Keith C. Perry on 24 Aug 2016 12:46:05 -0700


[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: [PLUG] RAID6 or RAID5+HS?


"SMP means that memory access from any core to any address range is the
same.  NUMA means that regions of memory are closer to some cores than
others.  They never were the same, and still are not."

The comparison was meant to be high level conceptual and not technical  :D

The main point is that NUMA is about about processors, not cores and there are "server" boards that can make use of real NUMA nodes.  Cool stuff to play around with especially in the context of CPUsets, cgroups and other container or VM tech.

~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 
Keith C. Perry, MS E.E. 
Owner, DAO Technologies LLC 
(O) +1.215.525.4165 x2033 
(M) +1.215.432.5167 
www.daotechnologies.com

----- Original Message -----
From: "Rich Freeman" <r-plug@thefreemanclan.net>
To: "Philadelphia Linux User's Group Discussion List" <plug@lists.phillylinux.org>
Sent: Wednesday, August 24, 2016 3:39:09 PM
Subject: Re: [PLUG] RAID6 or RAID5+HS?

On Wed, Aug 24, 2016 at 3:27 PM, Keith C. Perry
<kperry@daotechnologies.com> wrote:
> That's not quite right.  A NUMA system is essentially what SMP used to be.  Its multiple processors (which have multiple cores) so for instance if you type...

SMP means that memory access from any core to any address range is the
same.  NUMA means that regions of memory are closer to some cores than
others.  They never were the same, and still are not.  I just don't
think anybody tends to use SMP, since it typically involves slowing
down everything to the least common denominator (which is worse than
treating a NUMA system as if it were SMP).

>
> numactl --hardware

Won't this just tell you how the kernel is handling the system, and
not how the actual hardware is wired up?

Clearly it is showing in my case that it is handling the system as if
it were SMP.  That doesn't mean that the actual hardware works that
way.  The kernel strictly speaking doesn't have to know that it is
NUMA because the hardware takes care of everything.  If the kernel
sends a CPU off to fetch some other CPU's memory it will just politely
ask the other CPU to interrupt what its doing and lend it a hand, and
wait a while for the answer.  It is slower, but it still works.

I would need to read up a lot more about how AMD HT works to sort it out...

-- 
Rich
___________________________________________________________________________
Philadelphia Linux Users Group         --        http://www.phillylinux.org
Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce
General Discussion  --   http://lists.phillylinux.org/mailman/listinfo/plug
___________________________________________________________________________
Philadelphia Linux Users Group         --        http://www.phillylinux.org
Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce
General Discussion  --   http://lists.phillylinux.org/mailman/listinfo/plug