MathGroup Archive 2006

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Illusory Multicore Support in 5.2?

  • To: mathgroup at smc.vnet.net
  • Subject: [mg66166] Re: Illusory Multicore Support in 5.2?
  • From: Mark Laney <mlaney at exnconsulting.com>
  • Date: Tue, 2 May 2006 02:43:29 -0400 (EDT)
  • Sender: owner-wri-mathgroup at wolfram.com

Contrablue wrote:
> Greetings,
>
> One of the reasons I upgraded to Mathematica 5.2 was to enjoy the
> benefits of multicore support. However, despite my running 5.2 on a
> Windows XP x64 dual Opteron system, I have yet to see evidence that
> more than one kernel thread runs in parallel (the evidence, on my
> system, would be to see processor utilization rise above 50% while only
> the Mathematica kernel was active). This is despite trying quite a few
> evaluations of code that "should" -- according Wolfram documentation --
> initiate parallel execution on my system (numerical dot products, etc.)
>
> Please note, my system fully supports true multithreaded execution on
> two CPUs, as evidenced by 100% CPU utilization on relevant Sandra
> benchmarks and on the new multithreaded POV-Ray releases for AMD64.
>
> Has anyone verified multithreaded kernel operation of Mathematica 5.2
> on a dual processor Windows platform? If so, could you post a snippet
> of your Mathematica expressions here?
>
> Thanks much!
>
>>Try using SingularValueDecomposition on a sufficiently >>large matrix

>>Pratik
>>Wolfram Research, Inc.
>>Tech Support

As a new user of Mathematica, I am very disappointed by this flip answer. Why was some sample code not provided   as requested? What about some broader information for those of us with similar issues? What is the issue with 5.2 with respect to Contrablue's scenario? is there a fix for that? is there a fix coming in an upcoming release? Which release is targeted? What is the improvement? I could go on....


  • Prev by Date: Re: Conditions with Statistical Functions
  • Next by Date: Problem using Solve with many unknown variables
  • Previous by thread: Command completion maxes out CPU
  • Next by thread: Problem using Solve with many unknown variables