Date: Thu, 30 Jun 2005 11:03:58 +1000
From: Raymond <[EMAIL PROTECTED]>
Subject: RE: [LIB] Libretto U-100
At 05:34 PM 29/06/2005 -0700, you wrote:
Date: Tue, 28 Jun 2005 22:25:32 +0400
From: Vitaly Pavlenko <[EMAIL PROTECTED]>
Subject: RE: [LIB] Libretto U-100
> Date: Fri, 24 Jun 2005 12:28:29 -0700 (PDT)
> From: John <[EMAIL PROTECTED]>
> the 723 cpu intel makes runs at
> under 7 watts (if I remember right) and normally the
> libby itself should run at under 1-2watts total?
Well, the system comes with the 60 W power source which is a good estimate
of the total power. Of course, it would charge the battery at the same
time, so maybe something like 30 W. My lib SS1000 came with 30 W PS but
runs fine from third-party 15 W one.
Actually just because the power source can give X watts of power doesn't
mean it actually GIVES that all the time - case in point, you plug it into
the wall but don't plug the laptop in, it gives exactly 0 watts of power.
All that rating means is that's the amount of power that can be drawn
whilst guaranteeing the output voltage and dissipated power specifications.
To actually get the amount of power being drawn, you need to measure the
current going into the computer (and voltage if you've got particularly
poor regulation - switchmodes like all computer PSUs nowadays are generally
nice for that). A multimeter in series with the power supply (low voltage
side of course!) will give you a good indication (although the purists will
argue that you need an oscilloscope since the current drawn by the computer
is hardly likely to be dead flat as it has its own switchmode regulator -
despite the filtering caps). Also just because it "runs fine" from a 15W
one doesn't mean it isn't drawing MORE than 15W - those can stand a fair
amount of overloading before showing signs of huffing and puffing and even
if the output voltage drops, the switcher in the laptop can often handle
being a few volts under spec. Some laptops like my current one even test
the adapter when you plug it in and will do things like disable
simultaneous charging and running if it detects an adapter that isn't up to
spec (not sure how it does it - it could briefly cause a current surge to
test the capabilities of the adapter or it may send some sort of data back
down the power cable to query some smarts in the adapter).
Maybe a simpler test is to use battery capacity - take the AH rating (or
divide the mAh rating by 1000) and multiply it by the battery voltage and
you've got the number of rated watt-hours in the battery. Divide this by
the runtime you get on the computer and you've got a rough average for the
power drawn by the computer during the run. I say rough because the output
voltage generally isn't constant (although for lithium it's more so than
others) and the capacity rating is rarely accurate firstly because cells
deteriorate and secondly because the internal circuitry will probably shut
off the power before fully draining the battery (lithiums have a habit of
not waking up again after being drained too far).
IMHO Intel is cheating us somehow. They no longer include true dissipated
power in the specs but give "thermal design guideline" instead. This
matter is beyond my qualification, so please correct me if I am wrong.
The difficulty with providing "true dissipated power" is nowadays
processors employ all sorts of weird and wonderful ways of saving power.
For instance, the transistors on the chip (FETs of various types) are
highly capacitive so they only draw (and dissipate) power when they switch.
To save power, the processor is divided into a pile of different sectors.
If a sector isn't being used, the clock signal to that sector is switched
off - nothing in that sector switches anymore and so it dissipates almost
no power. IIRC this is called "clock gating".
Now if you take all the combinations of ways in which a processor may save
power, combined with things like clock throttling and the like, you end up
with thousands of combinations, each with a different "true dissipated
power". I guess Intel *could* publish all of these but then it'd just
confuse the consumer even more - which combinations actually get used more?
At the end of the day, it really depends on what you do with it.
Thermal design guidelines are an attempt to combine what Intel knows about
power dissipation of their chips with what they know about what parts of
their chips will get used in which cases to find some sort of nice average.
Basically they tell the equipment manufacturers how much power they need to
source (electrically) and sink (thermally) in what situations to avoid
having the chip overheat. Of course, manufacturers do get this right and
wrong to varying degrees - my Dell Inspiron 9300 monster has a Pentium M
2.0 and a GeForce 6800 but even when playing 3D games the base and air
vents only get mildly warm. In contrast, we've got a Sharp AL3DU at work
(Pentium M 2.0, GeForce 6600), the base of which gets uncomfortably hot
even when idling.
</rant> :-)
- Raymond
---
/~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\
| | "Does fuzzy logic tickle?" |
| ___ | "My HDD has no reverse. How do I backup?" |
| /__/ +-------------------------------------------|
| / \ a y b o t | [EMAIL PROTECTED] |
| | Need help? Visit #Windows98 on DALNet! |
| ICQ: 31756092 | www.raybot.net |
\~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~/
