DDR pressure measurement device and DDR pressure measurement method

By designing the DDR pressure measurement device, using the embedded access generation engine unit and the data comparison logic unit for high bandwidth read and write access, the problems of low DDR pressure measurement efficiency and complex bit error rate detection are solved, and full coverage of the DDR memory array is achieved The bit error rate statistics and automatic correction of dynamically optimize link quality.

CN119068962APending Publication Date: 2024-12-03WUXI STARS MICRO SYSTEM TECHNOLOGIES CO LTD
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411254106.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-06
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

DDR pressure measurement is low, has limited coverage of memory space, complex bit error rate detection and cannot be automatically corrected.

Method used

A DDR pressure measurement device is designed, including a DDR controller and a DDR PHY, with embedded access generation engine unit, test code generation logic unit and data comparison logic unit, which can generate high bandwidth read and write access and cover the DDR storage array. All storage space and the data sampling clock delay and discriminative voltage value are automatically adjusted by parameter correction logic unit to optimize link quality.

Benefits of technology

It significantly improves the efficiency and accuracy of DDR pressure measurement, can cover all storage space of the DDR storage array, provides real-time bit error rate statistics, and supports automatic correction and dynamic optimization of link quality.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119068962A_ABST
    Figure CN119068962A_ABST
Patent Text Reader

Abstract

The invention provides a DDR (Double Data Rate) voltage measurement device and a DDR voltage measurement method. The DDR voltage measurement device comprises a DDR controller and a DDR PHY (Physical Layer), wherein an access generation engine unit, a test code pattern generation logic unit and a data comparison logic unit are embedded in the DDR controller; the access generation engine unit is used for generating write access to a configuration address range after the DDR pressure measurement process is started; the configuration address range is any storage address range in a pre-configured DDR storage array; the test code pattern generation logic unit generates a test code pattern after the DDR pressure test process is started, and stores the test code pattern in a write data queue of the access generation engine unit; and after the write access is completed, the access generation engine generates read access to the configuration address range, and the data comparison logic unit intercepts return data corresponding to the read access, performs data verification by using the return data and calculates an error rate.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application belongs to the technical field of DDR stress testing, and in particular, relates to a DDR stress testing device and a DDR stress testing method. Background Art

[0002] DDR (double data rate sdram) stress testing is an important means to test the reliability of the link of SOC (System on chip) accessing DDR, that is, the SOC constructs a test pattern, initiates a large number of read and write accesses to DDR, collects and analyzes the bit error rate, and finally obtains the link quality information. DDR stress testing generally requires the stress testing device to have the following characteristics: 1) It can generate high-bandwidth read and write data streams to ensure the efficiency of stress testing; 2) The stress testing range can cover all storage spaces of DDR particles; 3) The test pattern (also called test pattern) must be diverse; 4) The bit error rate statistics must meet the accuracy and real-time requirements. Summary of the invention

[0003] The purpose of the present application is to provide a DDR stress testing device and a DDR stress testing method, aiming to solve the problems existing in the related art such as low DDR stress testing efficiency, limited coverage of storage space, complex bit error rate detection, and inability to automatically correct.

[0004] According to a first aspect of the present application, a DDR stress test device is provided, comprising: a DDR controller and a DDR PHY; wherein the DDR controller has an embedded access generation engine unit, a test pattern generation logic unit and a data comparison logic unit; the DDR PHY is an interface between the DDR controller and a DDR storage array;

[0005] The access generation engine unit is used to generate a write access to a configuration address range after the DDR stress test process is started; the configuration address range is any storage address range in the pre-configured DDR storage array;

[0006] The test pattern generation logic unit is used to generate a test pattern after the DDR stress test process is started, and store the test pattern in the write data queue of the access generation engine unit;

[0007] After the write access is completed, the access generation engine generates a read access to the configuration address range, and the data comparison logic unit intercepts the return data corresponding to the read access from the data stream between the DDR controller and the DDR PHY, and uses the return data to perform data verification and calculate the bit error rate.

[0008] In an optional implementation, the DDR stress testing device further includes: a parameter correction logic unit; wherein,

[0009] The parameter correction logic unit receives the bit error rate from the data comparison logic unit, adjusts the data sampling clock delay and the data sampling discrimination voltage value of the DDR PHY, and determines a target parameter combination according to the bit error rate; the target parameter combination includes the data sampling clock delay and the data sampling discrimination voltage value after correcting the bit error rate.

[0010] In an optional implementation, the access generation engine unit has one or more combinations of the following functions:

[0011] The configuration address range for write access and read access can be preconfigured, and the configuration address range can cover part or all of the storage space range of the DDR storage array;

[0012] Can be configured as write mode, read mode or write-first-read-later mode;

[0013] The burst length for read access and write access to the DDR memory array can be selected from a variety of candidate lengths;

[0014] Priorities of read access and write access can be configured to adjust the waiting time when passing through the arbiter of the DDR controller by changing the priorities of the read access and write access;

[0015] The number of cycles of read access and write access to the configuration address range can be configured;

[0016] After the task is executed, the interrupt can be reported.

[0017] In an optional implementation, the test pattern generation logic unit is configured to implement a combination of one or more of the following functions:

[0018] Generate PRBS code;

[0019] Generate an increasing code or a decreasing code;

[0020] Pre-configured custom encoding via software.

[0021] In an optional embodiment, the data comparison logic unit is configured to implement one or more combinations of the following functions:

[0022] intercepting the return data from the data stream of the DFI interface according to the configured address range; the DFI interface is an interface between the DDR controller and the DDR PHY;

[0023] According to the transmission code type, select PRBS code, increasing code or decreasing code, or custom code to obtain the test code type;

[0024] Comparing the intercepted return data with expected data by a comparator; the expected data includes the test pattern;

[0025] The bit error rate is counted in real time and stored in a register, and the bit error rate is fed back to the parameter correction logic unit of the DDR stress test device.

[0026] In an optional embodiment, the parameter correction logic unit determines the minimum delay and maximum delay with a bit error rate of zero by adjusting the data sampling clock delay, and determines the center delay based on the average value of the minimum delay and the maximum delay; then, multiple candidate delays are taken in the positive and negative directions of the center delay at a fixed step, and for each of the candidate delays, the data sampling judgment voltage value is adjusted to obtain the maximum voltage value and the minimum voltage value with a bit error rate of zero corresponding to each candidate delay, and then a two-dimensional matrix including candidate delays and candidate voltage values ​​is obtained, and the matrix element with the largest margin of the candidate delay and the candidate voltage value is determined from the two-dimensional matrix, and the candidate delay and the candidate voltage value corresponding to the matrix element with the largest margin are respectively used as the data sampling clock delay and the data sampling judgment voltage value in the target parameter combination.

[0027] In an optional implementation, the access generation engine unit is at the same level as other application layer interfaces in the DDR controller and is presented to the arbiter of the DDR controller in the form of a common port. The priority of the access generation engine unit can be configured.

[0028] In an optional implementation manner, the data comparison logic unit verifies the return data on each data transmission link between the DDR controller and the DDR PHY to obtain the bit error rate.

[0029] According to a second aspect of the present application, a DDR stress testing method is provided. The method uses the DDR stress testing device according to the first aspect to perform DDR stress testing, including:

[0030] After the DDR stress test process is started, the access generation engine unit constructs the write access of the configuration address range of the DDR storage array;

[0031] In response to the start of the DDR stress test process, the test pattern generation logic unit generates a test pattern and writes the test pattern into the write data queue of the access generation engine unit;

[0032] After the write access is completed, the access generation engine generates a read access to the configuration address range;

[0033] The data comparison logic unit intercepts the return data corresponding to the read access from the data stream between the DDR controller and the DDR PHY, performs data verification using the return data, and calculates a bit error rate.

[0034] In an optional embodiment, the method further comprises:

[0035] If the data comparison logic unit detects a bit error and the bit error rate is less than a set threshold value, determining whether a parameter correction enable signal of the parameter correction logic unit of the DDR stress test device is turned on;

[0036] If the parameter correction enable signal is turned on, the data comparison logic unit reports the bit error rate to the parameter correction logic unit;

[0037] The parameter correction logic unit adjusts the data sampling clock delay and the data sampling discrimination voltage value of the DDR PHY, and determines a target parameter combination according to the bit error rate; the target parameter combination includes the data sampling clock delay and the data sampling discrimination voltage value after correcting the bit error rate.

[0038] In an optional implementation manner, the parameter correction logic unit adjusts the data sampling clock delay and the data sampling discrimination voltage value of the DDR PHY, and determines a target parameter combination according to the bit error rate, including:

[0039] The parameter correction logic unit determines the minimum delay and the maximum delay for a zero bit error rate by adjusting the data sampling clock delay, and determines the center delay based on the average value of the minimum delay and the maximum delay;

[0040] Taking multiple candidate delays in both positive and negative directions of the central delay at a fixed step;

[0041] For each candidate delay, adjusting the data sampling discrimination voltage value, obtaining a maximum voltage value and a minimum voltage value corresponding to each candidate delay with a zero bit error rate, and obtaining a two-dimensional matrix including candidate delays and candidate voltage values;

[0042] Determine from the two-dimensional matrix the matrix element with the largest margin for the candidate delay and candidate voltage value;

[0043] The candidate delay and candidate voltage values ​​corresponding to the matrix element with the largest margin are used as the data sampling clock delay and the data sampling determination voltage value in the target parameter combination, respectively.

[0044] Compared with the related art, the technical solution of this application has the following advantages:

[0045] The DDR stress test device in the embodiment of the present application generates test patterns by hardware, initiates read and write access to DDR, and performs data comparison and error statistics, while the software only needs to configure the mode and trigger execution, and feeds back the stress test results to the system by interruption, so that the DDR stress test process is simple. In addition, the stress test range in the embodiment of the present application can cover all storage spaces of the DDR storage array, and can give statistical results of the bit error rate for each data sampling link of the DDR storage array, making the stress test more complete. The access generation engine unit embedded in the DDR controller can generate read and write access of continuous addresses, reduce the DDRpage switching frequency, and access does not pass through the on-chip bus, but is directly hung on the controller internal bus, which can make the bandwidth utilization of DDR reach the highest, and can significantly improve the stress test efficiency. For the access of other application-side services, the read and write access pressure to DDR generated by the embodiment of the present application can be regarded as background traffic, which can be used for performance evaluation of normal services. The startup of the DDR stress test device will not affect normal business access, and with the parameter correction logic unit, the link quality can be dynamically optimized during the working process.

[0046] Other features and advantages of the present application will be described in the following description, and partly become apparent from the description, or be understood by practicing the present application. The purpose and other advantages of the present application can be realized and obtained through the structures and processes indicated in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. It is obvious that the drawings described below are certain embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0048] Figure 1 It is a structural block diagram of a DDR stress testing device according to the related art.

[0049] Figure 2 It is a schematic diagram of the structural framework of a DDR stress testing device according to an exemplary embodiment of the present application.

[0050] Figure 3 It is a schematic diagram of a structural implementation of a DDR stress testing device according to an exemplary embodiment of the present application.

[0051] Figure 4 4 is a test pattern generation logic diagram according to an exemplary embodiment of the present application.

[0052] Figure 5 It is a schematic diagram of delay and voltage parameter adjustment of DDR PHY according to an exemplary embodiment of the present application.

[0053] Figure 6 4 is a schematic diagram showing the delay and voltage parameter adjustment targets of the DDR PHY according to an exemplary embodiment of the present application.

[0054] Figure 7 It is a schematic diagram of a process implementation of delay and voltage parameter adjustment of DDR PHY according to an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0056] The common DDR stress test method in the related art is to construct a test pattern through the CPU, start the on-chip DMA to initiate read and write access to the DDR, and then use software to compare the read and write data to obtain the bit error rate, or obtain the probability of link error through the statistical value of the ECC and CRC check errors inside the DDR controller. Figure 1 As shown in the figure, the test process is as follows: 1) CPU controls PG (pattern generator) to construct a test pattern and cache it in on-chip SRAM (Static Random-Access Memory); 2) DMA (Direct Memory Access) channel writes the test pattern into DDR; 3) DMA reads data from DDR; CPU performs data comparison and bit error rate statistics.

[0057] Figure 1The above-mentioned related technologies shown have the following problems: 1) The stress test efficiency is low. The read and write accesses generated by the DMA used for stress testing need to pass through the on-chip bus to reach the entrance of the DDR controller. It faces competition for bus resources with other on-chip services, making it difficult for DDR to run at the limit bandwidth during the stress test, which makes the stress test cycle longer; 2) The stress test coverage is limited. Due to the division of the system address space, if the DMA is not dedicated to DDR stress testing, it is difficult for the stress test space to cover all storage units of DDR particles; 3) Error detection is complex and inefficient. The method of error detection is usually to compare read data and write data under the control of the CPU. This method has a long data path, is prone to introduce errors in other links, is not accurate enough, and requires CPU resources; Another method of error detection is to calculate the bit error rate based on the number of ECC errors or CRC errors counted by the hardware. This method is also difficult to obtain an accurate bit error rate; 4) It does not support automatic correction. The traditional method can only obtain the bit error rate within the stress test cycle, and cannot obtain the bit error rate of each DQ of the DDR The bit error rate of each lane cannot be automatically corrected for the link quality of each data lane; 5) Stress testing occupies a large amount of software and hardware resources. Traditional stress testing methods require the use of hardware resources such as CPU, on-chip general DMA, on-chip bus, etc., and software cooperation is required to refresh the bit error rate. The software and hardware overhead is large and affects the operation of other services.

[0058] Based on the above analysis, the present application exemplarily proposes a DDR stress test device and a DDR stress test method, wherein the DDR stress test device generates a test pattern (test pattern) by hardware, initiates read and write access to DDR, and performs data comparison and error statistics, while the software only needs to configure the mode and trigger execution, and feeds back the stress test results to the system by interruption, so that the DDR stress test process is simple. In addition, the stress test range in the present application can cover all storage spaces of the DDR storage array, and the statistical results of the bit error rate can be given for each DQ lane (data sampling link) of the DDR storage array, making the stress test more complete. The access generation engine unit embedded in the DDR controller can generate read and write access of continuous addresses, reduce the DDR page switching frequency, and access does not pass through the on-chip bus, but is directly hung on the controller internal bus, which can make the bandwidth utilization of DDR reach the highest, and can significantly improve the stress test efficiency. For the access of other application-side services, the read and write access pressure to DDR generated by the present application can be regarded as background traffic, which can be used for performance evaluation of normal services. The startup of the stress testing device will not affect normal business access. In conjunction with the parameter correction logic unit, the link quality can be dynamically optimized during the working process.

[0059] For example, Figure 2 Schematic diagram of the structural framework of a DDR stress test device according to an exemplary embodiment of the present application. Figure 2As shown, the DDR stress test device includes a DDR controller and a DDR PHY; the DDR controller is embedded with an access generation engine unit, a test pattern generation logic unit, and a data comparison logic unit; the DDR PHY is an interface between the DDR controller and the DDR storage array;

[0060] The access generation engine unit is used to generate a write access to a configuration address range after the DDR stress test process is started; the configuration address range is any storage address range in the pre-configured DDR storage array;

[0061] The test pattern generation logic unit is used to generate a test pattern after the DDR stress test process is started, and store the test pattern in the write data queue of the access generation engine unit;

[0062] After the write access is completed, the access generation engine generates a read access to the configuration address range, and the data comparison logic unit intercepts the return data corresponding to the read access from the data stream between the DDR controller and the DDR PHY, and uses the return data to perform data verification and calculate the bit error rate.

[0063] Exemplarily, after the DDR stress test process is started, the access generation engine unit can be triggered to generate a write access and the test pattern generation logic unit can be triggered to generate a test model at the same time.

[0064] In an optional embodiment of the present application, the DDR stress testing device further includes: a parameter correction logic unit; wherein,

[0065] The parameter correction logic unit receives the bit error rate from the data comparison logic unit, adjusts the data sampling clock delay and the data sampling discrimination voltage value of the DDR PHY, and determines a target parameter combination according to the bit error rate; the target parameter combination includes the data sampling clock delay and the data sampling discrimination voltage value after correcting the bit error rate.

[0066] For example, Figure 3 FIG. 1 is a schematic diagram of a structure implementation of a DDR stress test device according to an exemplary embodiment of the present application. Figure 3 As shown, in this embodiment, the access generation engine unit ⑥, the test pattern generation logic unit ⑦, and the data comparison logic unit ⑧ are embedded into a general DDR controller, and a parameter correction logic unit ⑨ is added. The parameter correction logic unit ⑨ automatically adjusts the clock delay of DQS (data sampling synchronization signal) and the value of the discrimination voltage value of DQ (data sampling) in the DDR PHY through the bit error rate of each DQ 1aane (data sampling transmission link) of the DDR fed back by the data comparison logic unit ⑧, so as to achieve the purpose of optimizing the link quality.

[0067] like Figure 3 As shown, application bus interfaces ①, ②, arbitrator ③, DDR controller core ④, DFI interface ⑤, and DDRPHY ⑩ are common components of the DDR controller and PHY, and the access generation engine unit ⑥, test pattern generation logic unit ⑦, data comparison logic unit ⑧, and parameter correction logic unit ⑨ are newly added components in this embodiment.

[0068] Among them, application bus interfaces ① and ② provide interfaces between the DDR controller and the application layer master, which are used to process the interface bus protocol, cache read and write commands and write data, and return read data and write responses. The number of application bus interfaces ① and ② can be selected according to actual conditions. Arbitrator ③ is a port arbitrator, which provides port arbitration based on priority or delay sensitivity, arbitrates the read and write accesses issued by each port and outputs them to the internal bus; DDR controller core ④ is responsible for the scheduling and system control of DDR commands, and the data comparison logic unit is embedded in the DDR controller core. The DFI interface is responsible for interface protocol conversion, converting the internal bus of the DDR controller into a bus interface that complies with the DFI protocol; DDR PHY ⑩ is used to implement the interface between the DDR controller and the DDR storage array, and its functions include link quality training, level conversion, data rate and data bit width conversion, etc.

[0069] During DDR stress testing, Figure 3 The DDR stress test device shown works as follows:

[0070] 1. The access generation engine unit ⑥ is at the same level as other application layer interfaces ① and ②, and is presented to the arbitrator ③ in the form of a common port, and the port priority is configurable.

[0071] 2. The access generation engine unit ⑥ generates a write access to a configuration address range according to the configuration; the configuration address range is any storage address range in the pre-configured DDR storage array.

[0072] 3. The test pattern generation logic unit ⑦ generates a test pattern according to the configuration. The types of test patterns include: PRBS code (also known as pseudo-random code), increasing code (or decreasing code), and user-defined code; the generated test pattern is cached in the write data queue of the access generation engine unit ⑥. If the test pattern is a user-defined code, the test pattern generation logic unit will also pre-store the test pattern in the data comparison logic unit ⑧;

[0073] 4. After the write access is completed, the access generation engine unit ⑥ can generate a read access to the configuration address range according to the configuration;

[0074] 5. The data comparison logic unit⑧ intercepts the return data of the read access of the access generation engine unit⑥ on the data flow between the DDR controller and the DDRPHY in the data output bypass, that is, the data transmission link, detects the bit error through the data verification logic, and calculates the bit error rate of each data transmission link (datalane); illustratively, if the test code pattern is PRBS encoding (that is, pseudo-random code) or increasing code (or decreasing code), the data logic verification is performed based on the return data and the test code pattern generated according to the same logic, and the same logic refers to the same logic as the test code pattern generated by the test code pattern generation logic unit. If the test code pattern is a custom encoding, the data logic verification is performed based on the return data and the test code pattern pre-existing in the data comparison logic unit;

[0075] 6. Feedback the bit error rate information of each data lane to the parameter correction logic unit in real time ⑨;

[0076] 7. The parameter correction logic unit ⑨ adjusts the data sampling clock delay and the data sampling discrimination voltage value based on the DDR PHY training parameters, and automatically finds the optimal target parameter combination according to the bit error rate fed back by the data comparison logic unit ⑧. The target parameter combination includes the data sampling clock delay and the data sampling discrimination voltage value after the bit error rate is corrected, and the corrected bit error rate is the minimum bit error rate, such as zero bit error rate.

[0077] In an optional embodiment of the present application, the access generation engine unit is configured to implement a combination of one or more of the following functions:

[0078] 1) The configured address range for write access and read access can be preconfigured, and the configured address range can cover part or all of the storage space range of the DDR storage array; that is, since the read and write address ranges of the access generation engine unit are configurable, the generated read and write access can cover part or all of the storage space of the DDR storage array according to the configured address range;

[0079] 2) It can be configured as a write mode, a read mode, or a write-first-read-later mode; that is, the access generation engine unit can be configured as any one of a write mode, a read mode, or a write-first-read-later mode;

[0080] 3) The number of consecutive transfer cycles for read access and write access to the DDR memory array can be selected from a plurality of candidate lengths; that is, the burst length (number of consecutive transfer cycles) of the read and write accesses can be selected, for example, read and write accesses with a burst length of 4, 8, or 16 can be sent;

[0081] 4) The priority of read access and write access can be configured to adjust the waiting time when passing through the arbiter of the DDR controller by changing the priority of the read access and write access;

[0082] 5) The number of cycles of read access and write access to the configured address range can be configured; that is, the number of cycles of read and write scans for a certain address range can be configured;

[0083] 6) After the task is completed, the interruption can be reported.

[0084] In an optional embodiment of the present application, the test pattern may be obtained by PRBS encoding, an increasing code or a decreasing code, or a pre-defined code; the test pattern generation logic unit is configured to implement a combination of one or more of the following functions:

[0085] 1) Generate PRBS encoding; in some embodiments, it can generate PRBS7, PRBS15, PRBS31, etc.;

[0086] 2) Generate an increasing code or a decreasing code;

[0087] 3) Pre-configure custom encoding via software.

[0088] In some embodiments, the structural block diagram of the test pattern generation logic unit is as follows: Figure 4 As shown, PRNG is a PRBS encoding module, ICG is an increasing code or decreasing code module, and USR PATTERN is a user-defined encoding module. Through configuration, PRNG, ICG or USR PATTERN can be selected to generate corresponding test patterns. After the test pattern passes through the ECC ENCODE module, it is encoded into ECC encoding; the test pattern and ECC encoding together constitute the write data. It should be noted that the ECC ENCODE module may not be set in the test pattern generation logic unit, that is, the test pattern generation logic unit may not perform ECC encoding on the test pattern, but directly output the test pattern, and the DDR controller performs ECC encoding on the test pattern. It can be set according to actual needs, and this application does not limit this.

[0089] In an optional embodiment of the present application, the data comparison logic unit is configured to implement one or more combinations of the following functions:

[0090] 1) intercepting the return data from the data stream of the DFI interface according to the configured address range; the DFI interface is an interface between the DDR controller and the DDR PHY;

[0091] 2) According to the transmission code pattern, select one of PRBS coding, increasing code or decreasing code, and custom coding to obtain the test code pattern;

[0092] 3) Comparing the intercepted return data with the test pattern pre-stored in the data comparison logic unit through a comparator; in some embodiments, the data comparison logic unit compares the read data of each DQ lane with the expected data; the expected data includes the test pattern, such as a test pattern regenerated from a PRBS code, an incrementing code or a decrementing code, or a custom code pre-stored in the data comparison logic unit;

[0093] 4) Count the bit error rate in real time and save it in a register, and feed the bit error rate back to the parameter correction logic unit of the DDR stress test device. The bit error rate in the register can be queried by the user; the bit error rate can be fed back to the parameter correction logic unit in real time.

[0094] In an optional embodiment of the present application, the parameter correction logic unit determines the minimum delay and maximum delay with the minimum bit error rate by adjusting the data sampling clock delay, and determines the center delay based on the average value of the minimum delay and the maximum delay; then, multiple candidate delays are taken in the positive and negative directions of the center delay with a fixed step, and for each of the candidate delays, the data sampling judgment voltage value is adjusted to obtain the maximum voltage value and the minimum voltage value with the minimum bit error rate corresponding to each candidate delay, such as zero, and then a two-dimensional matrix including candidate delays and candidate voltage values ​​is obtained, and the matrix element with the largest margin for the candidate delay and the candidate voltage value is determined from the two-dimensional matrix, and the candidate delay and the candidate voltage value corresponding to the matrix element with the largest margin are respectively used as the data sampling clock delay and the data sampling judgment voltage value in the target parameter combination.

[0095] The parameter correction logic unit is used to generate parameter modification instructions and modify the DQS delay (data sampling synchronization signal delay) and DQ (data sampling) judgment voltage value of the DDR PHY. Figure 5 FIG. 1 is a schematic diagram of delay and voltage parameter adjustment of DDR PHY according to an exemplary embodiment of the present application. Figure 5 As shown, the DDR PHY parameters adjusted by the parameter correction logic unit are RxClkDly and VrefDAC, which are the data sampling synchronization signal delay and the data sampling determination voltage value respectively.

[0096] The parameter correction logic unit supports two correction modes: one-dimensional correction and two-dimensional correction. The one-dimensional correction adjusts the data sampling clock delay parameter RxClkDly to find the minimum delay and maximum delay with zero bit error rate, and takes the average value to obtain the center delay as the correction result.

[0097] The two-dimensional correction is based on the one-dimensional correction. With the correction result of the one-dimensional correction, i.e., the center delay, as the center, N candidate delays are taken in the positive and negative directions with a fixed step. The data sampling discrimination voltage value parameter VrefDAC is adjusted on each candidate delay to obtain the maximum voltage value and the minimum voltage value corresponding to the bit error rate of each candidate delay with zero. Finally, on the two-dimensional plane composed of the candidate delay (including the center delay) and the voltage value, the point with the largest margin of the candidate delay and the voltage value is found, i.e., M v +M d The largest point, where M v is the shortest distance from the zero error boundary in the vertical direction, M d is the shortest distance from the zero error boundary in the horizontal direction, such as Figure 6 As shown in the figure, it is the correction result of two-dimensional correction, that is, the final correction result. Figure 6 ,The parameter correction logic unit supports parallel execution of corrections for multiple DQ lanes.

[0098] In an optional embodiment of the present application, the access generation engine unit is at the same level as other application layer interfaces in the DDR controller and is presented to the arbitrator of the DDR controller in the form of a common port, and the priority of the access generation engine unit can be configured.

[0099] In an optional embodiment of the present application, the data comparison logic unit verifies the return data on each data transmission link between the DDR controller and the DDRPHY to obtain the bit error rate.

[0100] Accordingly, the present application exemplarily provides a DDR stress testing method in a second aspect, the method using the DDR stress testing device proposed in the first aspect to perform DDR stress testing, the method comprising:

[0101] After the DDR stress test process is started, the access generation engine unit constructs the write access of the configuration address range of the DDR storage array;

[0102] In response to the start of the DDR stress test process, the test pattern generation logic unit generates a test pattern and writes the test pattern into the write data queue of the access generation engine unit;

[0103] After the write access is completed, the access generation engine generates a read access to the configuration address range;

[0104] The data comparison logic unit intercepts the return data corresponding to the read access from the data stream between the DDR controller and the DDRPHY, performs data verification using the return data, and calculates a bit error rate.

[0105] In an optional embodiment of the present application, the method further includes:

[0106] If the data comparison logic unit detects a bit error and the bit error rate is less than a set threshold value, determining whether a parameter correction enable signal of the parameter correction logic unit of the DDR stress test device is turned on;

[0107] If the parameter correction enable signal is turned on, the data comparison logic unit reports the bit error rate to the parameter correction logic unit;

[0108] The parameter correction logic unit adjusts the data sampling clock delay and the data sampling discrimination voltage value of the DDRPHY, and determines a target parameter combination according to the bit error rate; the target parameter combination includes the data sampling clock delay and the data sampling discrimination voltage value after correcting the bit error rate.

[0109] In an optional embodiment of the present application, the parameter correction logic unit adjusts the data sampling clock delay and the data sampling discrimination voltage value of the DDR PHY, and determines the target parameter combination according to the bit error rate, including:

[0110] The parameter correction logic unit determines the minimum delay and the maximum delay for a zero bit error rate by adjusting the data sampling clock delay, and determines the center delay based on the average value of the minimum delay and the maximum delay;

[0111] Taking multiple candidate delays in both positive and negative directions of the central delay at a fixed step;

[0112] For each candidate delay, adjusting the data sampling discrimination voltage value, obtaining a maximum voltage value and a minimum voltage value corresponding to each candidate delay with a zero bit error rate, and obtaining a two-dimensional matrix including candidate delays and candidate voltage values;

[0113] Determine from the two-dimensional matrix the matrix element with the largest margin for the candidate delay and candidate voltage value;

[0114] The candidate delay and candidate voltage values ​​corresponding to the matrix element with the largest margin are used as the data sampling clock delay and the data sampling determination voltage value in the target parameter combination, respectively.

[0115] An optional implementation process of the DDR stress test method can be found in Figure 7 shown.

[0116] For specific implementation details of the above DDR stress testing method, please refer to the description in the embodiment of the first aspect, which will not be repeated here.

[0117] It is understood that the circuit structures, names and parameters described in the above embodiments are only examples. Those skilled in the art can also easily combine and adjust the structural features of the above embodiments according to the use requirements, and should not limit the concept of the present application to the specific details of the above examples.

[0118] Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that he or she may still modify the technical solutions described in the aforementioned embodiments, or make equivalent substitutions for some of the technical features therein; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A DDR stress testing device, characterized in that: include: DDR controller and DDRPHY; wherein the DDR controller has an embedded access generation engine unit, a test pattern generation logic unit and a data comparison logic unit; the DDR PHY is an interface between the DDR controller and the DDR storage array; The access generation engine unit is used to generate a write access to a configuration address range after the DDR stress test process is started; the configuration address range is any storage address range in the pre-configured DDR storage array; The test pattern generation logic unit is used to generate a test pattern after the DDR stress test process is started, and store the test pattern in the write data queue of the access generation engine unit; After the write access is completed, the access generation engine generates a read access to the configuration address range, and the data comparison logic unit intercepts the return data corresponding to the read access from the data stream between the DDR controller and the DDR PHY, performs data verification using the return data, and calculates the bit error rate.

2. The DDR stress testing device according to claim 1, characterized in that: The DDR stress test device further includes: a parameter correction logic unit; wherein, The parameter correction logic unit receives the bit error rate from the data comparison logic unit, adjusts the data sampling clock delay and the data sampling judgment voltage value of the DDRPHY, and determines a target parameter combination according to the bit error rate; the target parameter combination includes the data sampling clock delay and the data sampling judgment voltage value after correcting the bit error rate.

3. The DDR stress testing device according to claim 1, characterized in that: The access generation engine unit has one or more combinations of the following functions: The configuration address range for write access and read access can be preconfigured, and the configuration address range can cover part or all of the storage space range of the DDR storage array; Can be configured as write mode, read mode or write-first-read-later mode; The burst length for read access and write access to the DDR memory array can be selected from a variety of candidate lengths; Priorities of read access and write access can be configured to adjust the waiting time when passing through the arbiter of the DDR controller by changing the priorities of the read access and write access; The number of cycles of read access and write access to the configuration address range can be configured; After the task is executed, the interrupt can be reported.

4. The DDR stress testing device according to claim 1, characterized in that: The test pattern generation logic unit is configured to implement one or more combinations of the following functions: Generate PRBS code; Generate an increasing code or a decreasing code; Pre-configured custom encoding via software.

5. The DDR stress testing device according to claim 1, characterized in that: The data comparison logic unit is configured to implement one or more combinations of the following functions: intercepting the return data from the data stream of the DFI interface according to the configured address range; the DFI interface is an interface between the DDR controller and the DDR PHY; According to the transmission code type, select PRBS code, increasing code or decreasing code, or custom code to obtain the test code type; Comparing the intercepted return data with expected data by a comparator; the expected data includes the test pattern; The bit error rate is counted in real time and stored in a register, and the bit error rate is fed back to the parameter correction logic unit of the DDR stress test device.

6. The DDR stress testing device according to claim 2, characterized in that: The parameter correction logic unit determines the minimum delay and maximum delay with zero bit error rate by adjusting the data sampling clock delay, and determines the center delay based on the average value of the minimum delay and the maximum delay; then, multiple candidate delays are taken in the positive and negative directions of the center delay with a fixed step, and for each of the candidate delays, the data sampling discrimination voltage value is adjusted to obtain the maximum voltage value and the minimum voltage value with zero bit error rate corresponding to each candidate delay, and then a two-dimensional matrix including candidate delays and candidate voltage values ​​is obtained, and the matrix element with the largest margin of the candidate delay and the candidate voltage value is determined from the two-dimensional matrix, and the candidate delay and the candidate voltage value corresponding to the matrix element with the largest margin are respectively used as the data sampling clock delay and the data sampling discrimination voltage value in the target parameter combination.

7. The DDR stress testing device according to any one of claims 1 to 6, characterized in that: The access generation engine unit is at the same level as other application layer interfaces in the DDR controller and is presented to the arbiter of the DDR controller in the form of a common port. The priority of the access generation engine unit can be configured.

8. The DDR stress testing device according to any one of claims 1 to 6, characterized in that: The data comparison logic unit verifies the returned data on each data transmission link between the DDR controller and the DDR PHY to obtain the bit error rate.

9. A DDR stress testing method, characterized in that: The method uses the DDR stress testing device according to any one of claims 1 to 8 to perform DDR stress testing, comprising: After the DDR stress test process is started, the access generation engine unit constructs the write access of the configuration address range of the DDR storage array; In response to the start of the DDR stress test process, the test pattern generation logic unit generates a test pattern and writes the test pattern into the write data queue of the access generation engine unit; After the write access is completed, the access generation engine generates a read access to the configuration address range; The data comparison logic unit intercepts the return data corresponding to the read access from the data stream between the DDR controller and the DDR PHY, performs data verification using the return data, and calculates a bit error rate.

10. The DDR stress testing method according to claim 9, characterized in that: The method further comprises: If the data comparison logic unit detects a bit error and the bit error rate is less than a set threshold value, determining whether a parameter correction enable signal of the parameter correction logic unit of the DDR stress test device is turned on; If the parameter correction enable signal is turned on, the data comparison logic unit reports the bit error rate to the parameter correction logic unit; The parameter correction logic unit adjusts the data sampling clock delay and the data sampling discrimination voltage value of the DDR PHY, and determines a target parameter combination according to the bit error rate; the target parameter combination includes the data sampling clock delay and the data sampling discrimination voltage value after correcting the bit error rate.

11. The DDR stress testing method according to claim 10, characterized in that: The parameter correction logic unit adjusts the data sampling clock delay and the data sampling discrimination voltage value of the DDR PHY, and determines a target parameter combination according to the bit error rate, including: The parameter correction logic unit determines the minimum delay and the maximum delay for a zero bit error rate by adjusting the data sampling clock delay, and determines the center delay based on the average value of the minimum delay and the maximum delay; Taking multiple candidate delays in both positive and negative directions of the central delay at a fixed step; For each candidate delay, adjusting the data sampling discrimination voltage value, obtaining a maximum voltage value and a minimum voltage value corresponding to each candidate delay with a zero bit error rate, and obtaining a two-dimensional matrix including candidate delays and candidate voltage values; Determine from the two-dimensional matrix the matrix element with the largest margin for the candidate delay and candidate voltage value; The candidate delay and candidate voltage values ​​corresponding to the matrix element with the largest margin are used as the data sampling clock delay and the data sampling determination voltage value in the target parameter combination, respectively.

Citation Information

Patent Citations

  • Dynamic memory parameter calibration method and device, equipment, medium and program product

    CN115691633A

  • Semiconductor unit serial flash memory signal test method and computer program product

    CN118398071A

  • Semiconductor memory devices and memory systems including the same

    US20160042809A1

  • Signal interference testing using reliable read write interface

    US20240112747A1