Determination and checking of floating-point arithmetic errors in IBM PC. (English. Russian original) Zbl 0963.65046
Mosc. Univ. Comput. Math. Cybern. 1998, No. 1, 62-70 (1998); translation from Vestn. Mosk. Univ., Ser. XV 1998, No. 1, 43-49 (1998).
It is well known that contrary to operations with integer arithmetic operations with real numbers (real operands) give some errors due to incorrect representation of the operands and methods of fulfilment of arithmetic operations. A method of operational analysis of errors of the real arithmetic which is oriented on an architecture of IBM PC is proposed.
Reviewer: Sergei Georgievich Zhuravlev (Moskva)
