m/c Precision vs. Desired Tolerance

From: monir <monirg@xxxxxxxxxxxx>
Date: Tue, 09 Oct 2007 08:35:37 -0700

Hello;
Assuming the m/c precision EPSILON() for real data type is:
EPSILON(real) = 1.192 092 9E-7
and
X1 = 0.123 456 78
X2 = 0.123 456 79
X3 = (X2-X1) ! smaller than EPSILON
TOL= 1.E-8 ! smaller than EPSILON
then, which statement label the following arithmetic IF statement
would transfer control to ??:
IF (X3 - TOL) 1,2,3

(I guess, label 1 NOT 2, since X3 is smaller than the m/c smallest
increment EPSILON in single precision. Correct ??)

Thank you.
Monir

.

Follow-Ups:
- Re: m/c Precision vs. Desired Tolerance
  - From: Chip Coldwell
- Re: m/c Precision vs. Desired Tolerance
  - From: Richard Maine

Prev by Date: Re: Problem with log function - Intel fortran compiler under Linux
Next by Date: Getting fed up with Intel Premier support for IVF
Previous by thread: LNK2001 error
Next by thread: Re: m/c Precision vs. Desired Tolerance
Index(es):
- Date
- Thread

Relevant Pages

Re: m/c Precision vs. Desired Tolerance
... which statement label the following arithmetic IF statement ... IF (X3 - TOL) 1,2,3 ... increment EPSILON in single precision. ... The standard would allow any of the three choices here. ...
(comp.lang.fortran)
Re: m/c Precision vs. Desired Tolerance
... Not sure if this qualifies as 'dim history', but the Microsoft Fortran ... Compiler v3.31 ... IF (X3 - TOL) 1,2,3 ... increment EPSILON in single precision. ...
(comp.lang.fortran)