IBM 610 Manual Of Operation

Also see for 610: Technical overview and introduction

OTHER MODES OF OPERATION 53tents of register Y and, in the same operation,multiply the quotient by the contents of register Z -the operation is ^Z the rule governing the loca-tion of the decimal point in the result (result in Aregister) is:Da = Dx -Dy + DzFor example:X Y Z A+ 16.0 divided by -4.0 times +3.0 equals -120When the decimal point is correctly positioned, Acontains - 12.0.To avoid excessive calculation time and overflowchecks, follow the magnitude rule applied to X/Y.If we take the square root of the contents of regis-ter X, the location of the decimal point in the answer(answer in the A register) is given by the followingrule:Da = Vz DxBecause decimal indices must always be whole num-bers, this rule implies that Dx must be an even num-ber.For example:X AThe square root of +25.0 is +5.0To take the square root of +160.0 is not per-missible, and the machine acts as though the numberis 16.00. Instead, it should be positioned as: +0160.0,and the correct answer will be obtained.If we try to take the square root of a negativenumber, the machine ignores the sign and behavesas though the number were positive.It should also be noted that the square root of+ 16.00, +1600.0, and +160000.0 appear as thesame number in the A register. Only operator knowl-edge of the problem enables him to position the deci-mal point in the correct place in the A register.Note: All of the above discussion applies to thecontrol of the octal point when the machine is operat-ing in the octal mode.Operators should be cautioned that intermediateresults may be sufficiently large to cause overfloweven though the input data is numerically small. It iswell, therefore, to pay careful attention during allstages of a problem.Operators should also note that the first set of datais often chosen so that the order of magnitude of thecalculations will be a good indicator of their correct-ness. Subsequent sets of data may not behave as wellat intermediate points in the calculation. Care mustbe exercised in positioning numbers so that overflow,division by zero, and like occurrences do not leadto the belief that the machine is not operating cor-rectly.All of these considerations for fixed-point calcula-tions are taken care of by the machine when operatingin auto-point.