Re: UltraSPARC versus AMD

From: Richard_D_Levine(at)raytheon(dot)com
To: Mike Mascari <mascarm(at)mascari(dot)com>
Cc: Ben <bench(at)silentmedia(dot)com>, pgsql-general(at)postgresql(dot)org, pgsql-general-owner(at)postgresql(dot)org, William Yu <wyu(at)talisys(dot)com>
Subject: Re: UltraSPARC versus AMD
Date: 2005-04-26 13:58:49
Message-ID: OF3FE55DA7.7A59E0FB-ON05256FEF.004BF166-05256FEF.004CCC00@ftw.us.ray.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

Mike Mascari <mascarm(at)mascari(dot)com> wrote on 04/25/2005 09:21:02 PM:

> Richard_D_Levine(at)raytheon(dot)com wrote:
> > In my *utter* lack of enthusiasm over this option, I was gathering
> > ammunition for better hardware. I went to spec.org for speed
comparisons,
> > and sun.com for price comparisons. Sun's *entry* level servers are
more
> > powerful when running AMD CPUs.
>
> Just in case people still hold a bias against CISC processors capable of
> running x86 code as necessarily inferior to more expensive 64-bit RISC
> processors, in spite of the overwhelmingly obvious specint results to
> the contrary, I'd like to offer up this little baby:
>
> http://www.cray.com/products/xt3/index.html
>
> Something everyone should have in their office...

I RTFA. Oh my god. Favorite excerpts:

AMD Opteron Processor
The industry leading Opteron microprocessor offers a number of advantages
for superior performance and scalability.
The Opteron processor's on-chip, highly associative 1 MB cache supports
aggressive out-of-order execution and can issue up to nine instructions
simultaneously. The integrated memory controller eliminates the need for a
separate Northbridge memory controller chip, providing an extremely low
latency path to local memory—less than 60 nanoseconds. This is a
significant performance advantage, particularly for algorithms that require
irregular memory access. The 128-bit wide memory controller provides 6.4
GB/s local memory bandwidth per processor,   or more than one byte per
FLOP. This balance brings a performance advantage to algorithms that stress
local memory bandwidth.

Service PEs run a full Linux™ distribution. Service PEs can be configured
to provide login, I/O, system, or network services.

Scalable Operating System
The Cray XT3 operating system UNICOS/lc is designed to run large complex
applications and scale efficiently to 30,000 processors.

The I/O architecture consists of data RAIDs connected directly to I/O PEs
which reside on the high-speed interconnect. The Lustre file system manages
the striping of file operations across these RAIDs.

>
> Mike Mascari

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Hugues-Vincent Grislin 2005-04-26 14:02:33 out of memory
Previous Message Stephane Bortzmeyer 2005-04-26 13:54:51 Re: Immutable attributes?