Power consumption of Pentium 4 570J (3.8Ghz)

  • Thread starter Thread starter ykhan
  • Start date Start date
ykhan said:
Intel quietly introduced the latest Prescott, a 3.8 Ghz chip with the
NX-bit support available (that's why it's got the "J" designation).
They seem to have introduced a new, improved halt state to keep things
cooler.

Intel's Pentium 4 570J 3.8GHz processor - The Tech Report - Page 15
http://techreport.com/reviews/2004q4/pentium4-570j/index.x?pg=15

Yousuf Khan

I had a long chat with someone who works for a review
site. A lot of what he wanted to see added to that
site's review was omitted because his boss didn't want
to piss off Intel.

The differences between idle and loaded power consumption
as reported by Tech Report are in the same ballpark as what
my friend and his co-workers saw. Unfortunately, the
"loaded" power consumption comes into play even when you
are doing something so trivial as playing Solitaire or
using a simple text editor. They spent a lot of time
unsuccessfully trying to find *anything* an XP user could
do without kicking the power consumption up to the same level
you would see when running something like Prime95.

"Loaded" also apparently comes into play for something as
trivial as the Windows XP log-on screen saver. I found it
quite amusing to hear that you could stop typing, mousing,
etc and immediately see the power consumption drop by about
90 watts, and then watch the power consumption jump all the
way back up to the "loaded" level - and stay there - when the
screen saver eventually kicked in. And no, they weren't
running Folding or anything like that. The only screen saver
that did not cause this behaviour was the simple screen blanker.

I /expect/ to eventually hear that they found they set up
something wrong in the BIOS options or in the XP options,
but until then I intend to take claim by Intel and the
review sites with a big grain of salt.
 
Intel quietly introduced the latest Prescott, a 3.8 Ghz chip with the
NX-bit support available (that's why it's got the "J" designation).
They seem to have introduced a new, improved halt state to keep things
cooler.

Just as a matter of interest, does anyone know the power consumption of
a 2.2 Ghz Celeron? Is there a processor power consumption table somewhere
out there in cyberspace?

Buck
 
Just as a matter of interest, does anyone know the power consumption of
a 2.2 Ghz Celeron? Is there a processor power consumption table somewhere
out there in cyberspace?

Intel lists a TDP (Thermal Design Power) for their Celeron 2.2GHz of
57.1W. It's tough to accurately compare this to the P4 3.8J GHz
processor mentioned above though, since Intel uses a somewhat
different definition of "TDP" for their newer chips, however that chip
is listed at 115W.

AMD, meanwhile, specs their latest and greatest chips for 95W TDP,
though again they are using a different definition of the term. As a
rough guesstimate, I would say that the 2.2GHz Celeron probably is in
the same ballpark as a Athlon64 3000+ or 3200+ (running at 2.0GHz,
built on a 130nm fab process, much like that Celeron 2.2GHz).
 
Intel quietly introduced the latest Prescott, a 3.8 Ghz chip with the
NX-bit support available (that's why it's got the "J" designation).
They seem to have introduced a new, improved halt state to keep things
cooler.

Intel's Pentium 4 570J 3.8GHz processor - The Tech Report - Page 15
http://techreport.com/reviews/2004q4/pentium4-570j/index.x?pg=15

I think the actual CPU consumption is probably lower than that by
maybe 50%? He's measuring at the wall plug after all and that would
include the PSU inefficiency (measured at 80% 250W, 76% 150W by some
website before), the extra stuff like motherboard, hard disk, graphic
card, RAM as well as the motherboard voltage regulator's inefficiency
(say 85%), so that's 242*0.8*0.85 = 164W at most for load and 130 *
0.76 * 0.85 = 84W.

Comparing that against a Prescott we're playing with overclocked to
3.8 with much higher than default vcore. The processor's idling power
draw is pretty much in the 4~5A 12V range, which puts it at 60W x 85%
motherboard regulator efficiency to 51W. The 3.3V and 5V used mainly
by the motherboard and RAM takes a consistent 25W, while the
connectors going to the hard disk, graphic card takes around 20W or
so. That would put us at around 96W.

But our vcore is like set to about 12.6% higher than the default
around 1.39V so that makes dynamic power 27% and static power 43% (did
I get that right from the previous power discussion?) higher than a
default processor from Intel. Assuming static is 40% and dynamic is
60% of power consumption, this makes a difference 33% or 13W from 51W
(51/1.33). So we should be seeing some 83W on idle, viola, same
ballpark figure

For load, we've got almost the same figures for non-CPU items, just a
bit higher on the 3.3V since the RAM are active. Make that 50W for
these stuff. The processor is drawing around 14~15A so that's 180W x
85% = 153W and adjusting for vcore, the default processor should be
seeing some 115W, which is way off from techreport's. But incidentally
is also Intel's TDP for a 3.8G processor

Comments anybody? Why is there such a huge difference in the load
power even though the idle is pretty close? Or is my speculative maths
just really bad?



--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me :)
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code
 
He's measuring at the wall plug after all and that would
include the PSU inefficiency (measured at 80% 250W, 76% 150W by some
website before)

I've yet to see a PC power supply that could come anywhere near 80% under
any load whatsoever. They're typically in the 60-70% range.
 
I've yet to see a PC power supply that could come anywhere near 80% under
any load whatsoever. They're typically in the 60-70% range.

Your head is so far up your ass one might postulate that you are
trying to clone your brain.
 
Back
Top