What is the right way to use ARM SIMD for iPhone to play vector / matrix?

Question

I am creating a vector / matrix library for a game that uses a SIMD device on an iPhone (3GS or later). How can i do this? I searched about this, now I know several options:

• Speed ​​up Apple's BLAS + LAPACK + ... (iPhone OS 4)
• ARM OpenMAX implementation library
• GCC Auto Vectoring Function

What is the most suitable way for a vector / matrix library to play with?

Answer 1

, , , SIMD-. Neon C intrinsics gcc, , .

Answer 2

, GCC , , !

, iPhone, ARM, NEON SIMD, . , ARM, NEON, .. , ARM, Apple Accelerate framework ARM OpenMAX- , ARM NEON SIMD.

Accelerate, OpenMAX , . 2, , , , ARM OpenMAX , Apple, ARM NEON! .

Answer 3

:

NEON SIMD Apple. http://bulletphysics.org/Bullet/phpBB3/viewtopic.php?t=8490

Answer 4

NEON Mat * Mat Mat * Vec, ASM. Oolong Engine, MIT, :

http://code.google.com/p/oolongengine/source/browse/trunk/Oolong%20Engine2/Math/neonmath/neon_matrix_impl.cpp