Benchmarks: AMD64 vs i386 on Dual 246 Opteron

Ken Gunderson kgunders at teamcool.net
Thu Jul 28 15:01:10 GMT 2005


On Thu, 28 Jul 2005 14:13:49 +0200
des at des.no (Dag-Erling Smørgrav) wrote:

> ray at redshift.com writes:
> > While this sounded like a long shot, I loaded FreeBSD 5.4 i386 on
> > the machine and after applying the exact same configuration to the
> > OS, Apache, PHP and MySQL, re-ran the benchmarks.  Much to my
> > surprise, just changing the OS from 64 bit to 32 bit caused the
> > machine to double in speed.  The results are attached in an Excel
> > spreadsheet.  So the exact same machine, running the identical
> > configuration, performed roughly twice as fast when running FreeBSD
> > 5.4 i386 vs FreeBSD 5.4 AMD64.  Something about this seems so wrong
> > to me :-)
> 
> 64-bit code uses up to twice as much CPU cache and twice as many
> memory accesses to do the same work as the equivalent 32-bit code.

Makes a lot of sense once you think about it but counter intuitive to
most people's expectations from their "shinny new 64bit powerhouse"...

Looking at the other side of the coin perhaps we should mention some
situations where 64bit os's outperform 32 bit ones, e.g.
analysis of larger, complex datasets...??

I've talked with a few folks from the SAP crowd and they're absolutely
in love with the Opterons, reporting far superior performance over
Xeons, and are rolling out quads and 8x in lieu of much, much, much more
costly heavy iron counter parts.  They are, however, running on Linux.

Ubench scores, for what they're worth, appear to be comparable between
Opterons and Xeon64T's , so I guess the answer depends on your
application.;-)

-- Best regards,

Ken Gunderson

Q: Because it reverses the logical flow of conversation.
A: Why is putting a reply at the top of the message frowned upon?



More information about the freebsd-amd64 mailing list