L3DT Users' Community

Hey Aaron,

I was running a SSE2 optimised version of blender and the renders were 30-40% quicker than normal.
Though im not really sure what it does, it reaped great rewards for me .

Would it be possible for you to implement optimisations for SSE2 and SSE3 in L3DT? and if so what ways would it be beneficial?
Thanks for looking

Sorry Metalliandy,

I don't yet know enough assember to be able to make use of MMX/SSE2/SSE3 extensions safely. Speaking of which, just a few days ago I cam across a set of handy tips regarding assembly optimisation:

1: Don't do it.
2: (for experts) Don't do it yet.

Regardless, I'll put MMX/SSE2/SSE3 optimisations on the to-do list.

As for whether it would be beneficial: definitely maybe. I would think that at least some of the calculations could be re-factored for stream processing, and in such cases there should be a speed-up. However, it's certainly neither a general fix-all nor an easy optimisation: lots of assembly coding.

Cheers,
Aaron.

Ok cool
Im sure a programmer of you status will have no problems
Thanks for adding it to the "to do" list