🚀 go-pugleaf

JK wrote:
> McGrandpa wrote:
>
>> JK wrote:
>>> http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2149&p=7
>>
>> Yeah, a 64 bit processor beats a 32 bit processor by 18%.  I knew
>> this one months ago, where you all been?
>
> The reason for the high score on the Athlon 64 is the on chip memory
> controller and other enhancements, and the 64 bit ability. Those
> benchmarks were done with 32 bit software.
>
>>
>>
>> So, what's AMD gonna do when Intel releases a 64 bit processor?
>
> When 64 bit software is released, I expect the Athlon 64 chips  to
> show
> even greater leads over comparably priced Intel 64 bit processors.
>
>> Or
>> perhaps they shouldn't so they can assist the competition in staying
>> alive?   AMD simply hasn't built anything they can scale the clock up
>
> High clock speeds are a bad thing. They are associated with high power
> consumption and hot chips. It is better to increase performance in
> other ways if possible, such as using on chip memory controllers.
> Notice that Intel is swiitching to model numbers for its processors.
> I guess Intel will now
> try to increase performance by other means than just ramping clock
> speeds. The best chips are probably those that have the greatest
> performance
> per unit of power consumed, which probably translates to the best
> performance per unit of clock speed. It looks like Intel has made some
> progress with the Centrino, although its very high price seems to be
> an issue.
>
>>
>> in, so they spread the architecture out to 64 bits.  And they STILL
>> can't scale the speed up in their hardware.
>
> The speed in terms of benchmarks is great. Clock speed tends to
> correlate with high power consumption and hot chips, so knowledgeable
> people will
> tend to want a great performing chip with a low clock speed.
>
>>
>>
>> This race is an Indy 500 in which the leader is forced to allow the
>> 2nd place car to advance, to use a new engine in mid lap.  All in
>> order for there to even be a race.
>
> Are you implying that the  P4 EE chips are like an aircraft engine
> mounted on a car?  If someone can't get a car engine to compete well
> enough against other car engines, then they take an aircraft
> engine and mount it on a car.

No.  I've been saying that transistor for transistor, the AMD core runs
hotter at lower core speeds than Intels.  And that AMD, rather than work
on that issue at mask level to *allow* use of higher clocks with lower
temps, simply chose to spread into 64 bit.  And, as you say, use onboard
memory controller.
Intels Prescott die is 9 nanometer, AMD's 64 bit is 11.  The technology
exists to run at higher freqs with lower temps.  Intel is still pushing
that end and hasn't yet gone to running the same sort of functions that
AMD uses now.  WHEN Intel does, they will have a similar level of
performance per clock, and a much faster clock.
And it bothers me that the 64 bit processor is outperforming the 32 bit
processor by only 18%.  Even with running under a 32bit opsystem, I'd
think that the AMD 64 should be performing some 30 to 40% faster.

My own chip is a 3.0E Prescott 800 fsb.  I saw the scores.  Mine is dead
in the middle of everything.  That seemed reasonable for a $184 chip.
The same day, I bought a Athlon 3200+, for $199.
So who is selling the top AMD processors for $160?  Nobody here in
Houston, Texas is.
McG.

>
>>
>>
>> It's no secret, Intel has the die waiting.
>
>> And waiting.
>>
>> McG.