Re: IEEE 754 double floats

2004-11-16 Thread Bill Coffman
Another way is to count the bits, as in the following: .sub _main N1 = 1 N2 = 0.5 I0 = 0 REPEAT: I0 = I0 + 1 N2 = N2 / 2 N3 = N1 + N2 ne N1, N3, REPEAT print I0 print " bits precision\n" end .end On Wed, 17

Re: IEEE 754 double floats

2004-11-16 Thread Leopold Toetsch
Adam Warner <[EMAIL PROTECTED]> wrote: > In other words sprintf is printing trailing garbage: Well, the underlaying architecture is defining the precision of floats. And due to the binary nature of the representation of floats they are mostly just and inprecise approximation of a given float valu

Re: IEEE 754 double floats

2004-11-16 Thread Adam Warner
Hi Leopold Toetsch, >> PI is (very) approximately: > 3.1415906535896946927266526472521945834159851074218750 > ^ > 3.141592653589793238462643383279502884197169399375105820974944 > > You might probably want to run more iterations ;) And you'll never get > 60 digits out of l

Re: IEEE 754 double floats

2004-11-16 Thread Leopold Toetsch
Adam Warner <[EMAIL PROTECTED]> wrote: > On a windows binary I downloaded the precision is indeed double (~16 > decimal places). With my current Linux binary it's extreme (here's 60 > decimal places): > PI is (very) approximately: 3.1415906535896946927266526472521945834159851074218750

Re: IEEE 754 double floats

2004-11-16 Thread Adam Warner
> This may be the long double version that I compiled :-) Note: I've rebuilt parrot-latest.tar.gz (I believe it was 2004-11-16_00) at default settings on Debian unstable 2.6.8.1 i686 GNU/Linux. I'm still printing floats that appear to be 128-bit precision! Regards, Adam

Re: IEEE 754 double floats

2004-11-16 Thread Adam Warner
Hi Brent 'Dax' Royal-Gordon, >> option. Yet the Leibniz summation for PI >> >> still appears to be performing its calculations using single floats and >> continues to print 3.141591. > > Parrot usually uses double as its floating-point type. The problem is

Re: IEEE 754 double floats

2004-11-15 Thread Brent 'Dax' Royal-Gordon
Adam Warner <[EMAIL PROTECTED]> wrote: > I've taken a first look at Parrot and I'm very impressed. I'd like to > perform some micro-benchmarking of double floating point code. I have the > latest CVS version of parrot. I've compiled it upon Debian unstable x86 > with the: > >perl Configure.pl

IEEE 754 double floats

2004-11-15 Thread Adam Warner
Hi all, I've taken a first look at Parrot and I'm very impressed. I'd like to perform some micro-benchmarking of double floating point code. I have the latest CVS version of parrot. I've compiled it upon Debian unstable x86 with the: perl Configure.pl --floatval=double option. Yet the Leibniz