[RC5] Help with justification of running Dnet at work

Marcin Sochacki wanted at gnu.univ.gda.pl
Sun Mar 19 14:00:03 EST 2006


While it is true that d.net can be responsible for some, non-zero
slowdown, the actual difference is unlikely to be noticeable. In your
lab you can e.g. select two identical computers, one with d.net, the
other without one and run PowePoint measuring the startup time and maybe
loading a large file (repeat a couple of times to average the results).
This might convince the guy, but from what you write it seems that he's
just one of those close-minded selfish types, who in fact know very
little, but they play an important person. I'm afraid nothing can be
done to convince them, unless you convince their manager to do it.

Also, most accounts of dnet visibly slowing down applications are not
backed up by any evidence. Killing dnetc is just the first, simple thing
to make you happy that you've done anything to speed up the app. In fact
the change is tiny, but it appears to be visible, just like placebo effect.

One of the few justified reasons to not run dnetc is the increased power
consumption. Current CPUs use quite a lot of energy and saving it is IMO
an important aspect. But your lecturer didn't use this argument -- one
more point that he doesn't have a clue.

Marcin


More information about the rc5 mailing list