What really is the power consumption of a PC ?

[Sleepy]

Popped into Maplin recently and saw a 'power meter thingy' - you plug it
into the wall socket and the appliance
plugs into the meter. The meter provides displays of minimum and maximum
wattage used - timed usage and if you
input the price of your kwh it'll tell you the cost of running said
appliance.

I've been meaning to do this for ages so I bought it and tried it with my
PC and got a real surprise. The box consumes between
120 idle and 168 watts when gaming (STALKER, Crysis etc). The LCD screen is
25w and speakers another 5w.

Like Tony (see post below this) I've been looking at power consumption of
graphics card and imagining I've been
using 300w to 400w regularly but this meter clearly shows a different
picture.

my system is C2D E7200, 2x1gb DDR2 800, 8800GT, 2x SATA II HDDs, DVDRW, SB
Audigy and couple of 92mm case fans for
airflow. use that only PSU calculator (Extreme) and it says my power usage
is close to 400w which is why I have a 550w.

I ran this meter from dawn til dusk and did everything on the PC - gaming,
scanning and printing (scanner runs off the USB)
and burned a DVD too. Max wattage was only 192w so now I'm really puzzled.

Can anyone enlighten me?

 

[terryc]

Can anyone enlighten me?

The 300-400 watt probably comes from the capacityof common power supplies.
but as you have found out, it varies depending upon what components you
are actually using. Also, assuming that the meter is correct.

 

[Paul]

Popped into Maplin recently and saw a 'power meter thingy' - you plug it
into the wall socket and the appliance
plugs into the meter. The meter provides displays of minimum and maximum
wattage used - timed usage and if you
input the price of your kwh it'll tell you the cost of running said
appliance.

I've been meaning to do this for ages so I bought it and tried it with
my PC and got a real surprise. The box consumes between
120 idle and 168 watts when gaming (STALKER, Crysis etc). The LCD screen
is 25w and speakers another 5w.

Like Tony (see post below this) I've been looking at power consumption
of graphics card and imagining I've been
using 300w to 400w regularly but this meter clearly shows a different
picture.

my system is C2D E7200, 2x1gb DDR2 800, 8800GT, 2x SATA II HDDs, DVDRW,
SB Audigy and couple of 92mm case fans for
airflow. use that only PSU calculator (Extreme) and it says my power
usage is close to 400w which is why I have a 550w.

I ran this meter from dawn til dusk and did everything on the PC -
gaming, scanning and printing (scanner runs off the USB)
and burned a DVD too. Max wattage was only 192w so now I'm really puzzled.

Can anyone enlighten me?

Is the power supply a high efficiency one, or a regular one. I can try my
crude calculation method, and see what number it gives.

Processor 65W / 0.90 = 72.2W (includes waste heat from 90% efficient Vcore)
Motherboard_memory = 50W (covers the low power of modern RAM, plus the hot chipset.)
(with an Intel chipset, there is data available for this.)
Hard drive 12W x 2 = 24W (some modern drives do much better than this now)
CDROM = 25W (only if media present, 5W otherwise perhaps)
8800GT = 78 to 86W (Xbitlabs 3D peak measurement, two different designs)
Fans = 6W (say, up to three fans)
Standby = 10W (covers some USB loading on +5VSB)
Add-in cards = 0W (low power and not worth estimating)
Total 265W (at output side of ATX PSU)
AC input power = 265/0.80 = 331W (AC input to computer, assuming an 80% efficient ATX supply)

And you got 192W.

Paul

 

[VanguardLH]

Popped into Maplin recently and saw a 'power meter thingy' - you plug it
into the wall socket and the appliance
plugs into the meter. The meter provides displays of minimum and maximum
wattage used - timed usage and if you
input the price of your kwh it'll tell you the cost of running said
appliance.

I've been meaning to do this for ages so I bought it and tried it with my
PC and got a real surprise. The box consumes between
120 idle and 168 watts when gaming (STALKER, Crysis etc). The LCD screen is
25w and speakers another 5w.

Like Tony (see post below this) I've been looking at power consumption of
graphics card and imagining I've been
using 300w to 400w regularly but this meter clearly shows a different
picture.

my system is C2D E7200, 2x1gb DDR2 800, 8800GT, 2x SATA II HDDs, DVDRW, SB
Audigy and couple of 92mm case fans for
airflow. use that only PSU calculator (Extreme) and it says my power usage
is close to 400w which is why I have a 550w.

I ran this meter from dawn til dusk and did everything on the PC - gaming,
scanning and printing (scanner runs off the USB)
and burned a DVD too. Max wattage was only 192w so now I'm really puzzled.

Can anyone enlighten me?

What is the hysteresis in measurement? That is, how much lag is there
when there is big jump in wattage and then a big drop? An averaging
power meter won't help you figure out how much you need to handle the
peak loads, like when you first power on or when you load an application
that pumps up the load on the CPU, memory, mobo, video card, and hard
disks all at once. I've seen folks actually try to use their house
power meter to figure out how much an appliance consumes but that never
shows peak consumption over a short interval. Did you see a spike when
you powered on your computer? If not, the hysteresis is too long so all
you could ever measure is the typical operational power consumption
where there isn't much change in what components are powered on and how
hard the host is working. Along a similar vein, a window air
conditioner may read low enough when it is running to be on a 15A
circuit but the spike when you first activate the compressor would blow
the breaker if a couple of lamps were also on at the time.

Unless you are getting very high quality power supplies and have some
independent testing to show they actually support their rated maximum
load for indefinite time intervals (and actually the best ones are
underrated in that they supply more power than they are rated for by
anywhere from 10% to 25%), figure on getting maybe 65% to 75% of the
rated load (for a constant load at that value).

Did you have every IDE, SATA, SCSI, or other port connected to a hard
drive? Was every tap on the PSU connected to something? Was every
memory slot filled? Was every PCI slot filled? Were you running games
or Prime95 during testing of power consumption? While your load isn't
close to what you may have expected, remember that you probably want
expandability later without having to buy a bigger PSU when you later
add drives, add more memory, play a super-resource intensive game or
video editing app, etc. Of course, without an oscilloscope, you can't
see how much the ripple voltage increases as you approach the maximum
load capacity of the PSU, so you really want to make sure you never come
close to its max.

By the way, did you measure the wall outlet wattage and then remember to
multiply by the PSU's efficiency rating to figure out what the PSU was
delivering for power level?

 

[Sleepy]

What is the hysteresis in measurement? That is, how much lag is there
when there is big jump in wattage and then a big drop? An averaging
power meter won't help you figure out how much you need to handle the
peak loads, like when you first power on or when you load an application
that pumps up the load on the CPU, memory, mobo, video card, and hard
disks all at once. I've seen folks actually try to use their house
power meter to figure out how much an appliance consumes but that never
shows peak consumption over a short interval. Did you see a spike when
you powered on your computer? If not, the hysteresis is too long so all
you could ever measure is the typical operational power consumption
where there isn't much change in what components are powered on and how
hard the host is working. Along a similar vein, a window air
conditioner may read low enough when it is running to be on a 15A
circuit but the spike when you first activate the compressor would blow
the breaker if a couple of lamps were also on at the time.

Unless you are getting very high quality power supplies and have some
independent testing to show they actually support their rated maximum
load for indefinite time intervals (and actually the best ones are
underrated in that they supply more power than they are rated for by
anywhere from 10% to 25%), figure on getting maybe 65% to 75% of the
rated load (for a constant load at that value).

Did you have every IDE, SATA, SCSI, or other port connected to a hard
drive? Was every tap on the PSU connected to something? Was every
memory slot filled? Was every PCI slot filled? Were you running games
or Prime95 during testing of power consumption? While your load isn't
close to what you may have expected, remember that you probably want
expandability later without having to buy a bigger PSU when you later
add drives, add more memory, play a super-resource intensive game or
video editing app, etc. Of course, without an oscilloscope, you can't
see how much the ripple voltage increases as you approach the maximum
load capacity of the PSU, so you really want to make sure you never come
close to its max.

By the way, did you measure the wall outlet wattage and then remember to
multiply by the PSU's efficiency rating to figure out what the PSU was
delivering for power level?

Thanks to Terry, Kony, Paul and Vanguard. It is a cheap meter - £8 from
Maplin
isn't going to get me a professional quality meter I know but it does seem
to be accurate.
I watched the small LCD screen on booting up my PC and it did show
fluctuations
in wattage before settling down to around 120w at desktop. For gaming I
chose
STALKER because it uses the PC pretty thoroughly including the HDDs as it
continually
loads data as you move around. It never went above 170w when gaming.

Yes I am aware of the need to check the individual amps on the rails in
particular the 12v.

I'm not about to dump my 550w PSU for a 300w - no fear !!

I was just pleasantly surprised to see my general power consumption was less
than I feared and
less than most hardware reviews indicate. Also by switching off the
multi-sockets that feed my
PC, TV and hi-fi I found I was saving 18w at night.