Error correction in main memory of a fast computer

The paper is concerned with the properties and ways to develop uniform checking matrices of modified Hamming codes with a minimal number of units which correct single and detect double independent errors and bytes of errors of length 4. A number of uniform checking matrices for (72, 64)- and (137, 128)-codes are described which lead to maximally fast LSI-based coder and decoder. The structure of error correcting units in main memory of fast computers is discussed.