[DGD] Full double precision floats?

[Raymond Jennings]

Actually DGD uses "truncated doubles" by chopping off the last 16 bits.  It
uses 6-byte floats.

What I'm thinking of doing is having it go up to 64-bit floats that


On Thu, Apr 25, 2013 at 9:58 PM, Littlefield, Tyler <tyler at tysdomain.com>wrote:

> I'm not sure of the specifics as to why DGD's floats would be 36 as
> opposed to 32. I know that libraries tend to use doubles when higher
> precision is actually required, but on 32 bit systems pushing them to the
> stack takes more cycles than it would to push a 32 bit (it has to push it
> in halves). I don't even know if this was the idea behind DGD using float
> as opposed to double, but it's generally a tradeoff between performance and
> the need for a higher precision float value.
>
> On 4/25/2013 10:27 PM, Raymond Jennings wrote:
>
>> Hey, just curious if there's any cost/benefit factors not immediately
>> obvious if DGD were to expand its floats from 36 bits to the full 64 bits.
>>
>> I think it could gain on precision and if converted properly wouldn't
>> cause
>> anything to choke in conformant code that checks float.h
>> ______________________________**______________
>> https://mail.dworkin.nl/**mailman/listinfo/dgd<https://mail.dworkin.nl/mailman/listinfo/dgd>
>>
>
>
> --
> Take care,
> Ty
> http://tds-solutions.net
> The aspen project: a barebones light-weight mud engine:
> http://code.google.com/p/**aspenmud <http://code.google.com/p/aspenmud>
> He that will not reason is a bigot; he that cannot reason is a fool; he
> that dares not reason is a slave.
>
> ______________________________**______________
> https://mail.dworkin.nl/**mailman/listinfo/dgd<https://mail.dworkin.nl/mailman/listinfo/dgd>
>