Re: [OT: This is not the thread you're looking for. You can goabout your business. Move along, move along.]

From: Mok-Kong Shen (mok-kong.shen_at_t-online.de)
Date: 07/22/03

  • Next message: Gregory G Rose: "IVs versus Nonces" 
    Date: Tue, 22 Jul 2003 23:23:52 +0200

    Simon G Best wrote:
    >
    > Mok-Kong Shen wrote:
    > >
    > > Simon G Best wrote:
    > >
    > >>So, you're still ignoring the stuff about interprocessor communications,
    > >>etc.
    > >
    > > There are interprocessor communication when a code
    > > runs on a number of processors and there are data to
    > > be exchanged, i.e. I/O. That means the theoretically
    > > achievable performance is generally not achievable by
    > > such a code and in fact most codes can't achieve the
    > > peak performance. But, though a good mix of different
    > > codes, the I/O time of one taks could be concurrent to
    > > the processing time of another task, thus making the
    > > average performance better. This is similar to the
    > > case where one has only a single processor with
    > > a number of tasks doing time-sharing.
    >
    > If a supercomputing job is saturating, say, interprocessor
    > communications resources, running additional supercomputing jobs
    > wouldn't help. But there's no real wastage of supercomputing resources
    > in such a situation anyway, as the important supercomputing resource of
    > interprocessor communications is not being wasted at all (and it's those
    > kinds of resources within supercomputers that make supercomputers special).
    >
    > More generally, *it's not just the processors by themselves* that carry
    > out computations in things like massively-parallel,
    > highly-interconnected multiprocessor supercomputers. Things like
    > interprocessor communications resources play an important part in the
    > actual computations. It's *those* resources, *not* the individual
    > processors themselves, that make supercomputers special. Consequently,
    > measuring performance *only* in terms of what the processors
    > *individually* do themselves is insufficient for determining whether or
    > not those supercomputing resources are being wasted.

    As I said, this is a load-balancing problem. The issue
    is well-known in CS. How could a machine both be
    'saturated' and run at 4% of its peak performance?
    There are known problems of getting good performance
    in parallel computing that drive down the average
    figure in comparison to the case of mono-processor.
    But that extreme 4% is in my view clearly due to the
    exclusiveness of the class of problems that are allowed
    (for whatever reasons) to be run on the hardware. If
    you have some acquaintances working in such institutions,
    have a talk with them and you would likely be told of
    the general rules of selection (admission) of jobs
    (problems) that are to be run there. (In fact, this
    is nothing uncommon. For example, I don't participate
    in projects of internet-wide computing like factorization
    of huge numbers etc. for diverse personal reasons, hence
    my PC is by far not utilized to its full capacity.)

    >
    > {Once upon a time, you'd said:]
    > >>>>>In this case I do think it is bad. That's why I
    > >>>>>repeatedly said that one could use a better mix of jobs
    > >>>>>to enhance the average performance. But I don't see
    > >>>>>how this opinion could constitute a self-contradiction
    > >>>>>as you claimed.
    > [...]
    > > I 'confirmed'?? Where?? You claimed expressly that my
    > > 'previous posts in this thread clearly indicated otherwise'.
    > > Doesn't that mean what I wrote in the quote above is
    > > (in your view) contradictory to the (said) 'previous posts'?
    >
    > Yes, but it wasn't a "self-contradiction". A *self*-contradiction is
    > something that contradicts *itself*. The "quote above" didn't
    > contradict itself, and I never claimed that it contradicted itself.
    >
    > As you've got yourself horribly confused, I'll disregard the rest...

    Hmm, I meant 'self-contradiction' in the sense that
    one contradicts 'oneself' though statements at different
    occassions that are not in agreement with one another.
    Anyway, was you 'accusation' that what I wrote in
    the one post didn't agree with what I wrote in some
    previous post correct or not?

    M. K. Shen

  • Next message: Gregory G Rose: "IVs versus Nonces"
    Loading