Collecting SBUs now

Ronald Hummelink maillist at hummelink.xs4all.nl
Wed May 29 06:43:11 PDT 2002


On Wed, 2002-05-29 at 13:39, Rui Ferreira wrote:
> From: "Ronald Hummelink" <maillist at hummelink.xs4all.nl>
> 
> > Host system is LFS 20011130 CVS and compiling running in X and
> typing
> > this mail in some Evolution window and copy pasting instructions
> from
> > mozilla and a few Xterms. I leave the system untouched while
> compiling.
> > (imho: a real life situation)
> 
> Yes, real life.......
> 
> > Compiling on reiserfs host, but installing to an ext2 loop device
> > My 256MB swap is also in loop (my second hardisk which holds the
> swap
> > and a partition for this stuff is gone for RMA)
> 
> ...real life also, but of no good for stats since HD access is the
> main time-eater.

Idealize the builds by building from a VGA console with not even de
kernel loggers running is as far from ideal as almost nobody will be
building from that.
I know that building in loop devices is far from ideal and that is why i
stated it. It is up to Gerard to decide what to do with it.  Compare my
times with the times someone may get from building on a laptop (which
have VERY slow harddisk) and i might not even be very far off.
Different filesystems might also have a major influence, take a look
here: http://hummelink.xs4all.nl/download/FS-bench.html where i did some
informal testing with bonnie++ on Ext2, Ext3, XFS, JFS and Reiser.

Should the book use just 1 set of times or the average of many users?
Should the book use 1 carefully crafted build situation or the average
of what users may happen to have?

I think the last for both questions, but will let Gerard decide, it is
his book after all. He is free to use my results, he is free to send
them to /dev/null. 
I will submit results using my replacing hardisk if it is still useful
to do so when i get it.

My 2 cents.




-- 
Unsubscribe: send email to listar at linuxfromscratch.org
and put 'unsubscribe lfs-dev' in the subject header of the message



More information about the lfs-dev mailing list