> If you are using lookup tables to convert floating point to integers, I have
> found that you can use the lower bits to linearly interpolate a much smaller
> number of entries.
>
> You can also eliminate all the negative numbers and all numbers greater than
> 1 and all NAN, making the table 1/4 s
If you are using lookup tables to convert floating point to integers, I
have found that you can use the lower bits to linearly interpolate a
much smaller number of entries.
You can also eliminate all the negative numbers and all numbers greater
than 1 and all NAN, making the table 1/4 size.
On Tue, 2010-10-05 at 09:38 +0200, Soeren Sandmann wrote:
> How must of a performance improvement is that table really over just
> using regular conversions?
It's substantial on at least PowerPC and ARM using GCC, for differing
reasons. I can't remember whether I saw much of a difference on x86 a
Dmitri Vorobiev writes:
> As for the conversion table, what comes into my mind is keeping a
> global table, and accessing it using inter-process communication
> mechanisms such as a shared memory area. Any comments on this?
A pre-generated 'static const' table would certainly be better, but
half
On Sat, Oct 2, 2010 at 4:45 AM, Soeren Sandmann wrote:
>
> We need to figure out how a floating point implementation fits in
> before it makes sense to review the code in detail, especially
> considering that it clearly isn't yet close to being mergable (see
> below).
Soeren, thanks for your feed
Dmitri Vorobiev writes:
> Finally I finished combing through the code that we had, and now the
> floating point-based core pixel pipeline is available from our Git
> server:
>
> http://sandbox.movial.com/gitweb?p=dmvo/pixman.git;a=shortlog;h=refs/heads/float
>
> Please review. Note, however, th
Hello all,
Finally I finished combing through the code that we had, and now the
floating point-based core pixel pipeline is available from our Git
server:
http://sandbox.movial.com/gitweb?p=dmvo/pixman.git;a=shortlog;h=refs/heads/float
Please review. Note, however, that this implementation shoul