Processors
Intel® Pentium® Processor
Statistical Analysis of Floating Point Flaw: Intel White Paper - Appendix A

Appendix A
In the process of the characterization, a list of scientific constants that might commonly be used in floating point calculations were examined as potential divisors. These constants are shown in Table A-1 below. Two of these constants, indicated by the shaded rows, were identified as potentially problematic based on the analysis in Section 4.2. To verify if these two constants were really at risk each was used as a divisor with 100 billion random dividends with no errors found.

Table A-1 Engineering and Scientific Constants Analyzed.  

 

 

Table A-2 below contains the list of combinations of the aforementioned constants that were examined additionally for possible occurrences of an error due to the flaw. No problem was found.

Table A-2 Constants Used in Combination  

 

Table A-3 below indicates the commonly used multiples of the important constants and their multiples that were checked. None of these multiples were found to be at risk.

Table A-3 Multiples of Common Constants  

 

This applies to:

FDIV Replacement Program



Solution ID: CS-013020
Last Modified: 11-Dec-2014
Date Created: 08-Jul-2004
Back to Top