gcc321 and doubles and c++

gcc321 and doubles and c++

[Chris Croswhite]

How are doubles represented in memory under c++?  The reason I ask, is that
I need to get  the value of a double striped of the mantissa and then split
into two 32 bit words inorder to perform bitwise operations on the
individual parts (as bitwise operations only work up to long ints).  BTW, I
am tweaking for better math performance.

TIA,
Chris

Re: gcc321 and doubles and c++

[John Love-Jensen]

Hi Chris,

The representation of a double varies from platform to platform.  There is
no canonical (platform agnostic) format.

Some platforms (perhaps yours) represents doubles in IEEE 754 format.  Which
can be affected by your platform's endian.

qv: http://grouper.ieee.org/groups/754/
qv: http://research.microsoft.com/~hollasch/cgindex/coding/ieeefloat.html

Sincerely,
--Eljay

Re: gcc321 and doubles and c++

[Nathan Sidwell]

Chris Croswhite wrote:
> How are doubles represented in memory under c++?  The reason I ask, is that
> I need to get  the value of a double striped of the mantissa and then split
> into two 32 bit words inorder to perform bitwise operations on the
> individual parts (as bitwise operations only work up to long ints).  BTW, I
> am tweaking for better math performance.
It uses the underlying hardware representation.

nathan
-- 
Nathan Sidwell    ::   http://www.codesourcery.com   ::     CodeSourcery LLC
          The voices in my head said this was stupid too
nathan@codesourcery.com : http://www.cs.bris.ac.uk/~nathan/ : nathan@acm.org

Re: gcc321 and doubles and c++

[Zane Dodson]

On Tue, Feb 18, 2003 at 10:35:13PM -0800, Chris Croswhite wrote:
| How are doubles represented in memory under c++?  The reason I ask, is that
| I need to get  the value of a double striped of the mantissa and then split
| into two 32 bit words inorder to perform bitwise operations on the
| individual parts (as bitwise operations only

frexp(3) and modf(3) may do part of what you need portably.

As has already been indicated in other replies, any bit manipulations
will require that you determine the byte ordering and alignment for your
platform.  If you choose to use GNU Autoconf for your application, there
are already some builtin tests for these platform-specific issues.

Best regards,

-- 
Zane Dodson
zdodson@zdodson.com