pre-9: somethings not right here...

Samuli Kaski (samkaski@cs.Helsinki.FI)
Sat, 20 Sep 1997 18:54:20 +0300 (EET DST)


Hello!

I will try not to make a bible out of this message. Just hang on.

After I booted my Linux from pre-6 (15 days of slavery) some 8 days
ago I have noticed that pre-9 somehow (don't ask me why) caches a lot
more memory. I can't figure this out so I tell you.

I have seen a lot of "pre9 leeks memory" -messages so I just wanted to
be polite :)

Usually my cache shrinks after running apps that "don't use the same
ram" and require a lot of "new fresh memory" (ex. starting qwcl after
X), but on pre-9 I have constantly 9-15 MB in cache (on a 32/64) even
if I have recently ran a process that needs a lot of memory. Secondly,
running any heavily kernel-executing application somehow takes longer
time then it did on pre-6 + patches. Just running "top" while playing
mp3's makes the sound break. Before I could run just about anything
without the sound braking.

If this looks right to you, twist my arm (don't laugh!! :))

samkaski@pasuuna:~$ free
total used free shared buffers cached
Mem: 30860 30424 436 13312 5072 10228
-/+ buffers: 15124 15736
Swap: 66460 8092 58368
samkaski@pasuuna:~$

Now, I have a 7 day uptime, so if you want me to try a patch/patches
that takes/take some time to show up on statistical figures, I guess
it will take some time on a 32MB-P5, but...

In case the active kernel developpers want to do some testing, let me
know. I'm 24/7 connected, so you can make your own tests if you want.

If this problem is just on my platform, I'm sorry for the time you
took to read this message, I however think that it is forth of telling
the kernel developpers every strange experience one might have.

If this is ignored on the list, I will just go back to pre-6 (+
Werner's patch that included patches from Krzysztof and Benjamin) that
worked 100% for me. Thank you for taking the time to read this.

The usual crap from the mortal Linux users:

I'm just using Linux as my OS as it is clearly the best
performing/stable one. I'm sorry for not contributing to the coding
part, mainly due to studying/working 2-3 days/week.

--
Samuli Kaski, samkaski@cs.helsinki.fi
Department of Computer Science, University of Helsinki, Finland.