[petsc-users] Obtaining bytes per second

[Matthew Knepley]

On Wed, May 6, 2015 at 6:43 AM, Jed Brown <jed at jedbrown.org> wrote:

> Justin Chang <jychang48 at gmail.com> writes:
>
> > Thank you guys for your responses. If I want to estimate the number of
> > bytes that come down, would -memory_info give me that information?
>
> No.  There is no substitute for a performance model.  Hardware
> performance counters can tell you something, but without a cache-aware
> performance model, it can give you information that is off by an order
> of magnitude or more.


I think the idea is to be explicit. I would just use something like

  # Vecs * Vec size * 8 bytes/double + <same for Mat>

and forget about small stuff. This model is uncached, and you just state
that up
front. Someone else can take the data and produce a more nuanced model.

You can use the Flop counts (not Flop/s) from PETSc,  and get a rough
estimate
of flop/byte ratio.

> And with this information plus the total number of logged flops, i can get
> > the ratio of flop to bytes and hence the (crude estimation of) upper
> bound
> > FLOPS/s based on the reported stream BW?
>
> Flop/s is logged (for PETSc operations) and you can use hardware
> performance counters for that (much better than for bandwidth, but still
> erroneous due to speculative execution and incomplete/gratuitous
> vectorization).  But to compute an arithmetic intensity or find
> performance bounds based on memory bandwidth considerations, there is no
> substitute for understanding what you are doing.  Sorry.
>

Do not let Jed's pessimism get you down.

   Matt

-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20150506/9f2003e3/attachment-0001.html>