> > > Perhaps someone could at least run a benchmark to measure this
> > > effect on two otherwise identical systems with two different OS
> > > releases installed, say 1.4.3 vs. 1.5?
> >
> > I have just set up such a machine:  MVII, 13mb ram, 4x 1 gig identical
> >                                     scsi drives, 1.4.3 generic and
> >                                     1.5 generic loaded on individual
> >                                     drives with room for 2 more setups
> >                                     using stripped kernels.
> >
> > IF someone will send me what is considered to be the most highly stripped
> > and bare and fastest MVII kernel config, I will run that up on 1.4.3 and
> > 1.5 on the other two disks.  You gurus decide what is needed.  The
> > config must contain tk50, deqna, radisk and ka630 support, and really
> > nothing else.
> 
> I don't think the size of the kernel matters all that much speed-
> wise, unless you are severely memory restricted.

OK, but I wanted to compare stripped identical kernels to see what
bloat difference there was between the two.  Using stripped kernels
and configs that were about as identical as I could make them, the
sizes of kernels has grown from 380K to well over twice that, on
MVII stripped kernels from 1.0A through 1.5, that I have run off
over the past couple of weeks on my MVII box.  I will rerun that
on 1.4.3 vs 1.5 over the weekend to confirm the numbers and some
simple tests.
 
> What's otherwise the important thing is to run a kernel config
> that's comparable (as few differences as possible) on the two
> releases.

Granted, but I wanted the gurus to send me their chosen config.
If not, I will choose my chosen stripped config and run with that.

``Who can name/write that config in 20 lines or less?''.....(:+}}.

> I think I would recommend you run lmbench on the two systems, and
> merge the results (give them different names so that the results can
> be summarized in a single table for easy comparison).

OK.  Where is that, in the system or in the packages?  I looked in
the packages tree and only saw an lmbench-2alpha11.tar.gz or
something like that.  Is that it?
 
> And of course any other relevant benchmark you can think of which
> excercises the kernel/user interface in various ways.

Well, the simple things I was doing was logging boot times,
compile times, tcpip times, and that sort of thing.  I need
to get better system call overhead numbers and that sort of
thing to help find out where the bottlenecks are.  I will
work on that.

> As for the interpretation and explanation of the results I'm not
> qualified to comment, but it does at least help to have some
> measurements to rely on.  As an aquaintance of me say: "pair in
> facts beats a full house of suppositions".

OK, but I don't know enough deep in the systems to offer much
other than run the tests and report some numbers.  Others will
have to interpret them, for me, too.
 
> However, as long as suppositions and rumours is all we're
> discussing, yes, I seem to remember others mentioning that something
> also changed on the path to 1.5 in the network stack which caused
> worsening in the performance, especially noticeable on slow
> machines.  However, that's probably a different issue than the one
> we're discussing above.

Well, I am not trying to push rumors, but I am running into these
bottlenecks on my MVII.  Anything I can do to speed it along is good.
Surely I am not the only bozo running into this.....(:+{{...

Query... I just downloded todays sources, and was wondering if I
should compile those for the kernels or use the stock 1.5 suite?
Anyone have any feeling on that, so as to provide the best info?
 
Thanks

Bob