| 摘要 |
The method and apparatus provides a parity bit for every m multiples of b bits, a group of b bits being the smallest number of bits that can be manipulated by the CPU. The parity bit is computed for the entire m x b bits during a write operation, even if only a subset of the m multiples of b bits is being stored. The write operation is implemented as a read-modify-write operation of the entire m x b bits, with parity error reporting suppressed for the read portion of the operation. However, the parity bit is set factoring in whether a parity error is detected during the read portion of the operation. The parity bit for the entire m x b bits is checked during a read operation, even if only a subset of the m multiples of b bits is needed. Any detected parity error is reported to the CPU. As a result, hardware cost is substantially reduced with minimal degradation to data integrity. Furthermore, the method and apparatus is completely transparent to the CPU and the operating system. |
