Andrey Kharuk wrote:
> Hi Jeff,
>
> Thank you for information. I'll continue to investigate that problem
> and will make some more experiments with another MPI implementations.
> I haven't got yet results for 8nodes/16cpus. I'm going to do that
> later.
>
> Regards,
> Andrey
>
>
>>>>jsquyres_at_[hidden] 20/11/2005 6:58:47 a.m. >>>
>
> Sorry for the delay in replying here -- we were totally swamped with
> SC.
>
> In general, I have not heard good experiences with hyperthreading in
>
> HPC. I'm more of a software person than a hardware person, so I
> won't
> try to delve any deeper than that; you might want to google around to
>
> see what other people's experiences are with hyperthreading processes
>
> that run at/close to 100% of the CPU. In short, based on my
> [admittedly limited] experience with hyper threading, I would expect
>
> lower performance for your 32 process/8 node run vs. your 16
> process/8
> node run.
Unless you have one of the most recent kernels (e.g. SuSE 9.3 or
EL4_U2), you are likely to require HT disabled in BIOS to see full
performance with multiple CPUs per node. We have seen cases where HT
enables a 10% performance improvement when running MPI on a single node,
but it's difficult to sustain that on a cluster, even when all nodes
have sufficient RAM. Doubling up processes in hope of taking advantage
of HT also exposes faulty memory more frequently.
|
