[Beowulf] newbie

Gus Correa gus at ldeo.columbia.edu
Sun May 3 14:16:31 EDT 2009


Thank you Chris, Bill, Greg, and Joe.

Bill Broadley wrote:
> Chris Samuel wrote:
>> In the sense that they have no desire to support
>> competitors hardware, yes. Not really surprising,
> 
> Sure, they could be nice enough to have a flag to disable the check for
> non-intel cpus.  That way intel could avoid the cost of testing/certification
> of AMD cpus and folks that want to take the risk could.  There is a binary
> floating around that patches binaries to avoid the check.  


This is gone:

http://www.swallowtail.org/naughty-intel.html

> Improvements were
> on the order of 0-15% I believe, nobody reported wrong answers as a result.
> 

The (familiar?) Slashdot/2005 discussion:
http://yro.slashdot.org/article.pl?sid=05/07/12/1320202&tid=142&tid=118&tid=123
another on Ubuntu/2008:
http://ubuntuforums.org/showthread.php?t=824046
Inquirer/2007 article:
http://www.theinquirer.net/inquirer/news/1042378/amd-slaps-intel-over-spec-only-compiler#

>> if AMD made compilers I doubt they'd try and do
>> Intel specific optimisations either..
> 
> Well the issue wasn't intel not doing AMD specific optimizations, it was intel
> enabling optimizations that would benefit both CPUs, only when running on intel.
> 

Currently -xW (SSE,SSE2) seems to be the highest architecture-dependent
optimization the Intel compiler allows for Opterons.
Shanghai, Barcelona, and others have more than SSE2, right?

"-xW" is also what AMD recommends when using Intel compilers:
http://support.amd.com/us/Processor_TechDocs/32035.pdf
Compiler Usage Guidelines, p.25:

"3.3.2         Generic Performance Switches
The switches -xW -ipo -O3 -static are generally recommended."

Gus
_______________________________________________
Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.



More information about the Beowulf mailing list