On Wed, 2008-12-24 at 15:44 +0530, varun shrivastava wrote:
> hi
>         I was trying to use tidy adjustment which has ClutterFixed as
> members.
> So if try to modify the adjustment value > 32767, it overflows.
> When i checked ClutterFixed implementation its a 16.16. 
> 
> So is there a design issue for making ClutterFixed as a gint32
> (16.16)?

GLES uses a 16.16 fixed point notation in the public API, to which we
have to conform.

> Can 't it be made gint64(32.32)?
> Will it give a performance hit?

on ARMs, it's quite possible.

we are currently re-evaluating part of the decisions we made with
regards to fixed points and units, but that is all 1.x material.

ciao,
 Emmanuele.

-- 
Emmanuele Bassi, Intel Open Source Technology Center

-- 
To unsubscribe send a mail to [email protected]

Reply via email to