[Wolves] Kernel 2.6

[Jon Masters]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Old Dan wrote:

jcm>>Which distribution image please.

| 2.6.3-1 from Debian Sarge.

Ouch er well ok. I don't like Debian kernels any further than I can
throw them, and that aint very far cos you can't throw them (unless you
were to print them out on a lot of pieces of paper and then hurl that
out of the window).

jcm>>The new Linux scheduler (which is not a clone of the Slowaris
jcm>>one

When I first read about the new scheduler work, I happened to have
recently read bits of ``Solaris Internals'' (a great book which Sun seem
to have convieniently let die, as there is nothing since version 7.
Unfortunately the only other good SunOS book is called ``Panic'').

jcm>>makes some reasonable assumptions for people on desktops
jcm>>by simply delaying io operations sometimes.

| Yeah I heard about that.

A good book is ``Linux Kernel Development'' by Robert Love. I pestered
him by sending a bunch of corrections (email me for them).

jcm>>the schedule main loop is O1 in its choice
jcm>>of which processto run next time

| ...but must confess to not having the foggiest idea what that statement
| means.  I will google for a Clue(tm).  :)

It is a thing Computer Scientists rant on about and many others who have
no idea what it really means. Big Oh notation simply refers to the
complexity of an algorithm. In the case of the Linux scheduler, it can
be constant time in some executions although I could be really anal and
pedanty by pointing out that some of the instructions that it uses can
have variable execution times on some architectures...yay pendantry!

Perhaps you could send us the output from these commands:

	*). dmesg
	*). lspci -v -v
	*). cat /proc/cpuinfo
	*). cat /proc/interrupts

jcm>>improved sense of response from the desktop I use at home
jcm>>but I do not overall rate the system as being "faster".

| Lots of other people seem to, as do I.  I suppose YMMV always seems to
| apply in these cases.

I find that a certain first person 3D shooting game is much more
responsive so that an old ATI card is even working usefully.

However in general my desktop at home is never pushed enough to really
make any particular difference.

|>Define big processes please.

| Er - I suppose 'big processes' is rather unscientific.  I just meant big
| apps really.  As I say though, this may simply be my imagination.

Ah ok. The key point is that a big app is relative, especially if it has
lots and lots of shared libraries because these are likely to be used
for something else anyway. IO performance will mean that a large binary
perhaps loads a tiny weeny bit faster but meh.

Jon.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFAi8VheTyyexZHHxERArd1AJ9RdUKfE1wGvs7XknH/35Go3eY6ggCfVEpi
YMNIIk9YfCCY4NWfNiveivU=
=RIgX
-----END PGP SIGNATURE-----