[SM-Discuss] License Choices
warl0k at lvcm.com
Mon Jul 22 14:23:26 EDT 2002
I think I may have not been as clear as possible last time.
Allow me to explain my position a little fuller (sp?).
I just think it should be *optional* to use it. I do not advocate
removing it completely; I just think that for those who have the
need/want to do it, they should be able to do so.
One more if/then statement couldn't possibly hurt that bad.
As far as file system fragmentation goes good luck on convincing me on
that one. I have a 486 machine with 8M of RAM that I had running LFS for
over a year with 0.8% fragmentation of the file system. And it had
XFree86, apache, and all the other good stuff compiled on it. And I
wasn't even using reiserfs at the time. I kept it up to date (3
recompilations of XFree86 as newer versions came out) and everything was
compiled on disk, not put into a pseudo-ramdisk.
There are those of you out there that the tmpfs actually works for. In
my case, it doesn't. My system is constantly choking for physical RAM
(it hovers at 5-6M free 24/7) and as a result when I open a larger
process it has to shift things around. Yes, even with 1.5G of RAM. This
has been my biggest beef with tmpfs is that it seems to use every
available iota of RAM available and lets the system use swap for
"normal" (read: non-tmpfs mounted) processes. As a result things get
slowed down unnecessarily as my system goes to figure out what can be
swapped where, etc. eating precious clock cycles doing so. When a
non-source based distro (RedHat, Debian) outruns a source based distro
like sourcemage, something is wrong. Unmounting all the tmpfs mounts
that get left behind by the ever-accursed freeamp compilation that has
never worked makes my system scream. Leaving it there slows it down
immensely. The difference between mozilla (not galleon, full-on mozilla)
opening faster than I can perceivably count (blink of an eye literally)
, and taking upwards of 12 seconds to load up. Hence why I am on the
warpath to get an option to disable this broken (in my case) feature
set. Perhaps I should fix the freeamp spell to make the catalyst go
away, but for me the more efficient thing to me would be to disable
tmpfs as it does not seem to work in my favor.
Just my .020078143865 (adjusted for inflation)
From: sm-discuss-admin at lists.ibiblio.org
[mailto:sm-discuss-admin at lists.ibiblio.org] On Behalf Of Sergey A.
Sent: Monday, July 22, 2002 10:58 AM
To: Glenn Shannon
Cc: 'Phil/CERisE/KG6MBQ'; sm-discuss at lists.ibiblio.org
Subject: RE: [SM-Discuss] License Choices
Actually, 256 MB is more than enough, even for XFree.
I'm the witness.
Glenn Shannon <warl0k at lvcm.com>:
> I believe the choice to use tmpfs was because
compiling in memory is
> Unfortunately for those of us who have fast enough
machines, the benefit
> of tmpfs is minimal. For those that don't actually
have 1G of RAM to
> toss around like a ragdoll on May Day, it slows
things down as it has to
> swap stuff out to disk which is less efficient than
using the disk as a
> medium for compilation.
> For the 4 or 5 people in the world who are on the
right machines for
> tmpfs compiling (medium range machines with 768+M of
RAM) tmpfs gives
> them a .05% increase in speed in return for its hefty
SM-Discuss mailing list
SM-Discuss at lists.ibiblio.org
More information about the SM-Discuss