On Sun, Apr 28, 2019 at 06:15:51PM -0500, Scott Harvey wrote:
> On Sun, Apr 28, 2019 at 5:06 PM Ken Moffat <[email protected]> wrote:
> 
> > Oh, and if the machine's CPU is not the same as, or a superset of,
> > the machine on which you built LFS, you may need to rebuild gmp if
> > you didn't use the configfsf scripts.  And although I'm a fan of
> > using '-march=native' in my CFLAGS, that is fairly catastrophic for
> > trying to build for a different micro-architecture :)
> >
> 
> This might be part of my problem. My initial LFS build was constructed on a
> Core i7. The machine I want to continue LFS'ing on is only a Core i5. I
> don't know enough about processor architecture or the intricacies of the
> Linux kernel to know how much that may matter.
> 
> Regardless, this (older, i5) machine has a fresh Manjaro install and I'm
> starting LFS over, using the development branch this time around. I know
> the risks and hazards and further complications that may entail, but
> breaking and fixing things is my definition of "fun". The remaining
> important question is: when I get this version booting and running, do I
> get to be Counted a second time? :-)

I've long-since lost whatever number I had on the counter, so no
advice on that.  But for processor architectures - unless you pick
up experimental patches originally at gentoo, the kernel should not
care (as I understand it, it looks at runtime to check the details
for things like caches. etc).

But for some userspace packages (gmp in LFS, if using its default
configure files, more generally some multimedia packages in BLFS)
the _micro_ architecture matters : on intel, primarily the first
digit of modern CPU identities.  I have a SandyBridge i3 (2xxx),
Haswell i7 (4xxx), Skylake i3 (6xxx) - the higher the number, the
more new op-codes [ although sometimes low-end, e.g. modern pentium,
do not have the full package - that has caused problems with gmp in
the past ].

And a warning: you can have a lot of this sort of "fun" with LFS -
I'm currently trying to persuade most things that I build in BLFS to
use my own CFLAGS, initially looking for '-march=native' in the
build logs to see if they are using my CFLAGS - later, I'll try some
more exotic things to see if they help (compilation these days is
often so slow!), but I've already seen indications that using
-march=native in *LFS* might stop me building for other
micro-architectures.

Oh well, if that happens I can currently go back to a system before
I used -march=native (probably LFS-8.1) to build a generic system
for initial binary installs.

I already knew that AMD Ryzens have dropped a few opcodes and cannot
run systems built for a Kaveri, but for that usage I built on an old
AMD.  Sometimes, you have to keep your wits about you.

Have fun!

ĸen
> -- 
> http://lists.linuxfromscratch.org/listinfo/lfs-support
> FAQ: http://www.linuxfromscratch.org/blfs/faq.html
> Unsubscribe: See the above information page
> 
> Do not top post on this list.
> 
> A: Because it messes up the order in which people normally read text.
> Q: Why is top-posting such a bad thing?
> A: Top-posting.
> Q: What is the most annoying thing in e-mail?
> 
> http://en.wikipedia.org/wiki/Posting_style


-- 
With a few red lights, a few old bits, we made the place to sweat.
No matter what we get out of this, I know, I know we'll never forget.
Smoke on the water, a fire in the sky.  Smoke, on the water.
-- 
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Do not top post on this list.

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

http://en.wikipedia.org/wiki/Posting_style

Reply via email to