MathGroup Archive 2006

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

Search the Archive

Re: Comparison of Mathematica on Various Computers

  • To: mathgroup at smc.vnet.net
  • Subject: [mg64413] Re: Comparison of Mathematica on Various Computers
  • From: carlos at colorado.edu
  • Date: Thu, 16 Feb 2006 03:05:21 -0500 (EST)
  • References: <200602090745.CAA19418@smc.vnet.net> <200602100713.CAA15024@smc.vnet.net> <dsk8m8$i9l$1@smc.vnet.net> <dsmueo$da0$1@smc.vnet.net> <dspft3$cg1$1@smc.vnet.net>
  • Sender: owner-wri-mathgroup at wolfram.com

Why look at compilers and libraries?  Note that the better performance
of AMD chips is consistent across different OS (Window, Linux,
Solaris).
Pressumably different compilers and libraries are being used.

We notice similar across the board superiority in testing  a suite of
applications to benchmark supercomputers, as noted in a previous post.

IMO the main reason is hardware: a better memory controller
integrated into the 64-bit chipset.  Mathematica tends to be memory
traffic intensive because of the list structures.  Our tests on 128-CPU

clusters were also memory traffic bound for a different reason:
each processor talks to each other through a VM system.


  • Prev by Date: Re: Joining List of Lists
  • Next by Date: Re: Mathematica:recursion with 2 arguments?
  • Previous by thread: Re: Comparison of Mathematica on Various Computers
  • Next by thread: Mathlink