[Beowulf] Woodcrest Memory bandwidth

Kozin, I (Igor) i.kozin at dl.ac.uk
Tue Aug 15 09:31:11 PDT 2006

Oops. I misinterpreted what Keith was saying.
I thought he was trying to justify the "one third" empirical 
rule by referring that every read needs to check against
the other socket thus effectively reducing the bandwidth by 3.
It's obvious that the "three quarters" rule which worked perfectly 
for Opteron and Itanium does not apply to Woodcrest.
Mind you, it did not work too well for Pentium either.
It's bizarre but if you triple every read then
3/(3+3+3+1) is approx 1/3.

> > Good point which makes perfect sense to me.
> > Given that the theoretical maximum is actually 21.3 GB/s
> > the real maximum Triad number must be 21.3/3 = 7.1 GB/s.
> > And that's the best number I've heard of.
> Then how do you explain a dual opteron with two 6.4GB/sec (peak)
> memory system, 12.8GB/sec total per node managing 9-10GB/sec?
> 12.8/3=4.26GB/sec.  People are seeing well over twice that.
> If the opteron manages 75% efficiency or so on a 12.8GB/sec
> memory system, why does woodcrest manage 32% efficiency?

More information about the Beowulf mailing list