Replies to post #61474 on Advanced Micro Devices Inc (AMD)

[mmoy]

> But it will also depend on where the software is at the
> time. Even if Intel could release a 16 core processor,
> what kind of market would benefit? I think there is a
> software evolution that will limit the multi-core race.

I made a guess that Firefox and Thunderbird would be out around
the Fall to late 2005 but Makoto came out with a working Firefox
in the Spring and I have beta Firefox and Thunderbird right now.

But a 64-bit Java Plugin is not available yet and as of June,
Sun didn't even have it in their plans. Ditto on Linux though
on Linux, there is a 3rd-party plugin available. Don't even know
if it is available for Solaris.

HP is supposed to come out with a 64-bit driver pack within 1.5
months which will make me happy (there's one small bug in my
SoundMax driver).

I posted on this earlier this week and I think that the multicores are great for server machines but getting the
average desktop software vendor to port to multithread is
a tough sell as it increases a variety of software engineering
and maintenance costs. Those area that will truly benefit
from the change and that can afford the software expenses
will port over. I think that multimedia applications fall
into this category if there's money or help to do the
porting.

Intel could do something like grants to large software vendors
in terms of money, training or loaner engineers to help with
porting, maintenance and support. Once a major company has this
in their code base, their competitors have to follow suit to
be competitive.

Open Source is a great place to do this in.

[alan81]

AMD innovations ahead...
I you did a great job laying out some of what AMD has ahead. In addition to DDR-2 for notebook and desktop, they have FBDIMM for server/workstation. They need to get presidio and Pacifica (virtualization and security) out. They need to get 65nm shrink designs out.

In terms of overall performance enhancements I expect them to ramp clock rate to about 3.4Ghz by late next year resulting in a specint of about 2450... this is just a simple extrapolation based on the past couple of years. Perhaps with IBM's help and 65nm they will ramp faster/ higher than this. We will see.
--Alan

[jhalada]

wbmw,

At any rate, in spite of the FUD from some people, Intel looks like they have a compelling roadmap going forward, and AMD does not seem to have the "next best thing" waiting as a response.

Intel always has compelling roadmaps, which don't always lead where they're telling you they are leading. Right now, all the road maps go through a single narrow mountain pass called Merom. So, I would not yet count the chickens. The eggs look fine, very pregnent, but it is not like having chickens. The harrow mountain pass may still turn out unpassable, or the road conditions there may be bad, causing delays.

In the meantime, BTW, the CPU that you called desktop CPU pretending to be a notebook CPU is giving Dothan a run for its money. Wait a minute more than run for its money, Dothan is barely keeping up with the second best Turion CPU line - the ML line:
http://www.laptoplogic.com/resources/articles/42/1/1/

Their conclusion:

After 21 benchmarks, 2 comparisons, and a detailed architecture analysis, the conclusion is clear. Unless battery life while performing CPU-intensive tasks is paramount, choosing a Turion-based laptop over a Dothan-based laptop is recommended.

A bit surprising, if you asked me, but what do I know. I hate notebooks. But you are raving about notebooks practically non-stop, and if there is no substantial flaw in the review, it looks like your analysis could not be have been more off base.

Joe