[CentOS] Re: Is a new HP-dl380 dual XEON 64 bit or 32 bit -- AMD64 v. EM64T

[Bryan J. Smith]

On Wed, 2005-06-15 at 19:59 -0400, Robin Mordasiewicz wrote:
> I will download the ia64 bit version then.
On Wed, 2005-06-15 at 18:04 -0700, Hilliard, Jay wrote:
> You want the x86_64 version, not ia64.
On Wed, 2005-06-15 at 21:06 -0400, Ignacio Vazquez-Abrams wrote:
> No, you'll download the x86_64 version.

As everyone has mentioned, Intel Xeons are IA-32e aka "x86-64
compatible," so they run binaries built for the "x86_64" target.
>From a programmer perspective, both x86-64 (AMD64) and IA-32e (EM64T)
use PAE52 (52-bit virtual addresses into 48-bit "Long Mode" addressing)
with a 64-bit ALU and registers (there are some 128-bit XMM registers,
but that's another story).

> EM64T is AMD64 compatible

<Anal>EM64T is compatible with a _subset_ of AMD64T</Anal>

EM64T still currently runs on Intel 32-bit AGTL+ interconnect.  There are
some serious limitations to the physical interconnect _outside_ the CPU
-- both the "memory hub" approach as well as legacy AGTL+ addressing
issues, especially for I/O.  The "dumb hub" and 32-bit addressing
limitations are why EM64T processors lack some serious features of AMD64,
like the I/O Memory Management Unit (MMU).

AMD64 is based on Digital 40-bit EV6 interconnect.  It is capable of safe
memory addressing up to 40-bit/1.1TB (1TiB) for _both_ programs _and_
memory mapped I/O.  Intel proponents downplay this feature, but it's very
much a major different in the Linux/x86-64 kernel.

> Xeon's aren't compatible with ia64

IA-64 is a completely different architecture, byte code, etc...

<flammatory>IA-64 deploys CS ideals such as EPIC and Predication, things
that are designed to address limitations with optimizing machine code at
the compiler-only.  The reality is that machine code (like boolean
logic, long story) are legacy 1970s concepts for integrated circuits
designed by CS majors, before physicists and engineers took over in the
1980s.  You can't optimize math and algorithmic approaches that aren't
ideal at the semiconductor-level anyway, and run-time optimizations in
the processor design itself are the best way along with an optimizing
compiler that leverages those tricks (especially in the x86 future of
virtual cores of virtual, out-of-order PAE36/PAE52
machines).</flammatory>

On Wed, 2005-06-15 at 21:17 -0400, Robin Mordasiewicz wrote:
> oh, ok. I just torrented the ia64.

As others said, ain't gonna run.  ;->

> I am still blown away at how short a time it takes to download a dvd iso.
> this is good becuase the rest of my white box servers are amd_64 bit.

AMD64 has the advantage of being the only commodity PC platform with
a partial mesh of interconnects as well as being truly beyond the 32-bit
issues at the board-level as well as the register.

Ironically, it looks like Linux isn't the best server OS for AMD64 right
now (as much as we don't like to admit it).  Linux is rather immature
when it comes to platforms with multiple interconnects, whereas traditional
RISC/UNIX platforms have been perfecting their kernels for a good decade.


-- 
Bryan J. Smith                                     b.j.smith at ieee.org 
--------------------------------------------------------------------- 
It is mathematically impossible for someone who makes more than you
to be anything but richer than you.  Any tax rate that penalizes them
will also penalize you similarly (to those below you, and then below
them).  Linear algebra, let alone differential calculus or even ele-
mentary concepts of limits, is mutually exclusive with US journalism.
So forget even attempting to explain how tax cuts work.  ;->