Re: How to View Mathematica and Hardcopy Books
- To: mathgroup at smc.vnet.net
- Subject: [mg62402] Re: How to View Mathematica and Hardcopy Books
- From: "Steve Luttrell" <steve_usenet at _removemefirst_luttrell.org.uk>
- Date: Wed, 23 Nov 2005 01:12:19 -0500 (EST)
- References: <firstname.lastname@example.org> <email@example.com>
- Sender: owner-wri-mathgroup at wolfram.com
I used to criticise Mathematica for its poor number crunching performance, but that was several years/versions ago now. For instance, in image processing it is amazing what you can now do with the g and h functions in ListCorrelate (see the Help Browser for details). I used to write C/C++ code, but I don't do it any more because the speed-up C/C++ gives me over Mathematica is either non-existent or at best marginal. The type of coding I am talking about here is large image processing types of applications. The only sorts of operation that one doesn't yet seem to be able to do efficiently in native Mathematica are low-level operations that cleverly manipulate data at the byte or bit level; the sort of thing that is easy to do in C/C++. I appreciate the reason why we users are not allowed to do low-level programming within Mathematica, but it would be really useful if power hungry users were allowed to write low-level code within Mathematica (they take the risk for the consequences, lack of portability, etc). As for MathCode C++. I have had this application since its original release several years ago, in the days when Mathematica's compute speed was not as good as now, so MathCode C++ did indeed give a very useful speed improvement. However, I have stopped using it now because it doesn't give me any speed advantage, except for really messy code that doesn't map well onto Mathematica. Another problem I found was that it translates only a subset of Mathematica to to C++, which is quite frustrating when you want to use Mathematica as it was intended to be used. On the other hand, a very big plus for MathCode C++ is that it produces code that can be distributed and run for free. Steve Luttrell "Steven T. Hatton" <hattons at globalsymmetry.com> wrote in message news:dluq88$ni1$1 at smc.vnet.net... >...DELETED... > For purposes of number crunching Mathematica is quite often the wrong tool > for the job. If you are doing things which involve very intensive mumeric > calculation which are likely to be repeated in form, you are probably much > better off translating your work to something like C++, or FORTRAN. My > understanding is that MathCode C++ can produce executables which run > orders > of magnitude faster than the Mathematica implementation of the same > algorithm >...DELETED...