From: Andi Kleen <ak@suse.de>
To: Christoph Lameter <clameter@engr.sgi.com>
Subject: Re: [PATCH 1/5] cpuset memory spread basic implementation
Date: Tue, 7 Feb 2006 18:51:38 +0100
User-Agent: KMail/1.8.2
Cc: Ingo Molnar <mingo@elte.hu>, steiner@sgi.com, Paul Jackson <pj@sgi.com>,
       akpm@osdl.org, dgc@sgi.com, Simon.Derr@bull.net,
       linux-kernel@vger.kernel.org
References: <20060204071910.10021.8437.sendpatchset@jackhammer.engr.sgi.com> <200602071828.49370.ak@suse.de> <Pine.LNX.4.62.0602070941040.24841@schroedinger.engr.sgi.com>
In-Reply-To: <Pine.LNX.4.62.0602070941040.24841@schroedinger.engr.sgi.com>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200602071851.39330.ak@suse.de>
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 903
Lines: 22

On Tuesday 07 February 2006 18:42, Christoph Lameter wrote:

> > > and therefore there are only small latencies  
> > > involved. NUMA only gives small benefits.
> > 
> > That's also not true. Everytime I get memory placement for 
> > process memory wrong users complain _very_ loudly and there 
> > are clear benefits in benchmarks too.
> 
> What are the latencies in an 8 way opteron system? I.e. Local memory, next 
> processor, most distant processor?

The NUMA factor is surprisingly good because of the way the cache coherency
works even the local memory access gets slower with more
nodes @) iirc it's <3. Worst case latency tends to be <200ns.

-Andi
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/