Some notes/problems

Guillermo Pereyra Irujo gpirujo at
Mon Jul 2 05:14:15 PDT 2001

> > Define today's average PC. I myself have a Celeron 533 with 320 MB ram.
> > The LFS server is a PIII-750 with 512 MB RAM. I use both machines often
> > to test LFS. Which one would be more average? The LFS server would
> > probably be more average when it comes to CPU speed, but my workstation
> > would probably be closer to the average RAM amount these days. But the
> > LFS server has SCSI disks and my workstation IDE. Hard to decide which
> > one to use then and we can't start including 10 different estimates.
> > It'll be like "Glibc can take 20 minutes, or it can take 60 minutes,
> > depending on your cpu speed, ram, scsi or ide, one or more cpu's, etc,
> > etc, etc, etc". It just doesn't tell you a lot. And it's not feasible to
> > add specific configurations. How many should we add. A list of 5
> > different systems that are actively used for LFS testing?
> Compiling software takes primarily CPU power. Their aren't big differences
> between 32 and 64 MB and IDE and SCSI disks.
> We should give specifications for the most common processor speeds (66, 166,
> 500 & 1000 MHz), but I believe we should wait with this until the
> fill-in-a-form-and-the-server-generates-a-book-special-for-you idea has been
> realized.

We can take a small package (let's take ed), time its compilation and time
all the other packages' compilation in 'eds'. I don't know if these time
quotients will be consistent among computers, but perhaps any of you
since-long-ago compilers knows that.

This could be an optional script. Before you start, you run it and it
tells you how long an ed is in, say, minutes. Before each package
compilation instructions, there's a 'this pack takes 11.5 eds' note. At
the beginning of each chapter, there's a 'this chapter takes 140 eds'
note. If you want to know how long the whole building will take (or the
glibc, for example) you just have to know how long an ed is and calculate.

I can make it if you think it could work.

Guillermo Pereyra Irujo
Mar del Plata, Argentina

Unsubscribe: send email to lfs-discuss-request at
and put unsubscribe in the subject header of the message

More information about the lfs-dev mailing list