[gentoo-amd64] Re: tmpfs help

[Duncan <1i5t5.duncan@cox.net>]

Volker Armin Hemmann <volker.armin.hemmann@tu-clausthal.de> posted
200802131717.57766.volker.armin.hemmann@tu-clausthal.de, excerpted below,
on  Wed, 13 Feb 2008 17:17:57 +0100:

> Not real benchmarks. But kdepim with enablefinal, 1gb of ram and -j2
> several hours. With j1 2h.
> .
> kdepim with 2gb of ram and j2 30m
> Just because first case swap storn, last case no swap at all.

But that's measuring the effect of the additional memory as much as it is 
the effect of swap.  What you need to do is kdepim with enablefinal, a 
gig of RAM and -j1 with PORTAGE_TMPDIR set to someplace on disk, against 
the same thing but with PORTAGE_TMPDIR set to a tmpfs of sufficient 
size.  Then and only then are you measuring the effect of tmpfs vs no 
tmpfs.  He (and I) are arguing that the tmpfs case will take less time 
because some files will get deleted before they end up being written to 
disk at all, as opposed to /all/ of them being written to disk (well, if 
they remain more than a few seconds anyway).

If you like you can of course do the test with -j2, but then you must use 
-j2 for both cases, so again PORTAGE_TMPDIR on tmpfs vs on disk is the 
only variable.

This of course assumes similar system load otherwise, presumably either 
an idle system (in X/GNOME/KDE/whatever both times or at the CLI both 
times) or only something steady, say streaming Internet radio, the same 
station at the same bits per second without visualizations, going on, 
with the same apps loaded in the background.  No answering mails or 
browsing or anything else interactive or variable since it's possible you 
could then be doing something different at a critical time in one case 
vs. the other.

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman

-- 
gentoo-amd64@lists.gentoo.org mailing list