On Mon, 28 May 2012 16:05:28 +0900
Pavel Holoborodko <pa...@holoborodko.com> wrote:
> Are there any technical difficulties for that? Maybe I can help with
> that if my skills are enough.
The next version of MPIR has a Python based build generator that allows
any of the assembler options to be selected and built.
To build a FAT version would require a significant further development
of this build generator.
Here is an outline of what has to be done (this is probably wrong in
detail as I have not looked at this closely - Bill Hart has given a
better description).
In outline rather then picking up a single set of assembler code, a FAT
build has to pick up a number of sets, one set for each of the
architectures that the FAT build will contain.
Then for each such set, the assembler source code files are renamed and
copied to a FAT directory (for example, the file mpn_add.asm will
become amd_k8_mpn_add.asm, intel_i7_mpn_add.asm, ... one for each
architecture in the FAT build).
Then each of these files is scanned for external symbols and the
files are edited to change these symbols so that, for example, such
symbols in the assembler files for the amd k8 have a prefix such as
amd_k8_ added to them.
Then a C source code table has to be generated for each set of
assebler files, one such table being:
amd_k8_mpn_add
intel_i7_mpn_add
intel_core2_mpn_add
...
and these tables have to be added into the C file that initialises
the FAT build on forst use.
I have thought about doing this but it is quite a long way off at
the moment so if you want to have a go at it, please do. But it is a
very complex process and a pretty complex bit of Python code.
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-devel@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.
