An Open Platform Project for Embedded Multicore Systems

Hi Stefan,

Once again thank you very much for your help!
It seems that now, after applying your advice, I am close enough to the original distribution and the mean starts to resemble the proper value from the Amalthea benchmark.

From the conceptual point of view, it is interesting for me how to understand such high value of pRemainPromille as 0.5 in AMALTHEA_Democar.amxmi (whereas in another file that I have it's reasonable 0.005). Does it mean that the Weibull disctrubution (and, at least, its parameters) is, in fact, a poor choice for this particular case?

Best regards,
Piotr Dziurzanski

Hi Piotr,

glad to hear that the approach worked for you!

It should be said that the unit of pRemainPromille is permille, so a value of 0.5 denotes a probability of 0.0005 which is quite reasonable from my perspective.

The Weibull distribution is in general particulary suited for modelling measured runtimes for the use in timing simulations or probabilistic analyses. However there might be some special cases where a constant runtime or an uniform distribution is more suited.

Kind regards,
Stefan Schmidhuber

Hi Stefan,

I wasn't sure if this probability was given in permille (despite its name), thank you thus for clarify it. Maybe this also should be highlighted in future versions of the documentation.

There is a similar problem with frequency - I've assumed it is Hz, but compare it with time, where you have the proper unit given explicitly.

Anyway, thank you once again for such big help and clarification!

regards,
Piotr