Re: [boost] uBLAS formal review

From: Matthew Hurd (matt_at_[hidden])
Date: 2002-06-27 19:06:45

> From: "Fernando Cacciola" <fcacciola_at_[hidden]>
>
> > Not using valarray<> prevents fast 'memcpy' optimizations...
>
> valarray isn't the only route to those optimizations is it? Couldn't one
> equally say "not using std::copy() prevents fast 'memcpy'
> optimizations..."?
<snip>

Well, Duff's Device, (a cute type of unrolling) is faster than memcpy in
some situations... a winner seems compiler and size dependent... perhaps
another argument for platform specific optimisation support, offline or with
runtime support.  Memcopy doesn't go beyond PODs.

http://www.cuj.com/experts/1910/alexandr.htm?topic=experts&topic=experts
  -- "If you are confused, so am I. It seems hard to decide which method is
best across all compilers and all data amounts. "

matt.