Fwd: Re: Opinion question (Core2 Duo)

[Francesco Pietra]

I have just completed unrestricted MP2/6-31G* energy procedure (after
DFT/M05-2X) for a 98-atoms (first row) molecule in 19 hours with four-node
amd64 Debian amd64 etch, NWChem suite. I understand (if I understand correctly)
from your email that should I have had Core 2 I would not have had the time to
take a coffee in between launching the MP2 procedure and getting the
computational results. Interesting.

Thanks

francesco pietra

 --- Jo Shields [EMAIL PROTECTED] wrote:
 
  Subject: Re: Opinion question (Core2 Duo)
  From: Jo Shields [EMAIL PROTECTED]
  To: debian-amd64 debian-amd64@lists.debian.org
  Date: Tue, 18 Sep 2007 15:51:34 +0100
  
  
  On Tue, 2007-09-18 at 10:41 -0400, Lennart Sorensen wrote:
   On Tue, Sep 18, 2007 at 03:08:48PM +0100, Jo Shields wrote:
Or more? Buy an Altix! ;)
   
   Ehm, well the Altix uses either the itanium (why would anyone want that
   crap) or a dual socket core 2 based cpu.  That hardly matches a 4 or
   more cpu opteron server.
  
  Let's assume I have large examples of both IA64 and AMD64. Plus further
  benchmark data we collected ourselves.
  
  IA64 is fast, for floating point code. On paper, it offers the same
  per-core-per-Hz FLOP count as Core (twice that of AMD64). And in
  practise, Altix scales, whilst the competition, well, doesn't. In our
  benchmarks, IA64 was not only faster per-GHz than POWER5 or AMD64, but
  faster in absolute terms too, with an 8-way test absolutely dominated by
  a 1.6-GHz-Montecito-based Altix, whilst AMD64 didn't even register a
  pulse.
  
  However, for IA64, compiler choice is key - using GCC to compile test
  code isn't just crippling the system, it's throwing away hundreds of
  thousands (if not more) of investment
  
   SGI has nothing of any real interest.  No wonder they went under not
   that long ago. :)
  
  They've got SMP machines that don't choke at 4 cores. For some
  applications, that's of great interest.
  
  -- 
   __
  / Jo Shields [EMAIL PROTECTED] \
  | Systems Manager,  |
  \ Oxford Supercomputing Centre  /
   ---
 \   ,__,
  \  (oo)___
 (__))\
||--|| *
  
  
  -- 
  To UNSUBSCRIBE, email to [EMAIL PROTECTED]
  with a subject of unsubscribe. Trouble? Contact
 [EMAIL PROTECTED]
  
  
 
 
 
  


 Check out the hottest 2008 models today at Yahoo! Autos.
 http://autos.yahoo.com/new_cars.html
 



  

Don't let your dream ride pass you by. Make it a reality with Yahoo! Autos.
http://autos.yahoo.com/index.html
 



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]

Re: Fwd: Re: Opinion question (Core2 Duo)

[Jo Shields]

On Wed, 2007-09-19 at 00:08 -0700, Francesco Pietra wrote:
 I have just completed unrestricted MP2/6-31G* energy procedure (after
 DFT/M05-2X) for a 98-atoms (first row) molecule in 19 hours with four-node
 amd64 Debian amd64 etch, NWChem suite. I understand (if I understand 
 correctly)
 from your email that should I have had Core 2 I would not have had the time to
 take a coffee in between launching the MP2 procedure and getting the
 computational results. Interesting.

Xeon/Core2 cheats slightly - it has combined units for both adding and
multiplying on the chip - meaning if you do a multiplication immediately
followed by an addition, then it'll happen in 1 cycle instead of 2,
hence the twice as many FLOPs thing. Honestly, in desktop
applications, it doesn't matter much - but in matrix operations, as used
by most chemistry packages such as Gaussian or NWChem, it makes a full
impact.

So yes, you'd probably see a ~90% speed boost moving from an AMD64 to a
Core2 of identical clock speed.

If you have access to the NWChem source (you might not, I don't think we
have any login credentials lying around to check with) you would see
even bigger improvements with a commercial compiler and math library -
you may find a cheaper option to improve performance than buying a new
Core2 rig is to buy Portland C (or Pathscale C), and link against the
free AMD Core Math Library instead of conventional open-source
BLAS/LAPACK routines.

-- 
 __
/ Jo Shields [EMAIL PROTECTED] \
| Systems Manager,  |
\ Oxford Supercomputing Centre  /
 ---
   \   ,__,
\  (oo)___
   (__))\
  ||--|| *


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]