yade-dev team mailing list archive

[Bruno Chareyre <bruno.chareyre@xxxxxxxxxxxxxxx>]

On 12/03/15 16:16, Jerome Duriez wrote:
> The diff is attached. The line numbers do not correspond to current
> gitHub version, changes in fact would apply in this version to l. 178
Thanks. It looks safe.
>
> But, it seems speed measurements significantly fluctuate. This is the
> only reason I see to explain the slight speedup obtained observed in
> j1 between Old and NewCode : p.1 of the pdf.
Fluctuations are expected. That is why measurements are done on the
average of many runs usually.
If both old and new were compiled by yourself with the same options it
was probably a random fluctuation (or an obscure reason leading to
differences in compiler optimizations), a bug introduced by you (and
leading to less bridges) is very unlikely since the results with -j1 are
strictly the same.

> I do not have a high experience of speed execution variability, but
> this variability I observe (between comparable simulations) makes me
> wonder if "1.15 - 1.2" speedup is beyond natural variability.
It is expected. If the 1.15 is obtained by comparing only one run for
each case it doesn't say much. Even comparing the old with the old could
give such differences depending on what is measured.

Bruno