MathGroup Archive 2001

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

Search the Archive

Re: MultiProcessor Kernel.

  • To: mathgroup at smc.vnet.net
  • Subject: [mg31195] Re: MultiProcessor Kernel.
  • From: Jens-Peer Kuska <kuska at informatik.uni-leipzig.de>
  • Date: Wed, 17 Oct 2001 05:35:17 -0400 (EDT)
  • Organization: Universitaet Leipzig
  • References: <200110140811.EAA02428@smc.vnet.net> <9qggk4$f4c$1@smc.vnet.net>
  • Sender: owner-wri-mathgroup at wolfram.com

Hi,

it will *never* be twice as fast. My best speed up was
1.8 times faster. However this will always depend on the
communication overhead of the algorithm.

"David Isaac." wrote:
> 
> Dear Friends and Mathematicians.
> 
> I am running Mathemtica 4.0.1 on a Windows2000 Pro with Dual Processor.
> 
> And I would like to know if any one has tried up the Parallel Computing
> ToolKit.
> 
> Has anyone ran a benchmark sequence on a system with Multiprocessor system
> with the Parallel Computing Toolkit. If so, can anyone share with me the
> results... so I can know
> if it would be atleast :-) twice as fast as working on a Single Processor
> system.
> 
> By the way, Does anyone know about any plans in Wolfram regarding compiling
> the Kernel on a Multiprocessor Compiler which I think would be much more
> efficient then having the Parallel Computing Toolkit.

Visual KAP is not longer distributed, Visual KAP sales has stoped at
31, August 2001, KAP is only for OpenVMS and Tru64 UNIX

SUIF is experimetal and has only a MIPS backend.

What plans did you expect with this compilers ?

Regards
  Jens


  • References:
  • Prev by Date: fonts
  • Next by Date: Re: Sorting and Selecting in MultiLevel Lists?
  • Previous by thread: MultiProcessor Kernel.
  • Next by thread: RE:data files