[GRASS5] AltiVec optimization

Eric G . Miller egm2 at jps.net
Tue Nov 28 21:03:23 EST 2000


On Tue, Nov 28, 2000 at 03:19:53PM -0800, Jeshua Lacock wrote:
> Hello,
> 
> I was looking in to optimizing GRASS for the AltiVec Velocity Engine 
> coupled within the G4 processor (running under MacOS X). One would 
> benefit from 1.5x to 11x fold improvement in performance!
> 
> I have started looking into it and it looks like if I used the 
> Altivec math libraries in place of the current math libraries, the 
> code would then be optimized.

I don't know what this AltiVec thing is, but I think it would hardly
benefit GRASS to link against some "optimized" math library.  Maybe for
some modules.  Mostly the biggest gains will be found by having fast
disks and a fast CPU with sufficient memory to avoid swap.  Disk I/O is
one of the biggest bottlenecks, especially when dealing with any
relatively large rasters.

GRASS has traditionally favored portability over optimizations and
platform dependent features.  It's always better to worry about solving
the problem in a robust fashion and then, when everything works,
thinking about improving performance.

> Unfortunately it looks like fully understanding a robust 
> implementation is just a bit over my head.
> 
> If someone could take a look at: 
> http://developer.apple.com/hardware/ve/vector_libraries.html and give 
> me an idea as to how to proceed with the optimizations, and which 
> code would benefit most from the AltiVec unit, I think that Macs 
> could become the choice platform for GRASS.

-- 
Eric G. Miller <egm2 at jps.net>

---------------------------------------- 
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo at geog.uni-hannover.de with
subject 'unsubscribe grass5'



More information about the grass-dev mailing list