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

Re: intel core 2 duo



Phil Meyer wrote:
> intel dead ended Pentium 4 due to heat and performance characteristics
> intel used the good heat and performance characteristics of the Pentium M 
> (which is an evolution of the Pentium 3) and moved it to 65nm process.
> In order to obfuscate the origins of the processor, the Pentium name was 
> dropped altogether and a new marketing series was developed for the newest 
> evolution of the Pentium 3 called Core.

Not quite. There have been a *lot* of tweaks involved in producing the
Core 2 core (and Intel's naming schemes are still awful...). It's
similar to the amount of changes between the Athlon XP *core* and the
Athlon 64 *core*.

> You might recall that the Pentium 3, as well as the Pentium M are 32bit.
> You may also recall the years spent developing Itanium, a true 64bit
> platform.

Itanium *is* a true 64 bit platform. It's also what's variously called
"VLIW" or "EPIC". It is *massively* different from any of the other
common CPUs out there. Intel was practically doing raw research while
they were trying to produce their new processor.

That's a lot different from widening datapaths and adding support for a
few more architectural registers (which is basically what's needed for
AMD64 support). There's nothing in the AMD64 instruction set that wasn't
well understood twenty-five years ago (with the probable exception of
stuff like SSE and MMX).

I maintain that making a processor support AMD64 is a *lot* easier than
making it fast by modern standards (which the Core 2 certainly is).

> How long in development from the EOF of Pentium 4 and the
> release of the 'Core' CPUs?

Irrelevant, really. Once Intel had decided to support AMD64 (under
whatever name), it was obvious that this would be wanted on laptops. So
Intel would have wanted to get support on laptop chips sooner or later.

It's not about whether the laptop can make use of more than 4 GB of
memory -- it's that there are some pretty big organisations that would
want to run the same binaries on laptops as on servers (which quite
probably would be 64 bit), or the same binaries as on desktops (which
are increasingly likely to be 64 bit). And those are the sort of
accounts where Intel would like to keep AMD out.

Why would laptops need to run server programs? Usually for
demonstration, troubleshooting, or programming purposes. It can be very
useful for field engineers to turn up on a site and be able to run their
own copy of the customer's programs on their own laptop.

> http://arstechnica.com/articles/paedia/cpu/core.ars

Good article, but it doesn't help your cause.

> Please notice the complete lack of any and all marketing data calling this a 
> 64bit architecture.

Marketing! Marketing, marketing, marketing! You've said it yourself!
Intel is a marketing-oriented company! They'll deny the sky is blue if
it suits their purpose!

And it *does* suit their purpose. Given that Intel has poured billions
of dollars into Itanium, given that one of the key marketing features
was that it was 64 bit, why would Intel want to detract from that? Intel
has a very clear separation in it's own corporate mind between the
Itanium, Xeon, Pentium (and now Core) and Celeron brands, and the
pricing reflects this. They do *not* want people who might buy a more
expensive brand to think that a cheaper brand would be just as good. So
they trumpeted 64 bits for Itanium, and kept quiet about it for EMT64.

(Intel had hoped that people would transition to Itanium as they ran out
of steam on 32 bit x86 -- making an AMD64-compatible processor scotched
that hope, but it died hard).

Incidentally, Intel are now talking about EMT64 at
http://www.intel.com/technology/64bitextensions/ and calling it 64 bit
computing -- their position has changed as 64 bit has become more
prevalent.

> The real question is:  Does the addition of EM64T make this a 64bit platform?
> My conclusion: NO.

Then what does it *%£&*£* take to make a platform 64 bits? It runs
Fedora 64 bit binaries.

But this isn't what you set out to prove in the first place. *Not* that
Core 2 couldn't run 64 bit binaries (it obviously can), but that Core 2
couldn't run 64 bit binaries *efficiently*.

To do this, we would need documentation showing that the speedup or
slowdown that Core 2 gets from running 64 bit binaries is significantly
worse than the speedup or slowdown that an Athlon 64 gets.

> And further, does every core CPU model feature EM64T, and how can you tell?
> If not, then what does EM64T on Core really mean?

Did every AMD processor based on the Athlon 64 core feature AMD64?

As a matter of fact, it didn't -- there were a number of Semprons based
on Athlon 64, which needed the original Athlon 64 Socket 754 interface,
which had AMD64 turned off -- it was initially reserved for the Athlon
64s and Opterons.

AMD was taking a leaf out of Intel's book -- various Intel CPUs from the
same family at the same time have had VT, Hyperthreading, EMT64,
different bus speeds, dual core, and probably other features I've
forgotten turned on or off as it suited Intel so they could aim
processors at different markets (and try to make sure that someone who
needed a particular feature was likely to pay more for it).

Hope this helps,

James.

-- 
E-mail:     james@ | A: Because people don't normally read bottom to top.
aprilcottage.co.uk | Q: Why is top-posting such a bad thing?
                   | A: Top-posting.
                   | Q: What is the most annoying thing in e-mail and usenet?


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