Browse Prior Art Database

Data Pattern for Testing Read/Write Storage Facilities

IP.com Disclosure Number: IPCOM000120530D
Original Publication Date: 1991-May-01
Included in the Prior Art Database: 2005-Apr-02
Document File: 3 page(s) / 119K

Publishing Venue

IBM

Related People

Iseminger, RG: AUTHOR

Abstract

Disclosed is a data pattern that has unique properties, when applied in the prescribed manner, that are desirable for testing read/write storage facilities. While the data pattern is particularly effective when used in a memory diagnostic, it is also useful for general data flow and I/O testing.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Data Pattern for Testing Read/Write Storage Facilities

      Disclosed is a data pattern that has unique properties,
when applied in the prescribed manner, that are desirable for testing
read/write storage facilities.  While the data pattern is
particularly effective when used in a memory diagnostic, it is also
useful for general data flow and I/O testing.

      The data pattern consists of a repetition of the seven bit
sequence, "1 0 1 1 0 0 0", and the inverse of same, "0 1
0 0 1 1 1". When the resultant serial string of bits is parcelled
into units (or groups of bits) such as hex digits, bytes, words,
etc., a seven unit repetition is apparent. This provides a total of
fourteen different units comprised of seven "true" units and seven
"complement" units (see Figure 1).  As an aid for remembering the bit
sequence, and to provide a name, the byte unit, '10110001'b or 'B1'x,
was chosen to identify this pattern as the "B1" data pattern.

      The B1 Pattern has a mirror image pattern that is the same
sequence of seven bits as "B1" (1011000'b) but in reverse order
('0001101'b).  It exhibits the same characteristics as the B1 Pattern
and can be utilized in the same manner.

      Since computer structure is typically based on "powers of two",
the application of a "repeat by seven" pattern, such as the B1 data
pattern, yields a ripple effect.  A ripple pattern, when applied to
storage arrays, e.g., memory chips, is more successful than an even-
repeat pattern at reproducing "cell disturb" failures due to the
physical distribution of the zero and one charges in the bit cells.
While a ripple pattern can be obtained through the application of any
odd-repeat pattern to a power-of-two structure, the B1 Pattern
exhibits unique characteristics that enhance the production and
identification of failures. For example, single- or double-bit
inversions in a B1 Pattern byte-unit will produce a byte that is
different from any of the 14 valid units.  To demonstrate, 'C5'x XOR
'21'x yields "E4'x which is not a valid B1 Pattern byte-unit.  The
significance of this characteristic is that the B1 Pattern units can
be randomly verified and modified (permuted) without the necessity of
associating a specific unit value with a specific position (memory
address or sequence number).

      It is recommended that the wider units, e.g., two-byte, four-
byte, etc., be employed for...