[Beowulf] CCL:Question regarding Mac G5 performance (fwd from mmccallum@pacific.edu)

[Eugen Leitl]

----- Forwarded message from Mike McCallum <mmccallum at pacific.edu> -----

From: Mike McCallum <mmccallum at pacific.edu>
Date: Wed, 19 May 2004 10:49:14 -0700
To: Joe M Leonard <jle at world.std.com>
Cc: chemistry at ccl.net
Subject: CCL:Question regarding Mac G5 performance
X-Mailer: Apple Mail (2.613)


On May 19, 2004, at 05:06, Joe M Leonard wrote:

>Folks, has anybody run QM or MD benchmarks on a dual-processor
>2GHz G5 that they care to share?  I'm particularly interested
>in how they stack up to the newest/newer Opteron and Pentium4
>machines.  If you've worked with the newer Mac's, (a) is it
>possible to get good, useful 2x processor performance for
>double-precision code?  (b) what compilers are required to get
>(a)?  (c) is Altivec useful for double-precision code, or
>merely for int/float work?
>

I had done some comparisons between price/performance, and I found dual  
G5s to be at or near the best in price/performance, especially if  
things are recompiled with the IBM compilers (between 8-10 % speed  
increase over pre-compiled (with apple gcc) versions of NAMD2, and  
using standard gcc for charmm).  I would expect that things like  
gaussian03 run very well (I believe gaussian uses the IBM compilers for  
macOS).  For MD, the speedup seems to be due to the on-chip square root  
eval.

The built-in Gigabit enet is attractive, also, as charmm and NAMD scale  
very well with gigabit, and it makes myrinet less price-effective (when  
used on any platform, wintel included, see
http://biobos.nih.gov/apps/charmm/charmmdoc/Bench/c30b1.html for  
example).  I decided that dual G5 xserve cluster nodes with gigabit  
switches were much more cost-effective for me than any other processor,  
especially any high-bandwidth specialty comm method (apple's gigabit  
has a pretty low latency also).

Additional considerations for us were the BSD environment which is more  
secure than windows, and the OS is arguably more stable and supported  
than a linux (though you can pretend to be linux with the fink layer).   
Being much more familiar and happy with OS X than windows or linux  
sealed it.

It is my impression that opterons, PIVs, G5s all have their advantages,  
and disadvantages, so your own comfort level may be a non-trivial  
consideration.  I don't feel there is any net performance penalty, and  
often a benefit from using G5s.


>A quick glance at apple.com suggests these machines go for about
>$3200 USD, which isn't that different from numbers gained from a
>quick glance at dell.com.  I know IBM chips were number-crunchers
>in the good old days, and I assume that's still the case now (which
>is why I'm looking to CCL for guidance :-).

I got my current dual 2.0 G5 for $2200.  This is acadmic pricing, and  
I'm sure your rep would give you a similar price, especially now that a  
new processor version is close to (?) production.  I was looking at at  
least $2500 for a similarly configured opteron node (w/o gigabit,  
though).

Hope this helps,

Mike
>
>Thanks in advance!
>
>Joe Leonard
>jle at theworld.com
>
>
>To send e-mail to subscribers of CCL put the string CCL: on your  
>Subject: line
>
>Send your subscription/unsubscription requests to:  
>CHEMISTRY-REQUEST at ccl.net
>HOME Page: http://www.ccl.net   | Jobs Page: http://www.ccl.net/jobs
>
>-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- 
>+-+
>
>
>
>
>
>
--
C. Michael McCallum                         
http://chem.cop.uop.edu/cmmccallum.html
Associate Professor
Department of Chemistry, UOP
mmccallum .at. pacific .dot. edu                (209) 946-2636 v  /  
(209) 946-2607 fax



-= This is automatically added to each message by the mailing script =-
To send e-mail to subscribers of CCL put the string CCL: on your Subject: 
line
and send your message to:  CHEMISTRY at ccl.net

Send your subscription/unsubscription requests to: CHEMISTRY-REQUEST at ccl.net 
HOME Page: http://www.ccl.net   | Jobs Page: http://www.ccl.net/jobs 

If your mail is bouncing from CCL.NET domain send it to the maintainer:
Jan Labanowski,  jlabanow at nd.edu (read about it on CCL Home Page)
-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+





----- End forwarded message -----
-- 
Eugen* Leitl <a href="http://leitl.org">leitl</a>
______________________________________________________________
ICBM: 48.07078, 11.61144            http://www.leitl.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE
http://moleculardevices.org         http://nanomachines.net
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: not available
Url : http://www.scyld.com/pipermail/beowulf/attachments/20040519/561263ef/attachment.bin