vectormath SSE optimisations are MSVC only

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Line 41 of 
http://code.google.com/p/bullet/source/browse/trunk/Extras/vectormathlibrary/inc
lude/vectormath/SSE/cpp/vectormath_aos.h includes "intrin.h"
2. The file "intrin.h" appears to only exist in MSVC
3. Therefore GCC users do not get SEE optimisations and have to import the 
scalar implementation directly.

What is the expected output? What do you see instead?
I'd expect the code to import "xmmintrin.h".  However, when doing this other 
errors are reported such as
"vectormathlibrary/include/vectormath/cpp/../SSE/cpp/vectormath_aos.h:78:8: 
error: ‘__forceinline’ does not name a type"
So it looks like "__forceinline" is MSVC specific too.

It looks like the update from r1056 introduces lots of MSVC dependencies. 

What version of the product are you using? On what operating system?
SVN HEAD on Fedora 14.

Please provide any additional information below.
-mno-sse is not a work around.  It should tell the compiler to undefine 
__SSE__, however somewhere an sse register is still returned.

Original issue reported on code.google.com by aidan.de...@gmail.com on 13 Feb 2011 at 11:03

• Merged into: #383

[GoogleCodeExporter]

Looks like there are a set of fixes on the forum

http://bulletphysics.org/Bullet/phpBB3/viewtopic.php?f=9&t=5091&p=18564&hilit=GC
C+vectormath#p18564

It would be great to have these applied to HEAD.

Original comment by aidan.de...@gmail.com on 13 Feb 2011 at 11:23

[GoogleCodeExporter]

Looks like I've just dup'ed #383

Original comment by aidan.de...@gmail.com on 13 Feb 2011 at 11:26

[GoogleCodeExporter]

Thanks for the feedback.

I hope there will be some gap in my schedule to process this patch soon.

Original comment by erwin.coumans on 15 Feb 2011 at 11:17

[GoogleCodeExporter]

Original comment by erwin.coumans on 27 Feb 2011 at 3:58

• Changed state: Duplicate