Re: [STDS-754] GRE: rogue comment: IEEE 754R and Reproducibility

[Nick Maclaren <nmm1@xxxxxxxxxxxxx>]

Mike Cowlishaw <MFC@xxxxxxxxxx> wrote:
> In that case, please be more specific in the future.  One Fortran library 
> out of (too many) thousands of IBM products at the time is hardly cause to 
> condemn 'IBM 1970s mainframes'.  I'm often amazed at how much code and how 
> many algorithms developed back then, on that hardware, are still relevant 
> today.  

I was.  Please do not (a) quote out of context and (b) confuse yourself
in jumping to conclusions by doing so.  Here is the COMPLETE paragraph
to remind you:

    Absolute, bitwise consistency has always been a nightmare to deliver
    and almost always an obstacle to fixing bugs or improving accuracy.
    Anyone with experience of IBM 1970s mainframes can brear witness
    to that!

That is not a condemnation of IBM mainframes in any sense.  All it
said was that anyone with experience of them can bear witness to the
fact that absolute, bitwise consistency has always been a nightmare
to deliver and almost always an obstacle to fixing bugs or improving
accuracy.  And that is true.


Regards,
Nick Maclaren,
University of Cambridge Computing Service,
New Museums Site, Pembroke Street, Cambridge CB2 3QH, England.
Email:  nmm1@xxxxxxxxx
Tel.:  +44 1223 334761    Fax:  +44 1223 334679