Hi out there!
I was really impressed when I first started to copy 1000 mails from my
linux-kernel mail folder to the archive/linux-kernel folder in netscape.
While under 2.2.16 it you could count the along with the first dozen
messages being transferred, 2.2.17 lets netscape do a jump start and
reduces the time needed to move 1000 messages out of a folder with 2000
to - not measured - one fourth.
The netscape mail folders reside inside an loopback-blowfish-encrypted
filesystem (astors patch) of just under 2G in size that is itself
located on a 3x1.5 raid5-array (mingos patch).
Impressed that I was I ran my 'swapping test': select some jpgs in the
kde file manager and let them be shown by kview. KFM is dumb so it
starts an instance of kview for every file that I selected, which in
results in excessive memory pressure.
I then noticed pauses of up to one minute when the system seemed dead. I
then cleared the buffer cache using a huge jpg and patch-2.2.17 and
patch-2.4.0-test2 each displayed in a separate netscape window and
restarted the test with vmstat 1 in the background. The first two vmstat
traces are with vmstst at nice 0, the third is with vmstat at nice -19.
The overall feeling I have of 2.2.17 vs 2.2.16 VM after half a day of
laying with it is that 2.2.16 remains responsive [1] under high loads
(with which I do not mean my swap test) whereas 2.2.17 becomes jumpy,
like working over a slow network. But then 2.2.17 increases the
performance of some operations dramatically. I don't know which I find
better. I think that 2.2.17 is far better than 2.2.16 when the load is
normal, but I'm afarid that maybe a background kernel compile will lead
to a jumpy mouse - will try that later.
Attached please find the vmstat outputs of the three tests I carried
out, where the first one seems to match the behaviour of 2.2.16 quite
closely.
My System:
K2-2 500, 128M, SuSE 6.0.
What has changed:
kernel: 2.2.16
applied patches: patch-int-2.2.16.2, raid-2.2.16-A0
kernel: 2.2.17
applied patches: patch-int-2.2.16.9, raid-2.2.17-A0 (rejects by hand)
Marc
--
Marc Mutz <[EMAIL PROTECTED]> http://marc.mutz.com/Encryption-HOWTO/
University of Bielefeld, Dep. of Mathematics / Dep. of Physics
PGP-keyID's: 0xd46ce9ab (RSA), 0x7ae55b9e (DSS/DH)
testing-kview-44-1.7.gz
testing-kview-68-3.3.gz
testing-kview-68-3.3-v2.gz
