On Dec 17, 2:09 pm, Cactus <rieman...@gmail.com> wrote:
> On Dec 17, 1:22 am, Jason <ja...@njkfrudils.plus.com> wrote:
>
> > Hi
>
> > A new mul_1 for the core2/penryn is in the trunk , it still runs at the same
> > optimal 4.0c/l but is a 1-way unroll rather than the previous 4-way unroll ,
> > this should save some I-cache.
>
> > Jason
>
> Hi Jason
>
> That is truly brilliant since you massively reduced the register usage
> and this allowed me to use a Windows leaf function.
>
> The more like that you do the better :-)
>
> In the conversion and testing of the Core2 popcount assembler I picked
> up an error in the Python AT&T to Intel translator. If you do any
> work with it be sure to use the latest version.
>
> Brian
In addition to making the mpn_redc_1 changes for Windows, I picked up
several issues during testing:
1. There was a subtle error in the Python AT&T to Intel assembler
syntax translator that I have corrected
2. Issue 1 above caused the Windows popcount core2 assembler to fail.
3. There was a missing export specification in gmp-impl.h that caused
the tests for the generic C DLL to fail in Windows.
I have now built and tested all win32 and x64 builds for static and
DLL libraries for the SVN trunk code using Visual Studio 2010.
Brian
--
You received this message because you are subscribed to the Google Groups
"mpir-devel" group.
To post to this group, send email to mpir-de...@googlegroups.com.
To unsubscribe from this group, send email to
mpir-devel+unsubscr...@googlegroups.com.
For more options, visit this group at
http://groups.google.com/group/mpir-devel?hl=en.
