[Beowulf] Re: dual core Opteron performance - re suse 9.3
kus at free.net
Wed Jul 13 07:58:32 PDT 2005
In message from Mark Hahn <hahn at physics.mcmaster.ca> (Tue, 12 Jul 2005
17:40:58 -0400 (EDT)):
>> >there are only 4 slots on the Tyan 2875 (I had mistakenly reported
>> I'm not seeing anywhere at Tyan an indication this board can take
>> of NUMA.
>node interleave is meaningless for the 2875, since the board only has
>memory attached to one CPU. while the bios probably does include the
>ACPI table that informs the kernel's k8-numa code, it's moot, since
>there's no way to arrange cpu-proc affinity to minimize non-local
>accesses. (except by not using the second socket, of course!)
>I'd expect NUMA support to make more of a difference on 4-socket
>since on them, a process can be >1 hop away from memory. on a
>system, it's probably still worth doing, but can't be all that
>naturally, latency-sensitive codes (big but with poor locality) will
>show a bigger difference.
>> >Bank interleaving "Auto"
>I tried to measure this on a dual, and couldn't. it's hard to see,
>based on the low-level hardware specs, why it would matter much.
>yes, bank interleave should reduce the amount of time waiting on
>bank misses, but it's certainly not visible to Stream.
>> >Node interleaving "Auto"
>turning this to on essentially defeats NUMA; it could be the right
>for some codes/systems, since it means that no process has any
>affinity for a particular socket.
The main practical result for dual-CPU single core Opteron server
is: if I turn "Node interleaving" to ON, STREAM results will be much
more worse. Therefore some applications will work more slow.
But what will be in the case of setting it to "AUTO" ? "Who"
(and how) will solve about real setting ?
>Beowulf mailing list, Beowulf at beowulf.org
>To change your subscription (digest mode or unsubscribe) visit
More information about the Beowulf