Abstract: A fault bits scrambling memory and method thereof relate to a memory including at least one memory bank. The memory bank includes a memory module, a scrambling-logic unit, a self-testing unit and a scrambling code generating unit. The memory module includes a plurality of pages. Each page has a plurality of memory cells, and each memory cell has a physical address. The scrambling logic unit receives a scrambling code and the physical address to generate a mapping address by logical calculation, and outputs the mapping address to the memory module. The self-testing unit detects the faulty memory cells of each page. The scrambling code generating unit is applied to generate the scrambling code to maintain the number of the faulty memory cells corresponding to the mapping address of the same page is up to a maximum tolerance.

Abstract: A fault bits scrambling memory and method thereof relate to a memory including at least one memory bank. The memory bank includes a memory module, a scrambling-logic unit, a self-testing unit and a scrambling code generating unit. The memory module includes a plurality of pages. Each page has a plurality of memory cells, and each memory cell has a physical address. The scrambling logic unit receives a scrambling code and the physical address to generate a mapping address by logical calculation, and outputs the mapping address to the memory module. The self-testing unit detects the faulty memory cells of each page. The scrambling code generating unit is applied to generate the scrambling code to maintain the number of the faulty memory cells corresponding to the mapping address of the same page is up to a maximum tolerance.

Abstract: A fault masking method is applied to a non-volatile memory array which includes a faulty cell and electrically connected to an address register providing a first address. The faulty cell can only output a fixed value. The content of the first address is not equal to the fixed value. The method includes: providing a complementer electrically connected between the address register and the faulty cell; providing a control word; writing the first address and the control word into the complementer; performing a complement operation on the first address and the control word by the complementer to obtain a second address, and storing the content of the second address into the faulty cell, wherein the content of the second address is equal to the fixed value. The method can reduce or eliminate the usage of redundancy in non-volatile memories, so as to reduce the manufacturing costs and improve the fabrication yield.

Abstract: A fault masking method is applied to a non-volatile memory array which includes a faulty cell and electrically connected to an address register providing a first address. The faulty cell can only output a fixed value. The content of the first address is not equal to the fixed value. The method includes: providing a complementer electrically connected between the address register and the faulty cell; providing a control word; writing the first address and the control word into the complementer; performing a complement operation on the first address and the control word by the complementer to obtain a second address, and storing the content of the second address into the faulty cell, wherein the content of the second address is equal to the fixed value. The method can reduce or eliminate the usage of redundancy in non-volatile memories, so as to reduce the manufacturing costs and improve the fabrication yield.