[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: instalation issue

On 2010-03-15 18:54, Stan Hoeppner wrote:
Ron Johnson put forth on 3/15/2010 5:07 PM:
On 2010-03-15 16:47, Stan Hoeppner wrote:
Ron Johnson put forth on 3/15/2010 4:11 PM:

Because of the way that AMD designed the specification, it's possible to
install a 64-bit kernel onto a 32-bit system.
You wanna take another stab at that statement Ron?

$ uname -m

$ file /bin/bash
/bin/bash: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV),
    dynamically linked (uses shared libs), for GNU/Linux 2.6.18,

$ apt-cache show linux-image-2.6.33-2-amd64 | grep ^Architecture
Architecture: i386

Methinks you were suffering momentary thought dyslexia. ;)

Nope.  The AMD engineers who designed that architecture had true

Yes, they did, to an extent, but it wasn't akin to leaping tall buildings in
a single bound.  If I'm reading your original statement correctly, I don't
think you really understand what that forethought was.


Just to make sure I'm reading your original statement correctly, and that
we're talking about the same thing here, when you state "32-bit system" in
your original statement above, are you referring to the hardware, or are you
referring to replacing a 32 bit kernel on an existing 32 bit Linux
distribution installation with a 64 bit kernel?

Replacing a 32 bit kernel on an existing 32 bit Linux distribution installation with a 64 bit kernel.

See my above examples from "3/15/2010 4:11 PM" which demonstrates that I'm running a 64 bit kernel on a 32 bit distro.

If the former, you're smoking crack because it's physically not possible.

It's foolish to attribute crackitude when you have half the needed information.

If the latter, you're giving credit to the wrong folks, and backing your
statement with a non-applicable reason.  The vast majority of the credit for
running 32 bit user space programs on top a 64 bit kernel goes to the Linux
kernel developer community, not the x86-64 engineers.

Without the AMD engineers building the needed features into the CPU, the kernel programmers wouldn't have been able to take advantage of them.

                                                      I think you're
attributing a bit of originality to these guys that doesn't apply.  A decade
before the x86-64 extensions were conceived and implemented, Alpha, MIPS,
SPARC, and PowerPC engineers did essentially the same thing.  In fact, many
Alpha engineers, including AMD's current CEO Dirk Meyer, went to work for
AMD after Compaq bought DEC and killed the Alpha off.

I've been working on DEC/Compaq/HP kit for 20+ years. Nothing you just wrote is new to me.

                                                       Ideas that went into
the 64 bit Alpha implementation were incorporated into x64-64.  The concept
wasn't new.  The "forethought" occurred a decade earlier when Meyer et al
designed the 21064, which, coincidentally, had to run the 32 bit VAX
instruction set in emulation.

Under OpenVMS it was called VEST, and under NT/Alpha it was FX!32.

                              All of that experience played a big role in
designing x86-64.

You're confusing "forethought" with "ingenuity".

While 32-on-64 wasn't original to AMD (after all, OpenVMS still isn't *fully* 64-bit even after 16 years on the Alpha!), it was still forethought on the part of the AMD engineers to see that PC users would need/want a smooth path from 32 bits to 64 bits.

Or maybe Intel was just gobbsmackingly stupid for thinking that everyone would just dump all their perfectly functioning h/w and s/w to buy Itanium kit and AMD management was rational.

Yeah, that's probably it.

Besides, the Intel guys didn't think it was possible to extend ia32 to 64 bits, but the AMD geeks had the ingenuity to figure out how.

Ron Johnson, Jr.
Jefferson LA  USA

"If God had wanted man to play soccer, he wouldn't have given
us arms."  Mike Ditka

Reply to: