Some Random Access Memory (RAM) chips have built-in error-checking functions that use a process called parity. Chips that use parity have an extra bit for every eight bits of data. In the parity process, as the eight bits receive binary data (data represented by 1s and 0s), the chip adds all the 1s, and if that total is odd, the extra bit is set to 1.
If the total is even, the extra bit is set to 0. When the computer tries to read each byte of data back from the RAM, it calculates the total number of 1s again and compares its findings to the parity bit. If the findings match up, the data is ruled error-free and can be sent to the CPU.
If the findings don't match up, the chip assumes the data is erroneous and dumps it. This is called even parity. Odd parity is the same process, except that the extra bit setting is 1 when the sum of all the 1s in the byte is even.
The computer checks a bit for parity each time a byte is transferred or transmitted. The process of parity uses simple arithmetic to determine whether stored data values are the same when they are read as they were when they were written. This is fine for detecting errors, but it doesn't do anything to correct them. If parity proves data to be faulty, the memory system simply rejects the data and starts over.
High-end computer servers need a more effective form of error-checking, such as error-correcting code (ECC). ECC uses more memory than parity, but with its special algorithm, it can actually fix most errors it finds. Amazingly, many computers operate with nonparty memory and manage to continue working despite lacking this check. The good news is that memory components used in PCs today have improved and checking the integrity of memory and data is less necessary than it once was for the average home computer