Re: More LTC timings...
From: Sander Vesik (sander_at_haldjas.folklore.ee)
Date: 06/24/03
- Next message: Sander Vesik: "Re: More LTC timings..."
- Previous message: Sander Vesik: "Re: More LTC timings..."
- In reply to: Shill: "Re: More LTC timings..."
- Next in thread: Erwann ABALEA: "Re: More LTC timings..."
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Date: Tue, 24 Jun 2003 19:38:15 +0000 (UTC)
Shill <nobody@example.com> wrote:
>> Could I know the name of one 32-bit hardware (if
>> none, then any other type) where unsigned char is
>> not 8 bits? Thanks.
>
> On SHARC systems, where a "byte" (defined by ANSI C
> to be the size of a char) is 32 bits, not 8 bits,
> this implementation exposed a portability problem
> for MPI codes.
>
> http://www.crpc.rice.edu/NHSEreview/MPI/node48.html
> http://mail.gnu.org/archive/html/lwip-users/2003-01/msg01481.html
> http://www.analog.com/processors/processors/sharc/index.html
>
> I'd look into embedded processors and DSP.
>
> Wasn't the Alpha only capable of dealing with 64-bit quantities?
> Thus a char would be 64 bits wide, right?
>
Early models of alpha did not have byte load or store instructions.
This does not mean char datatype would not be 8 bits wide - just that
load/store char are not automic and can have some interesting side effects
in threaded code that was not carefully written.
-- Sander +++ Out of cheese error +++
- Next message: Sander Vesik: "Re: More LTC timings..."
- Previous message: Sander Vesik: "Re: More LTC timings..."
- In reply to: Shill: "Re: More LTC timings..."
- Next in thread: Erwann ABALEA: "Re: More LTC timings..."
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Relevant Pages
|