🚀 go-pugleaf

In article <1992Sep3.131306.13751@tc.cornell.edu> bai@msiadmin.cit.cornell.edu (Dov Bai-MSI Visitor) writes:
>In article <92Sep03.032755.21494@acs.ucalgary.ca> bauwens@acs.ucalgary.ca (Luc Bauwens) writes:
>>In article <21529@optima.cs.arizona.edu> shack@cs.arizona.edu (David Shackelford) writes:
>>>In article <1992Sep2.163508.17347@tc.cornell.edu> bai@msiadmin.cit.cornell.edu (Dov Bai-MSI Visitor) writes:
>>>>The following short program compiled with MS-FORTRAN 5.0 with 
>>>>coprocessor crashes OS2 Ver. 2 on my machine:
>>>
>>>Sounds interesting.  Are you compiling to a DOS program and running
>>>it in a VDM, or an OS/2 program?  If OS/2, is it with the 1.x or 2.0
>>>libraries?  Please post more information...
>>
>>If he uses MS-Fortran, it got to be a DOS program.
>
>No, MS-Fortran 5.0 is both a real & protected mode product. I think
>that 5.1 merely adds Windows functionality that does not exist in 5.0.
>
>>
>>I tried with Ms Fortran 5.1 and with Watcom fortran 386 V9.
>>
>>Both return an answer 1.00000000.
>
>It that is true, it is really bad - I would not trust such 
>a compiler for any serious numerical work. 
>
>>
>>Luc B
>>
>>
>
>Dov 

Many motherboards (Micronics Rev E specifically), and I'm sure others,
have a hardware problem with trapping floating point errors. This causes
the problem, not the compiler. The compiler has no effect on this, I
dont believe. 

   ---Ken

-- 
Ken Bass (kbass@gmuvax2.gmu.edu)              |   Telecommunications  
George Mason University                       |   Techniques Corp, 
Student, Department of Electrical Engineering |   Software Engineer