faster code for neural network

Matthew Bernardini matthew at gandalf.rutgers.edu
Thu Jan 20 11:09:19 PST 1994


> >> Recently I received the code for a neural network. It is written in
> >> standard ANSI-C. However, running it on the Mac gives me a lot of coffee
> >> breaks ( it takes 10 min on a IIci, 3 hrs on an LC ). I was wondering if
> >> anybody out there has experience with this and knows of ways to make the
> >> code faster ( replacing ANSI routines with Mac-based routines.
> >> 
> >> If you are the woman or man with this experience mail me! I really need the
> >> help ( it's all for my master's thesis ).
> >
> >You might want to get  a match coprocessor for the LC.  The ci has one,
> >and that's probably your problem.  (I'm not sure if the LC can take one,
> >though!)  If my memory still works after the two pints I just had, the LC
> >has a 68020 without any FPU.  You might want to upgrade to an LC III and
> >find out if it has an FPU.   I believe neural nets use floats, no?  If so
> >you will most certainly need and FPU if you want decent speed.
> >
> 
> The LC does lack a FPU, but you can get one on a card.  I'm not sure if
> this would help though, because it would depend if your compilier would
> generate FPU code.  I know you can with Think.  You also might need to take
> into account which compiler you are using for this kind of program, there
> seems to be a vairation in efficency between Mac Compilers.
> 
> As for using toolbox calls, using Apple's SANE library might help. I
> haven't had much experience with it, but it would probably take advatage of
> the FPU.
> 
> E-mail me if I can help anymore. (I'll run the program on my Q700 if you
> just need a quick speed test.)
> ___________________________________________________________________________
> "I want to know God's thoughts...the rest are details."
>                                            -- Albert Einstein
> _________________________________________
> Ron Davis                                  rondavis at datawatch.com       
> Datawatch, Research Triangle Park, NC      (919)549-0711
> 
> 


That, or use an AV.

If you are adventurous you could use Aplle's Real Time Architecture to
offload processing to the 66MHZ DSP processor in the AV series.  I know of
some examples of ARTA.

Matthew Bernardini


-----------------------------------------------------------------------------
                        |               Rutgers University Computing Services
Matthew Bernardini      |               Hill Micro/Graphics Center
7804 McCormick          |               Site-Manager
(908) 878-0946          |               017 Hill Center
                        |               (908) 932-3129  (908) 932-4921
-----------------------------------------------------------------------------






More information about the cypherpunks-legacy mailing list