention. Those skilled in the art understand that, without departing from the spirit and scope of the present invention, a variety of equivalent substitutions, modifications or improvements may be made to the technical solution of the present invention and its implementation methods, all of which fall within the scope of the present invention. The scope of protection of the present invention shall be subject to the attached claims.
[0116] The contents not described in detail in the specification of the present invention belong to the common knowledge of those skilled in the art.
Claims
1. A single-particle functional error test method for a dynamically reconfigurable chip, the dynamically reconfigurable chip comprising a PE processing unit array, an on-chip interconnect network and a controller, It is characterized in that The following steps are involved: (1) Only the controller is irradiated to obtain the single particle sensitivity σ of the controller module 1 ; (2) Irradiate the entire chip to obtain the configuration layer sensitivity σ of the controller and on-chip interconnect network 2 ; (3) Irradiate the entire chip to obtain the configuration layer and application layer sensitivity σ of the controller and on-chip interconnect network 3 ; (4) Irradiate the entire chip to obtain the configuration layer sensitivity σ of the controller, on-chip interconnect network configuration layer, and PE processing unit array 4 ; (5) Irradiate the entire chip to obtain the sensitivity σ of the controller, the configuration layer and application layer of the on-chip interconnect network, and the configuration layer and application layer of the PE processing unit 5 ; (6) According to σ 1 , σ 2 , σ 3 , σ 4 , σ 5 , we get the controller module single particle functional error cross section σ CC , single-event functional error cross section σ of the on-chip interconnect network configuration layer CR , single-event functional error cross section σ of the on-chip interconnect network application layer AR , PE processing unit array configuration layer single particle function error cross section σ CPE , and PE processing unit array application layer single-particle functional error cross section σ APE ; In step (1), only the controller is irradiated, and the single particle sensitivity σ of the controller module is obtained according to the test procedure P1. 1 , the steps are as follows: (11) executing the configuration of the test program P1 on the on-chip interconnect network and the PE processing unit array through the controller; (12) determining whether the code stream in the PE processing unit array and the on-chip interconnect network is consistent with the golden code stream, and if not, recording a controller single particle function error as occurring, executing step (11) and recording the total number of controller single particle function errors; (13) Execute steps (11) and (12) repeatedly until the total number of controller single-particle function errors reaches 10 or the total ion injection reaches 1.0E+07 particles / cm 2 , then stop irradiation and the test ends; (14) Calculate the ratio of the total number of single-particle functional errors of the controller to the total ion injection to obtain the single-particle sensitivity σ of the controller module 1 ; In step (2), the whole chip is irradiated, and the configuration layer sensitivity σ of the controller and the on-chip interconnect network is obtained according to the test procedure P2. 2 , the steps are as follows: (21) executing the configuration of the test program P2 on the on-chip interconnect network through the controller; (22) determining whether the flag of the single-particle functional error is consistent with the on-chip interconnect network configuration code stream and the golden code stream, and if they are inconsistent, recording that a single-particle functional error has occurred in the controller and the on-chip interconnect network configuration layer, executing step (21) and recording the total number of single-particle functional errors; (23) Repeat steps (21) and (22) until the total number of single-element functional errors reaches 10 or the total ion injection reaches 1.0E+07 particles / cm 2 , then stop irradiation and the test ends; (24) The ratio of the total number of single-event functional errors to the total ion injection is calculated to obtain the configuration layer sensitivity σ of the controller and the on-chip interconnect network. 2 ; In step (3), the whole chip is irradiated, and the configuration layer and application layer sensitivity σ of the controller and the on-chip interconnection network are obtained according to the test program P3 and the test value required by the chip external input. 3 , the steps are as follows: (31) executing the configuration of the test program P3 on the on-chip interconnect network through the controller; (32) determining whether the configuration code stream of the configuration layer of the controller and the on-chip interconnect network is consistent with the golden code stream, and monitoring whether the application layer function of the on-chip interconnect network is correctly executed. If inconsistent or incorrect, it is recorded as a single-particle functional error of the configuration layer and application layer of the controller and the on-chip interconnect network, and step (31) is executed and the total number of single-particle functional errors is recorded; (33) Repeat steps (31) and (32) until the total number of single-element functional errors reaches 10 or the total ion injection reaches 1.0E+07 particles / cm 2 , then stop irradiation and the test ends; (34) The ratio of the total number of single-event functional errors to the total ion injection is calculated to obtain the configuration layer and application layer sensitivity σ of the controller and on-chip interconnect network. 3 ; In step (4), the whole chip is irradiated, and the sensitivity σ of the controller, the on-chip interconnection network configuration layer and the application layer, and the PE processing unit configuration layer is obtained according to the test program P4. 4 , the steps are as follows: (41) executing the configuration of the test program P4 on the on-chip interconnect network and the PE processing unit array through the controller; (42) determining whether the configuration code stream in the controller, the on-chip interconnect network, and the PE processing unit array is consistent with the golden code stream. If not consistent, it is recorded as a single-particle functional error in the configuration layer of the controller, the on-chip interconnect network, and the PE processing unit array. Execute step (41) and record the total number of single-particle functional errors. (43) Repeat steps (41) and (42) until the total number of single-element functional errors reaches 10 or the total ion injection reaches 1.0E+07 particles / cm 2 , then stop irradiation and the test ends; (44) The ratio of the total number of single-event functional errors to the total ion injection is calculated to obtain the configuration layer sensitivity σ of the controller, on-chip interconnect network, and PE processing unit. 4 ; In step (5), the whole chip is irradiated, and the sensitivity σ of the controller, the configuration layer and application layer of the on-chip interconnect network, and the configuration layer and application layer of the PE processing unit are obtained according to the test program P5 and the test value required by the chip external input. 5 , the steps are as follows: (51) executing the configuration of the test program P5 on the on-chip interconnect network and the PE processing unit array through the controller; (52) determining whether the configuration information in the controller, the on-chip interconnect network and the PE processing unit in the configuration layer is consistent with the golden code stream, and monitoring whether the function execution or data operation of the on-chip interconnect network and the PE processing unit in the application layer is correct. If they are inconsistent or incorrect, it is recorded as a single-particle functional error in the controller, the on-chip interconnect network configuration layer and the application layer, and the configuration layer and the application layer of the PE processing unit, and executing step (51) and recording the total number of single-particle functional errors; (53) Repeat steps (51) and (52) until the total number of single-element functional errors reaches 10 or the total ion injection reaches 1.0E+07 particles / cm 2 , then stop irradiation and the test ends; (54) The ratio of the total number of single-particle functional errors to the total ion injection is calculated to obtain the sensitivity σ of the configuration layer and application layer of the controller, the on-chip interconnect network, and the configuration layer and application layer of the PE processing unit. 5 .
2. A single-event functional error testing method for a dynamically reconfigurable chip according to claim 1, It is characterized in that In step (6), according to σ 1 , σ 2 , σ 3 , σ 4 , σ 5 , we get CC , σ CR , σ AR , σ CPE , and σ APE The method is:
3. A single-event functional error test system for dynamically reconfigurable chips, It is characterized in that A single-particle functional error test method for implementing the dynamically reconfigurable chip according to claim 1 or 2, comprising a programmable power supply module, a host computer control center module and a control area module; The program-controlled power supply module supplies power to the dynamically reconfigurable chip to be tested and the control area module according to the power control instruction input by the host computer control center module, obtains the current state of the dynamically reconfigurable chip to be tested and feeds it back to the host computer control center module; The host computer control center module outputs power control instructions to the programmable power module, and monitors the current according to the current state input by the programmable power module; outputs test programs P1, P2, P3, P4 and P5 and test configuration code stream to the control area module; P1 is used to test the single particle sensitivity σ of the controller module 1 , P2 is used to test the configuration layer sensitivity of the controller and the on-chip interconnect network σ 2 , P3 is used to test the configuration layer and application layer sensitivity of the controller and the on-chip interconnect network σ 3 , P4 is used to test the configuration layer sensitivity of the controller, the on-chip interconnect network configuration layer, and the PE processing unit array σ 4 P5 is used to test the sensitivity of the controller, the configuration layer and application layer of the on-chip interconnect network, and the configuration layer and application layer of the PE processing unit. 5 ; The control area module receives the test programs P1, P2, P3, P4 and P5 and the test configuration code stream input by the upper computer control center module and outputs them to the dynamically reconfigurable chip to be tested, obtains the data obtained by testing the dynamically reconfigurable chip to be tested according to the test programs P1, P2, P3, P4 and P5 and the test configuration code stream, determines whether a single particle functional error occurs in the dynamically reconfigurable chip to be tested, and calculates the single particle functional error cross section of each module.
Citation Information
Patent Citations
Dynamic reconfigurable subnetting method and system based on network on chip
CN102799560A
SRAM type FPGA single particle irradiation test system and method
CN103744014A
