> eg 17! (17Factorial) is the highest I can get exactly (17*16*15 etc.),
> 18! turns to a real number and I don't know if I am losing digits.
So: mathematical curiosity - cool. :)
> Another example of large numbers is time converted to milliseconds:
>
> Thisint64:=strtoint(formatDateTime('yyyymmddhhnnsszzz',now));
Scratching my head over this one - if you aren't doing arithmetic on the
resulting value then it appears to be simply a way to store a date/time
value to the same level of precision as offered by a TDateTime (1msec) in a
datatype that is the same size (8 bytes - a TDateTime is a Double) but in a
form that is harder to do useful work with (arithmetic).
To which the obvious next question has to be: Why?
> My main guess about int128 is that if it were there it could give real
> numbers to a higher number of precision digits (only 15-16 with current
> real).
I don't see how you make the connection. Integers are, by definition,
handled completely differently to floating point values "on the metal". The
provision of a Int128 would have no bearing on the precision of existing
real types.
Only the invention of a new REAL real type could do that.
> Reason I asked is I heard that after Windows 7 one of the supposed
versions
> of Windows being considered was a 128 bit version. Also 64bit Delphi on
the
> horizon suggests that larger integers might be coming.
That is a numerical extrapolation, not a logical conclusion. :)
The main driver behind general purpose 64-bit computing is not larger
integers but larger **pointers**.
The need for 128-bit pointers is very, very slight indeed, and no, this
isn't the same short sighted thinking that led to the apocryphal "no one
will need more than 640KB" comment.
Heck, (general purpose) 64-bit computing doesn't even use ALL of the 64-bits
yet!! :)
The most memory that any Windows system can use currently is 2TB, nowhere
near the theoretical 16 EXABYTE (!!) limit that 64-bits would make
available.
If you have a real need to crunch 128-bit integers then there are computers
and software specifically designed to do such things, and have been for a
long time, serving the sort of research applications that need such things.
But I think it will be a very, VERY long time indeed (if ever) before such
capabilities reach the general computing technology we are dealing with day
to day.
_______________________________________________
NZ Borland Developers Group - Delphi mailing list
Post: delphi@delphi.org.nz
Admin: http://delphi.org.nz/mailman/listinfo/delphi
Unsubscribe: send an email to delphi-requ...@delphi.org.nz with Subject:
unsubscribe
_______________________________________________
NZ Borland Developers Group - Delphi mailing list
Post: delphi@delphi.org.nz
Admin: http://delphi.org.nz/mailman/listinfo/delphi
Unsubscribe: send an email to delphi-requ...@delphi.org.nz with Subject:
unsubscribe
