[gmx-users] pentium III vs Pentium IVs and Intel compiler benchmarks

[Lynne E. Bilston]

Hi all,

Following up on Mark and Erik's comments about intel/AMD CPUs and 
compilers, you may find the following benchmark site useful.

http://www.dl.ac.uk/TCSC/disco/Benchmarks/comp_chem.html

There are several subdirectories iin that same directory with different ppt 
presentations they've put together too, which graph the various benchmarks.

It has some computational chemistry benchmarks, even some that are specific 
to MD (not gromacs unfortunately), which are a bit old, but include some 
athlons and intel p4 machines (along with every other kind of machine 
imaginable), with various compilers. You'll have to hunt around a bit to 
find all the info.

It seems for the MD-specific benchmarks, the Athlons definitely do much 
better for an equivalent speed. They have an Athlon 1400 killing a 1.5GHz 
P4.  Of course, this doesn't take into account the assembly issue, but it 
may be interesting nonetheless.

The comparison slides are 
http://www.dl.ac.uk/TCSC/disco/Benchmarks/ppoint_oct01/sld026.htm and
http://www.dl.ac.uk/TCSC/disco/Benchmarks/ppoint_oct01/sld042.htm

For those who do other things besides MD, they've looked at various other 
computational chemistry tasks too.

Mind you, the more I look at all the differences, the more confused I get 
between instruction sets, motherboards etc.

I'm leaning toward getting dual athlons right now, assuming that Erik's 
right in saying that double precision isn't critical for my type of 
simulations (similar to yours, by the sound of it, Mark!)

Cheers,

Lynne
________________________________________________________________________________
Lynne E. Bilston, PhD
Senior Scientist, Prince of Wales Medical Research Institute, and
Conjoint Associate Professor, Faculty of Medicine, University of New South 
Wales
Barker St, Randwick, NSW 2031 Australia
Tel: +61-2-9382-7924	Fax: +61-2-9382-2643
________________________________________________________________________________