Method for generating repair strategy under DRAM redundant resource layout and implementation device thereof

Abstract

A method and apparatus for generating a repair strategy under a DRAM redundant resource layout are disclosed. The DRAM storage area includes several BlockGroups, each BlockGroup comprising M Blocks and sharing column redundancy resources Ct, and each Block having row redundancy resources Rt. The method includes detecting and obtaining a distribution map of all fault points; dividing the fault points within each Block into several FailGroups according to their correlation, and calculating the solution space for each FailGroup; sequentially permuting and combining the solution spaces of all FailGroups and finding a solution that uses as many row redundancy resources within a Block as possible, limited to Rt row redundancy resources, thereby using the minimum shared column redundancy resource Cmin; after finding the minimum shared column redundancy resource Cmin of the M Blocks, performing statistical summation at the BlockGroup level; if the total Cmin of the M Blocks is less than or equal to Ct, the BlockGroup is determined to be repairable. Therefore, this invention optimizes the redundant resource utilization strategy.

Description

Technical Field

[0001] This invention belongs to the technical field of Automatic Test Equipment (ATE) for integrated circuits, and relates to a method for generating a repair strategy under DRAM redundant resource layout and its implementation device. Background Technology

[0002] During the manufacturing process of DRAM chips, memory cells inevitably fail. Therefore, redundant resources need to be arranged around the main memory area. After the fault point is detected in the testing phase, it is replaced by redundant resources. The search for this replacement solution is called redundancy analysis. This invention is applied to this redundancy analysis phase.

[0003] Currently, there are various designs for redundant resource layouts in the industry. A traditional layout is as follows: The DRAM redundant resource layout consists of several regions, each region including N block groups, each block group including M blocks, and each block having several row redundant resources Rt and several column redundant resources Ct. This form is relatively simple in circuit design and algorithm implementation. However, the above-mentioned redundant resource control circuit occupies a large wafer area and is not cost-effective.

[0004] The latest layout method involves a BlockGroup sharing several column redundancy resources, while each Block has several dedicated row redundancy resources. This layout is basically consistent with the traditional layout in terms of mathematical modeling of fault points within a Block, but the solution process differs significantly.

[0005] ① The mathematical modeling of fault points all adopts the sparse matrix method, and the row and column of the fault point are stored as a node in the sparse matrix;

[0006] ② Within the fault matrix, nodes are grouped together. Nodes with the same row address or column address are grouped together to form several fault groups (FailGroup).

[0007] ③ Solve for each FailGroup. Based on the maximum number of rows Rm and the maximum number of columns Cm occupied, find its solution space, i.e., {R1Cx, R2Cx...RmCx}. Here, RmCx means that after occupying Rm row redundancy lines, the fault group needs at least Cx column redundancy lines to cover it.

[0008] ④ For blocks with a traditional layout, based on the solution space of each FailGroup, find a solution to satisfy the constraints of the total number of row redundancy resources Rt and the total number of column redundancy resources Ct of the entire block. Figure 1 For example:

[0009] For the above layout, a solution can be found in the solution space of each failure group (FailGroup). These solutions are then arranged and combined to find any combination that satisfies the constraints of the total number of row redundancy resources Rt and the total number of column redundancy resources Ct of the block. This combination can then be used as a solution for the block. Figure 1 As shown, R2C0 of FailGroup1 + R1C1 of FailGroup2 + R1C0 of FailGroup3 = R4C1, which does not exceed the limits of the total number of row redundancy resources Rt and the total number of column redundancy resources Ct of the block, and can be used as the solution for the block.

[0010] ⑤ However, for the latest layout, there is only the total number of row redundancy resources Rt within a block, while column redundancy resources are shared by all blocks within a block group. Therefore, there is no total number of column redundancy resources Ct within a block, and the above method cannot be applied. A new strategy needs to be found. Summary of the Invention

[0011] To address the aforementioned technical problems, this invention proposes a method for generating a repair strategy under DRAM redundant resource layout. The proposed resource allocation strategy solves the problem of optimal allocation in DRAM redundant resource layout.

[0012] To achieve the above objectives, the technical solution of the present invention is as follows:

[0013] A method for generating a repair strategy under a DRAM redundancy resource layout, wherein the main storage area of ​​the DRAM is composed of several regions, each region includes N block groups, each block group includes M blocks and has shared column redundancy resources Ct, and each block has row redundancy resources Rt, comprising the following steps:

[0014] Step S1: Detect the main storage area of ​​the DRAM to obtain a distribution map of all fault points in the storage area of ​​each DRAM;

[0015] Step S2: Store each row and column of all the fault points as a node in a sparse matrix. That is, use a sparse matrix method to mathematically model the distribution map of all the fault points. Divide the fault points in each block into several fault point groups (FailGroup) according to their association relationship, and calculate the solution space of each fault point group (FailGroup). The solution space includes any combination that satisfies the block's RxCy resource combination to repair the constraint. Rx is the x-row redundant resource and Cy is the shared y-column redundant resource.

[0016] Step S3: Arrange and combine the solution space of each FailGroup in sequence, and find a solution. The solution is to use as many row redundancy resources as possible within each block as possible, given the row redundancy resource Rt of each block, thereby using the minimum shared column redundancy resource Cmin.

[0017] Step S4: After finding the minimum shared column redundancy Cmin of the M blocks, perform statistical summation at the BlockGroup level. If the total number of minimum shared column redundancy Cmin of the M blocks is less than or equal to the total number of redundancy resources of the BlockGroup, it is determined that the BlockGroup can be repaired; otherwise, it is determined that the BlockGroup cannot be repaired.

[0018] To achieve the above objectives, the technical solution of the present invention is as follows:

[0019] A repair apparatus under a DRAM redundancy resource layout that implements the above-described repair strategy generation method, wherein the DRAM redundancy resource layout includes at least one storage region of the DRAM, each region includes N block groups, each block group includes M blocks and has shared column redundancy resources Ct, and each block has row redundancy resources Rt, which include:

[0020] The detection module detects the main storage area of ​​the DRAM and obtains a distribution map of all fault points in the storage area of ​​each DRAM.

[0021] The solution space generation module is used to store the rows and columns of all the fault points as nodes in a sparse matrix. That is, it uses a sparse matrix approach to mathematically model the distribution map of all the fault points and calculates and obtains the solution space of each fault point group FailGroup. The solution space includes any combination that satisfies the RxCy resource combination of the block to repair the constraint, where Rx is the x-row redundant resource and Cy is the shared y-column redundant resource.

[0022] The solution selection module is used to sequentially arrange and combine the solution space of each FailGroup and find a solution. The solution is to use as many row redundancy resources as possible within each Block, given the row redundancy resource Rt of each Block, thereby using the minimum shared column redundancy resource Cmin.

[0023] The repair judgment module is used to find the minimum shared column redundancy resources Cmin of the M blocks, and then perform statistical summation at the block group level. If the total number of minimum shared column redundancy resources Cmin of the M blocks is less than or equal to the total number of redundancy resources of the block group, the block group is determined to be repairable; otherwise, the block group is determined to be unrepairable.

[0024] As can be seen from the above technical solutions, the DRAM redundant resource layout repair strategy generation method and its implementation device in the embodiments of the present invention have only the total number of row redundant resources Rt and no total number of column redundant resources Ct in the block. When the column redundant resources are shared among the block groups, the restrictions on the total number of row redundant resources Rt and the total number of column redundant resources Ct of the entire block can be met, thus optimizing the redundant resource usage strategy. Attached Figure Description

[0025] Figure 1 The diagram shows a method for finding the solution space of a block in the prior art.

[0026] Figure 2 The diagram shown is a flowchart illustrating the method for generating a repair strategy under DRAM redundancy resource layout in an embodiment of the present invention.

[0027] Figure 3 The diagram shown is a schematic representation of the method for finding solutions within a block in an embodiment of the present invention.

[0028] Figure 4 The diagram shown illustrates the method for finding a BlockGroup solution in an embodiment of the present invention. Detailed Implementation

[0029] The following is in conjunction with the appendix Figure 2-4 The specific embodiments of the present invention will be further described in detail below.

[0030] It should be noted that this invention is a resource allocation strategy proposed to address the latest redundant resource layout of DRAM. The latest redundant resource layout of DRAM includes at least one storage region of the DRAM, each region includes N block groups, each block group includes M blocks and has shared column redundancy resources Ct, and each block has row redundancy resources Rt.

[0031] Please see Figure 2 , Figure 2 The diagram illustrates the flow of a repair strategy generation method under DRAM redundancy resource layout in an embodiment of the present invention. Figure 2 As shown, the method for generating a repair strategy under this DRAM redundancy resource layout includes the following steps:

[0032] Step S1: Detect the storage area of ​​the DRAM to obtain a distribution map of all fault points in the storage area of ​​each DRAM; this step can be completed using any ATE testing equipment.

[0033] It should be noted that the DRAM redundancy resource layout includes at least one storage area of ​​the DRAM. This invention only describes one, and if there are several, the processing method is the same, so it will not be described in detail here.

[0034] Step S2: Store the rows and columns of all the fault points as nodes in a sparse matrix. That is, use a sparse matrix method to mathematically model the distribution map of all the fault points, and calculate and obtain the solution space of each fault point group FailGroup. The solution space includes any combination that satisfies the RxCy resource combination of the block to repair the constraint, where Rx is the x-row redundant resource and Cy is the shared y-column redundant resource.

[0035] Please refer to Figure 3 , Figure 3 The diagram illustrates the method for finding the solution space of a block in an embodiment of the present invention. In this embodiment, the row redundancy resource Rt of each block is 4. Assuming there are 3 failure groups (FailGroups) within the block, the solution space of each failure group is as follows: Figure 3 As shown, the next step is to find the minimum number of shared column redundant resources Cmin required when the number of row redundant resources Rt is 4.

[0036] Step S3: Sequentially arrange and combine the solution space of each FailGroup and find a solution that uses as many row redundancy resources as possible within a block, with a limit of Rt row redundancy resources, thereby using as few shared column redundancy resources as possible Cmin.

[0037] In other words, we can find one solution in the solution space of each failure group (FailGroup), take them out and perform permutations and combinations, search all combinations, and find the minimum value of C that satisfies the row redundancy resource limit Rt of the block, which is taken as the solution Cmin of the block. Figure 3 As shown, R2C0 of FailGroup1 plus R1C1 of FailGroup2 plus R1C0 of FailGroup3 = R4C1, which does not exceed the Rt limit of Block, and the C value is the smallest, that is, the solution of the block is Cmin = 1.

[0038] Step S4: After finding the minimum shared column redundancy Cmin of the M blocks, perform statistical summation at the BlockGroup level. If the total number of minimum shared column redundancy Cmin of the M blocks is less than or equal to the total number of redundancy resources of the BlockGroup, it is determined that the BlockGroup can be repaired; otherwise, it is determined that the BlockGroup cannot be repaired.

[0039] In other words, the prerequisite for selecting the solution in the embodiments of the present invention is:

[0040] A block contains at least one failgroup, all of which share a common characteristic: if the current failure can be repaired using RxCy resource combinations, and if one more row of redundant resources is provided, then the requirement for shared column of redundant resources will remain unchanged or decrease, and will not increase. That is, Rx+1Cy1, y1<=y.

[0041] Therefore, the solution of the present invention is as follows:

[0042] ① When repairing within a block, find a solution that uses as many row redundancy resources as possible within the block, thereby using as few shared column redundancy resources as possible (Cmin).

[0043] ② After finding the minimum shared column redundancy Cmin for each block, perform a statistical summation at the block group level. If the total is less than the total redundancy Ct of the block group, then the block group can be repaired.

[0044] Please refer to Figure 4 , Figure 4The figure shows a schematic diagram of the method for finding the block solution in an embodiment of the present invention. As Figure 4 shown, in an embodiment of the present invention, there are three blocks Block. The minimum shared column redundancy resources Cmin of each block Block are statistically summed up, and the sum value is compared with the total shared column redundancy resources Ct of the block group BlockGroup. If the sum of the minimum shared column redundancy resources Cmin of the three blocks Block is 1 + 3 + 2 = 6 < Ct, the block group BlockGroup composed of these three blocks Block can be repaired. Otherwise, it is determined that the block group BlockGroup cannot be repaired.

[0045] In addition, in an embodiment of the present invention, the above repair strategy generation method can be implemented by the following device. The DRAM redundancy resource layout includes at least one storage area of the DRAM. Each area includes N block groups BlockGroup. Each block group BlockGroup includes M blocks Block and has shared column redundancy resources Ct. Each block Block has row redundancy resources Rt. It includes a detection module, a solution space generation module, a solution selection module, a repair judgment module, etc. These above modules can be included in the ATE device or used separately.

[0046] The detection module can be used to detect the storage area of the DRAM to obtain the distribution map of all fault points. The solution space generation module is used to take the rows and columns of all the fault points as a node respectively and store them in a sparse matrix, that is, use the sparse matrix method to mathematically model the distribution map of all the fault points, and calculate and obtain the solution space of each fault point group FailGroup; wherein, the solution space includes any combination that satisfies the repair limit of the RxCy resource combination of the block, Rx is the x row redundancy resource, and Cy is the shared y column redundancy resource. The solution selection module is used to permute and combine the solution space of each fault point group FailGroup in turn and find a solution. The solution is limited by Rt row redundancy resources, and uses as many row redundancy resources in the block Block as possible, so as to use as few shared column redundancy resources Cmin as possible; the repair judgment module is used to find the minimum shared column redundancy resources Cmin of the M blocks Block, and do a statistical sum at the block group BlockGroup level. If the total number of the minimum shared column redundancy resources Cmin of the M blocks Block is less than or equal to the total redundancy resources of the block group BlockGroup, it is determined that the block group BlockGroup can be repaired, otherwise, it is determined that the block group BlockGroup cannot be repaired.

[0047] The above description is merely a preferred embodiment of the present invention. The embodiments are not intended to limit the scope of patent protection of the present invention. Therefore, any equivalent structural changes made based on the description and drawings of the present invention should also be included within the scope of protection of the present invention.

Claims

1. A method for generating a repair strategy under a DRAM redundancy resource layout, wherein the DRAM redundancy resource layout includes at least one storage region of the DRAM, each storage region includes N block groups, each block group includes M blocks and has shared column redundancy resources Ct, and each block has row redundancy resources Rt, characterized in that, Includes the following steps: Step S1: Detect the main storage area of ​​the DRAM to obtain a distribution map of all fault points in the storage area of ​​each DRAM; Step S2: Store each row and column of all the fault points as a node in a sparse matrix. That is, use a sparse matrix method to mathematically model the distribution map of all the fault points. Divide the fault points in each block into several fault point groups (FailGroup) according to their association relationship, and calculate the solution space for each fault point group (FailGroup). The solution space includes any combination that satisfies the RxCy resource combination of the block to repair the constraint. Rx is the redundancy resource in row x, and Cy is the shared redundancy resource in column y. Step S3: Arrange and combine the solution space of each FailGroup in sequence, and find a solution. The solution is to use as many row redundancy resources as possible within each block as possible, given the row redundancy resource Rt of each block, thereby using the minimum shared column redundancy resource Cmin. The permutation and combination includes: finding one solution in the solution space of each FailGroup and permuting and combining them, searching all combinations, and finding the minimum value of the row redundancy resource number C that satisfies the Block, which is the solution of the block. Step S4: After finding the minimum shared column redundancy Cmin of the M blocks, perform statistical summation at the BlockGroup level. If the total number of minimum shared column redundancy Cmin of the M blocks is less than or equal to the total number of redundancy resources of the BlockGroup, it is determined that the BlockGroup can be repaired; otherwise, it is determined that the BlockGroup cannot be repaired.

2. A repair apparatus under a DRAM redundancy resource layout for implementing the repair strategy generation method of claim 1, wherein the DRAM redundancy resource layout includes at least one storage region of the DRAM, each region includes N block groups, each block group includes M blocks and has shared column redundancy resources Ct, and each block has row redundancy resources Rt, characterized in that, include: The detection module detects the main storage area of ​​the DRAM and obtains a distribution map of all fault points in the storage area of ​​each DRAM. The solution space generation module is used to store the rows and columns of all the fault points as nodes in a sparse matrix. That is, it uses a sparse matrix approach to mathematically model the distribution map of all the fault points. The fault points in each block are divided into several fault point groups (FailGroup) according to their association relationships, and the solution space of each fault point group (FailGroup) is calculated. The solution space includes any combination that satisfies the block's RxCy resource combination to repair the constraint. Rx is the x-row redundant resource and Cy is the shared y-column redundant resource. The solution selection module is used to sequentially arrange and combine the solution space of each FailGroup and find a solution. The solution is to use as many row redundancy resources within each Block as possible, given the row redundancy resource Rt of each Block, thereby using the minimum shared column redundancy resource Cmin. The repair judgment module is used to find the minimum shared column redundancy resources Cmin of the M blocks, and then perform statistical summation at the block group level. If the total number of minimum shared column redundancy resources Cmin of the M blocks is less than or equal to the total number of redundancy resources of the block group, the block group is determined to be repairable; otherwise, the block group is determined to be unrepairable.

Images