amforth-devel

[Amforth-devel] printing floats

[Leon N. M. <leo...@gm...>]

I've finally added a function to output floats (in scientific notation):

> 355 s>f 113 s>f f/ fs.
3.1415927  ok
> 1 s>f f2/ f2/ f2/ f2/ fs.
6.25E-2  ok
> 12345 s>f fs.
1.23449993E4  ok

You can see from the second example that rounding problems do happen. That can 
be avoided... if you don't mind coding with decuple length integers (need to 
be able to store up to 2^154). See the pdf at:

http://portal.acm.org/citation.cfm?id=93559

So I'm going to stick with this simpler -- but less accurate -- method for the 
moment (also in the pdf) -- I'll work on input before spending more time with 
output (similar problems exist -- maybe there are easier solutions). As usual, 
the code is available at:

http://github.com/lnmaurer/amforth-float

-Leon

PS I'm not sure if these type of updates are of general interest, but there's 
been a lot of floating point stuff going around, so I'll keep sending them for 
the moment.

Re: [Amforth-devel] printing floats

[pito <pi...@vo...>]

Leon, thanks! I am very happy that you publish the float updates!
Pito.

PS: Leon, is there any chance to name the .zip updates with e.g. a
date in it as it is uneasy to recognise and track the files based on
those codes like "..-float-9e1ace1.zip"
P.

> 
> PS I'm not sure if these type of updates are of
> general interest, but there's 
> been a lot of floating point stuff going around,
> so I'll keep sending them for 
> the moment.

Re: [Amforth-devel] printing floats

[pito <pi...@vo...>]

Leon,
I did a lot of various tests I am not going to publish and to make
the story short:
1. it seems the fs. routine presition itself is 7 digits which is
good
2. it seems the f/ most notably with larger operands (eg. +/- 1e15)
introduces errors so the result's precision drops to 6 digits or
less
3. what is interesting the fs. conversion took seconds (atmega
@25MHz)e.g.:
\
marker -test1
: measure
timer-start
-1234567890. d>f fs. cr
timer-stop prnt_elapsed_ms ." msec" ;
> measure
-1.2345674E9 
1174 msec ok
> 

\
marker -test1
: measure
timer-start
_-1 _pi  f* _1e15 f* fs.  cr
timer-stop prnt_elapsed_ms ." msec" ;
> measure
-3.1415915E15 
1709 msec ok
>

Pito


Pito

Re: [Amforth-devel] printing floats

[李雪愚 <sno...@gm...>]

hi, Leon,

Maybe you can port the float lib of PForth to amforth:

  http://pforth.googlecode.com/svn/trunk/fth/floats.fth

unitTest: http://pforth.googlecode.com/svn/trunk/fth/t_floats.fth

Best Regards


> I've finally added a function to output floats (in scientific notation):
> 
>> 355 s>f 113 s>f f/ fs.
> 3.1415927  ok
>> 1 s>f f2/ f2/ f2/ f2/ fs.
> 6.25E-2  ok
>> 12345 s>f fs.
> 1.23449993E4  ok
> 
> You can see from the second example that rounding problems do happen. That can 
> be avoided... if you don't mind coding with decuple length integers (need to 
> be able to store up to 2^154). See the pdf at:
> 
> http://portal.acm.org/citation.cfm?id=93559
> 
> So I'm going to stick with this simpler -- but less accurate -- method for the 
> moment (also in the pdf) -- I'll work on input before spending more time with 
> output (similar problems exist -- maybe there are easier solutions). As usual, 
> the code is available at:
> 
> http://github.com/lnmaurer/amforth-float
> 
> -Leon
> 
> PS I'm not sure if these type of updates are of general interest, but there's 
> been a lot of floating point stuff going around, so I'll keep sending them for 
> the moment.
> 
> ------------------------------------------------------------------------------
> This SF.net Dev2Dev email is sponsored by:
> 
> Show off your parallel programming skills.
> Enter the Intel(R) Threading Challenge 2010.
> http://p.sf.net/sfu/intel-thread-sfd
> _______________________________________________
> Amforth-devel mailing list
> Amf...@li...
> https://lists.sourceforge.net/lists/listinfo/amforth-devel