Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > comp.soft-sys.math.mathematica > #3430

Re: C code generation

From Richard Fateman <fateman@cs.berkeley.edu>
Newsgroups comp.soft-sys.math.mathematica
Subject Re: C code generation
Date 2011-07-01 00:45 +0000
Organization Steven M. Christensen and Associates, Inc and MathTensor, Inc.
Message-ID <iuj5bi$np3$1@smc.vnet.net> (permalink)
References <iug5os$9g0$1@smc.vnet.net>

Show all headers | View raw


On 6/29/2011 2:34 PM, Shane Di Dona wrote:
> Is there a way to make Mathematica generate C code for functions such as Position[] and MemberQ[]?

  I am writing a program for CUDAlink and would like not to have to 
reinvent the wheel (in C)

on these somewhat basic functions.
>

1. Code for a similar program for position and member can be found in 
other languages, though really, it should be easy to write yourself.
2. A full implementation of Position and MemberQ requires that you allow 
for the possibility of searching for a pattern, and so you would have to 
(re) implement Mathematica's matching.  I doubt you want to do that.
3. While I am not familiar with CUDAlink, it seems to me unlikely that 
copying a (long?) list into a graphics processor to test for membership 
is a really bad idea. A worse idea would be, I suppose, NOT copying the 
list into the graphics processor.  My suggestion: Read up on hash 
tables. Suggestion 2: If you are in a hurry, don't use Mathematica.

RJF

Back to comp.soft-sys.math.mathematica | Previous | NextPrevious in thread | Find similar | Unroll thread


Thread

C code generation Shane Di Dona <stdidona@ncsu.edu> - 2011-06-29 21:34 +0000
  Re: C code generation Richard Fateman <fateman@cs.berkeley.edu> - 2011-07-01 00:45 +0000

csiph-web