Intel 64 or AMD 64 POLL & comments...

[Acanthus]

If they had a chipset that didnt suck, id buy A-64.

 

[Viditor]

IBM is co-developing the x86-64 processors with AMD

Actually, I don't believe they are...they are co-developing the K9 (AMD's next-gen processors)

As for Sun being a "heavyweight in the industry". I assume that you haven't read their financial reports or seen the predictions that many are making that Sun will be out of business in a couple of years

Actually I have, but they are old news...If you read the more recent reports, Sun's future has radically turned around since October...
From Q4 04 report:
"Sun saw sales jump to 83,833 units in the quarter versus just 62,914 last year. The company holds 5.3 percent of the market compared to 4.9 percent last year"
Also, keep in mind that this the whole server market...Sun were very, very late in adopting Linux. For Unix servers they have close to 50% of the market still.
Because of this, Sun's share price has gone from $3.07 in October to $5.60 today...quite a turnaround!

Now as for the heat output of P4 and A64 processors. Let's take a look at this through facts provided by each company

OK...
1. Intel released only the average TDP, not the max...AMD release both the max and the min...
89W is MAX TDP on a 3200+ A64, 103W is AVERAGE TDP on a P4 3.2E
2. Anecdotal data and average test reviews tell the story in much greater detail...Average temp under load for a 3.2E is 70C+ as compared to 45C-50C on Northwood and A64...

 

[Viditor]

113 posts in 4 years, most of them in this thread..

It's a slow work week...

 

[Viditor]

Both values are TDP (Thermal Design Power; which is an industry standard way of representing thermal output) which if I'm not mistaken is the absolute worst case for the thermal output for that particular chip design

Not really...If you look at AMD's chart (that OddTSi kindly provided), you will see both the min TDP and the max TDP...

 

[redpriest_]

AMD is not developing any processor with IBM's assistance - the alliance is only for *process improvements*, specifically at the 65nm node.

Intel's TDP definition: Maximum expected output using common applications.
AMD's TDP definition: Maximum output given the laws of physics - input current and voltage.

Don't you find it curious how AMD's entire Opteron line has the same TDP? 89W? From 1.4 GHz - 2.2 GHz, that's kind of curious eh? On an Intel platform, your heat output could exceed TDP if you were running a hard working, very well optimized core. This isn't such a huge problem because the P4 can always throttle performance if it exceeds acceptable norms. AMD designs for worst possible case because Opteron only has a thermal switch rather than throttling.

 

[cm123]

So now that Intel has indeed said yes to 64 bit, does this change everyone's thought on AMD 64?

I guess that shoots the thought Intel will need years...

To all that voted who cares, does it matter to you now?

(below of course from AT):

CNET has yet ANOTHER article on Intel's 64-bit server chip. The chip is also compatible with AMD's x86-64 standard:

Intel will come out with a server chip next quarter that adds 64-bit processing power to its current x86 line of processors, the company's chief executive said Tuesday.


In a keynote speech at the Intel Developer Forum here, CEO Craig Barrett called the arrival of Nocona "one of the worst kept secrets in San Francisco." Intel had been widely expected to show off such a chip.

Nonetheless, having the chip next quarter is earlier than many analysts had predicted. Nocona will use the same core as Prescott, Intel's current desktop processor, meaning that Intel could quickly migrate the 64-bit architecture to desktop processors.


During his keynote, Barrett demonstrated a 64-bit x86 chip running on a desktop machine, a Dell Dimension XPS...

Intel's approach is compatible with AMD's, the representative said. "There will be one operating system that will support all (64-bit) extended systems," the representative said.

 

[Kell]

Originally posted by: OddTSi

Originally posted by: Cerb
The A64 specs are maximum possible output EVER (hence the 3000+ and 3400+ are the same). I am currently unable to find anything to back this claim, but I'm pretty sure on it...off to bed I go.
The Pentium 4 specs, as usual, appear to be maximum EXPECTED output. Big difference.

Both values are TDP (Thermal Design Power; which is an industry standard way of representing thermal output) which if I'm not mistaken is the absolute worst case for the thermal output for that particular chip design. It doesn't mean anything different for the P4 than it does for the A64.

"You're off the edge of the map, mate. Here there be monsters!"*

FYI, it's called "thermal DESIGN power" because it's the heat level that each company recommends system builders plan for. I even remember reading a TDP footnote in some of Intel's P4 datasheets years ago that clarified this somewhat--it stated that TDP was not the true max power draw, and directed you to another datasheet.

Intel can get away with "underspeccing" the cooling for two reasons:

1) P4 clock throttles itself rather than locking up. If clock throttling fails to keep temps down (which just about never happens), it trips off.

2) The P4's power draw usually doesn't exceed TDP, and when it does, it does only for very short bursts, easily handled by clock throttling. True max DC power is easily calculated as volts*amps, and even in some Northwood parts, that figure exceeds 100W (FWIW I think it actually got a hair over that in Willamette too.) It just rarely gets even close to that unless there's a processor malfunction or a specially crafted "power virus" (i.e. the Toast utility for the K7).

AMD has to spec its TDP closer in line with true max power, simply because AMD parts handle heat differently. K7 cores tend to fry or do latent damage to themselves when they overheat; K8 cores trip off to avoid damage. Neither solution is quite as ideal as clock throttling.

*[Disclaimer for those who don't recognize: this quote shamelessly ripped from Pirates of the Carribean. Credit where credit is due, y0!]

 

[Kell]

Oh, and by the way...

Originally posted by: OddTSi
IBM is co-developing the x86-64 processors with AMD, therefore their own profits (and reputation of being a good outsourcing partner for manufacturing processors) are at stake in the sales of Opteron-based servers. So them selling these servers is not the ringing endorsement that you make it seem. Of course a company is going to sell/hype a product that they helped make.

No, not really (not to anyone's knowledge, anyways, and not in any rumors I've heard). IBM and AMD are reportedly cooperating on "process technology," nothing more. K9? Maybe, but I haven't heard anything of that. It's years off anyways, and IBM knows better than anyone that a lot can change in a few years.

IBM is apparently just providing Opteron because customers demand Opteron HPC/server systems, and IBM is to some extent processor-agnostic (FYI IBM sells IA64 systems too).

As for Sun being a "heavyweight in the industry". I assume that you haven't read their financial reports or seen the predictions that many are making that Sun will be out of business in a couple of years. However, if you meant heavyweight in the sense that it's sinking like a stone then I'll agree with you.

If you take a moment to look beyond Sun's projected financial situation, and see what made SPARC great, you might notice why the endorsement of Sun counts for so much.

SPARC kit was (is) some of the most reliable stuff on the planet, and while it fell below the peak benchmark scores of some of its competitors, it pretty much outpaced the entire pack when it came to sustained performance serving multiple clients. That's one of the areas where x86 just hasn't been a great performer; I've seen piddling little single-processor SPARC boxen tear a new hole even for multiprocessor Xeons when it came to serving a metric assload of requests.

x86 is pulling ahead of SPARC primarily because it's cheap, and because SPARC technology, while technically fantastic, is sorely mismanaged. Alpha has apparently fallen prey to the same forces.

 

[Keysplayr]

Originally posted by: nemesismk2
Intel are playing catchup because AMD has a huge advantage with their excellent 64bit cpu. The question should be how long will it take Intel to release a 64bit chip which is equal to AMD's?

I noticed that nobody has voted for "intel forever", where are you intel owners?

Ever consider that the AMD64 is only "excellent" because there is nothing else to compare it with except an intel 32 bit CPU right now?