Boost logo

Ublas :

From: Gunter Winkler (guwi17_at_[hidden])
Date: 2006-08-01 14:09:50


John Maddock schrieb:
> emijah_s_at_[hidden] wrote:
>
>> Could you please elaborate on the last part? We are using uBlas for a
>> robotics application and the typical size of a matrix tends to be
>> pretty small(around 6x6 typical, 35x35 maximum). So it doesn't really
>> make sense to use atlas which tends to be effective for large
>> matrices.
>>
>
> I wonder have you looked at MTL ? It's optimised for small matices and
> known to be very fast in this case: http://osl.iu.edu/research/mtl/
>
this is only partially true. The performance of small MTL matrices was
quite bad until someone wrote short and lean specializations for small
matrix/vector products. I can give more details (and the files) if
someone is interested. The specializations for a given size were created
by a python script.

mfg
Gunter