Re: 3 byte variables [feature request]

24-bit integers would have to be 16-bit aligned for the cpu to be able 
to load them efficiently, those negating any storage benefits.  So to be 
able to store them efficiently, you would effectively have a struct 
containing three 8-bit bytes (or an array of three bytes).  I think the 
best you could do here is to make a C++ class containing these three 
bytes, with conversion operations to and from int20_t types.  It should 
be possible to do this with code that is almost as efficient as a real 
gcc implementation.

mvh.,

David


On 05/06/13 16:09, Peter Bigot wrote:
> Best to submit this to the gcc project; it's certainly not something that
> would be added through mspgcc.
>
> int20_t was an enhancement I thought useful since the MSP430 CPUX
> architecture enables it, but I don't have reason to believe other MSP430
> toolchains (including the upcoming RH GCC) will recognize it.  Native
> support for it on CPUX still requires 32-bits when it's written to memory,
> though, and that wouldn't change since its inherent in the way the
> instruction set works.
>
> Peter
>
>
> On Wed, Jun 5, 2013 at 8:53 AM, kuldeep dhaka <[email protected]>wrote:
>
>> while working on a program i faced a problem.
>> i need only 20 bit , and having a 32 bit only waste more memory(12 byte),
>> 24bit would much be better.
>> i need a uint24_t variable but gcc dont support it so i have to start with
>> uint32_t.
>> it would be very nice if gcc support such variables.
>> i tried a little program and found that gcc (on i686) dont support it.
>>
>> http://embeddedgurus.com/stack-overflow/2009/06/three-byte-integers/
>>


------------------------------------------------------------------------------
How ServiceNow helps IT people transform IT departments:
1. A cloud service to automate IT design, transition and operations
2. Dashboards that offer high-level views of enterprise services
3. A single system of record for all IT processes
http://p.sf.net/sfu/servicenow-d2d-j