On 21/02/2022 11:22, Robin Vowels wrote:
Has anyone been following this?
https://greenlab.di.uminho.pt/wp-content/uploads/2017/10/sleFinal.pdf
Has anyone seen any source?
It would be interesting to see it.
In previous language comparisons (notably, Tucker),
either the code has been much less than optimum,
or the wrong compiler options were used.
If the focus is on energy efficiency, I would want to see some
comparisons across various hardware platforms. Related to that. it would presumably be instructive to see at least a histogram of hardware
instruction usage (since ultimately that is where the energy use is) and
a discussion of how that usage relates to the source language and
performance. I am having difficulty understanding the point of the
present paper, but perhaps I am missing something, lengthy IBM mainframe experience may predispose me to a particular perspective.
On a broader issue, what seems an interesting question is whether
program performance can be analysed adequately by solely external means
(as contrasted with simply being time measured) or whether it requires
also some sort of internal instrumentation (as provided for instance by
some PL/I options).
Peter
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)