Message-ID: <4DC9A5C4.7080507@2ndQuadrant.com>
Date: Tue, 10 May 2011 16:53:24 -0400
From: Greg Smith <greg@2ndQuadrant.com>
User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US;
	rv:1.9.1.16) Gecko/20110307 Icedove/3.0.11
MIME-Version: 1.0
To: pgsql-performance@postgresql.org
Subject: Re: Question processor speed differences.
References: <BANLkTim8h2pLbhLmufQv+vSHBt-ha+VqSQ@mail.gmail.com>
In-Reply-To: <BANLkTim8h2pLbhLmufQv+vSHBt-ha+VqSQ@mail.gmail.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

On 05/10/2011 01:28 PM, Tory M Blue wrote:
> AMD Opteron(tm) Processor 4174 HE  vs  Intel(R) Xeon(R) CPU    E5345  @ 2.33GHz
>
> I'm wondering if there is a performance difference running postgres on
> fedora  on AMD vs Intel (the 2 listed above).
>
> I have an 8 way Intel Xeon box and a 12way AMD box and was thinking
> about migrating to the new AMD box, from the 4 year old Intel box. But
> I wasn't sure if there is some performance stats on AMD multi core
> procs vs the Intels for DB applications?
>    

The real limiting factor on CPU performance on your E5345 is how fast 
the server can shuffle things back and forth to memory.  The FB-DIMM 
DDR2-667MHz memory on that server will be hard pressed to clear 5GB/s of 
memory access, probably less.  That matter a lot when running in-memory 
database tasks, where the server is constantly shuffling 8K pages of 
data around.

The new AMD box will have DDR3-1333 Mhz and a much better memory 
architecture to go with it.  I'd expect 6 to 7GB/s out of a single core, 
and across multiple cores you might hit as much as 20GB/s if you have 4 
channels of memory in there.  Rough guess, new server is at least twice 
as fast, and might even hit four times as fast.

If you have access to both boxes and can find a quiet period, you could 
try running stream-scaling:  https://github.com/gregs1104/stream-scaling 
to quantify for yourself just how big the speed difference in this 
area.  That's correlated extremely well for me with PostgreSQL 
performance on SELECT statements.  If you're going to disk instead of 
being limited by the CPU, none of this matters though.  Make sure you 
really are waiting for the CPUs most of the time.

-- 
Greg Smith   2ndQuadrant US    greg@2ndQuadrant.com   Baltimore, MD
PostgreSQL Training, Services, and 24x7 Support  www.2ndQuadrant.us
"PostgreSQL 9.0 High Performance": http://www.2ndQuadrant.com/books