Operations involving floating point numbers are not always associative due to approximation and rounding errors: ((A + B) + C) is not always equal to (A + (B + C)).

Although not readily apparent, the non-associativity of floating point arithmetic can also affect aggregate operations: you can get different results each time you use an aggregate function on a given set of floating point data. When Teradata Database performs an aggregation, it accumulates individual terms from each AMP involved in the computation and evaluates the terms in order of arrival to produce the final result. Because the order of evaluation can produce slightly different results, and because the order in which individual AMPs finish their part of the work is unpredictable, the results of an aggregate function on the same data on the same system can vary.