Re: [boost] [boostcon][proto] Suggestion for LIAW session:fixed-pointnumbers

Subject: Re: [boost] [boostcon][proto] Suggestion for LIAW session:fixed-pointnumbers
From: Gruenke, Matt (mgruenke_at_[hidden])
Date: 2011-03-17 20:28:28

Eh?  So what you want is essentially (denormal) floating-point, except that the exponent is specified at compile time?  That's interesting.
 
I haven't had a need for higher range than precision, so it was never a problem for me to include bits all the way down to 2^0.
 
 
Matt
 

________________________________

From: boost-bounces_at_[hidden] on behalf of Ravi
Sent: Thu 3/17/2011 1:28 AM
To: boost_at_[hidden]
Subject: Re: [boost] [boostcon][proto] Suggestion for LIAW session:fixed-pointnumbers

On Wednesday 16 March 2011 19:24:31 Gruenke, Matt wrote:
> > In the system you proposed, what are the template parameters for the
> > fixed- point number with the following range?
> >
> >  { 0, 1, ..., 15 } x 2^{-11}
>
> 11 fractional bits.
>
> > Or, the fixed-point number holding values in the following range?
> >
> >  { 0, 1, ..., 15 } x 2^{11}
>
> 15 integer bits.

Suboptimal at best, and usually unacceptable because it leads to overflow of
the underlying type (int64, for example) much sooner than necessary in complex
expressions. The way it is usually avoided is to specify (in your notation)
the examples above as -7.11 and 15.-11, which, while consistent, is completely
unintuitive and unobvious.

Regards,
Ravi

_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost