[Bf-committers] [PATCH] Remove all MTC_* matrix/vector references and replace with BLI_arithb

Campbell Barton ideasman42 at gmail.com
Sun Feb 17 16:39:46 CET 2008


Just discussed this the patch is too late for inclusion in the next
release of blender (2.46) However cleanup patches are good, could you
submit to the patch tracker and this can be reviewed+applied after
2.46 release.

On Feb 14, 2008 8:14 PM, Konstantinos Margaritis <markos at codex.gr> wrote:
> Hi all, (new to the list)
>
> As discussed with some developers on #blendercoders (JesterKing, sorry
> mate, don't know your real name :-), I've made a patch to remove all
> references to the (supposedly obsolete) MTC_* matrix/vector
> functions, and I also removed some unused mul_* functions present in
> BLI_arithb.c (these aren't even prototyped, I guess they were
> leftovers from some previous versions). Blender compiles fine with
> this patch, and the MTC_*.{c,h} files can now be removed if you like.
> I've tested the resulting binary with some test scenes and it worked
> fine with no crash/core dump/etc.
>
> In the patch, I've included a CHANGELOG.markos file (just for
> reference, I don't expect it to be included of course), where I
> outline the functions removed and a substitution map listing the old
> functions and their replacements.
>
> The reason for this substitution is that I'm working on SIMD
> optimizations in Blender (SSE/AltiVec and Cell as an ultimate goal),
> and having a clean matrix/vector API to work on is a necessary step.
> Next step is to take out these matrix function in a separate library,
> which will hold all optimized functions, and then it will link with
> the the main program.
>
> Next step is to make some of the code SIMD-friendly, eg make the
> float[4][4] to be forcibly aligned on a 16-byte boundary, pad the
> vectors (float[3]) to be in fact float[4] (again aligned on a 16-byte
> boundary), etc. As discussed with JesterKing and others on IRC, there
> seems to be no objection in me doing this, so I'll proceed and I'll
> let you know of any patches/results.
>
> For me, the ultimate goal is the port to Cell, but I'll just can't
> avoid making some core changes to Blender. I just hope you have no
> objections in this.
>
> Thanks again for a wonderful program :-D
>
> --
> Konstantinos Margaritis
> CODEX
> http://www.codex.gr
> markos at codex.gr
> markos at debian.org
>
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers
>
>


More information about the Bf-committers mailing list