mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Hardware (https://www.mersenneforum.org/forumdisplay.php?f=9)
-   -   Electricity use (https://www.mersenneforum.org/showthread.php?t=1902)

PrimeCruncher 2004-01-12 00:51

Electricity use
 
Okay, pretty simple question here:

When running Prime95, does the electricity consumption of the computer increase?

The reason I ask is that the CPU temp rises, but is that because the FP is now in heavy use or is it because it needs more electricity to power it? This is a question I'm liable to be asked when I borg those P3s, so I'd like to go in with a difinitive answer rather than a 'I don't think so.'

Complex33 2004-01-12 01:02

First Law

Prime95 2004-01-12 04:30

Your electricity use will increase. The amount depends on the CPU type. There have been a couple of threads about this in the past with some concrete numbers.

outlnder 2004-01-12 04:55

At approximately 7.5 cents per kilowatt hour, it will cost $8 per month per machine.

E_tron 2004-01-12 06:11

i noticed about a $25 increase on the electric bill here(at 10cents per kilo).

I beleive that the P4 would use the most energy.

lycorn 2004-01-12 12:09

Re: Electricity use
 
[QUOTE][i]Originally posted by PrimeCruncher [/i]
[B]Okay, pretty simple question here:

When running Prime95, does the electricity consumption of the computer increase?

The reason I ask is that the CPU temp rises, but is that because the FP is now in heavy use or is it because it needs more electricity to power it? This is a question I'm liable to be asked when I borg those P3s, so I'd like to go in with a difinitive answer rather than a 'I don't think so.' [/B][/QUOTE]

In fact, the CPU temp rises because the FP is in heavy use, that meaning that many more transistors are being switched on and off constantly, which implies that more electricity is being used. The values measured by several people in this forum point to an increase of 40 to 50 W when doing LL testing comparing to idle mode. Trial Factoring increases the power consumption by a smaller amount. But those values are for P4s and Athlons. P3 should consume less. You may reassure the owners of the borged P3s they won´t go bankrupt by allowing you to run P95 on them:innocent: :rolleyes:...

TauCeti 2004-01-12 14:33

In addition to the previous posts:

'Normal' P-III (500-1130 MHz, 0.18u process) draw about 7A in stop-grant/sleep/deepsleep mode and 10 (500 MHz) to 20 (1130MHz) Amperes under load. Core Voltage is about 1.6 to 1.75 V

That means:

Compared to idle-mode, for a 500 MHz P-III you can expect about 5 W more with prime95. A 1 GHz P-III should draw about 22 additional watt.

So if you borg a typical office system running 8/5, the figures add up to:

P-III 500: About 8KWh extra/year --> 0.8$
P-III 1000: About 45 KWh extra / year --> 4.5$

If you run a P-IV 3060 24/7 you get about 525KWh extra --> about 50$/year extra.

All numbers compared to idle excluding other system components.

PrimeCruncher 2004-01-12 20:14

[QUOTE][I]Originally posted by TauCeti [/I]
[B]That means:

Compared to idle-mode, for a 500 MHz P-III you can expect about 5 W more with prime95. A 1 GHz P-III should draw about 22 additional watt.

So if you borg a typical office system running 8/5, the figures add up to:

P-III 500: About 8KWh extra/year --> 0.8$
P-III 1000: About 45 KWh extra / year --> 4.5$

If you run a P-IV 3060 24/7 you get about 525KWh extra --> about 50$/year extra.[/B][/QUOTE]So let me see here... 22 watts per hour x 8 hours per day = 176 watts x 180 days = 31680 watts per year / 1000 = 31.68 KWh x 0.0705 cents per KWh (current electricity cost) = $2.23 (approx) per computer x 60 computers (again, approx) = $134 per year.

Thanks for all the replies, this'll be very helpful.

Angular 2004-01-14 03:02

How do you measure the amount of power drawn by a computer? Excluding the shutdown everything else in the house method :smile:

I was going to use a clamp on multi-meter, but I have heard that they make power meters specially designed for measuring power draw. I asked four different Radio Shacks and they have no idea what I am talking about.

I have read about 'Kill a Watt' - ~$30, [url]http://www.spytown.com/kilwatwatkil.html[/url]

Are these devices accurate?

TauCeti 2004-01-14 17:40

I (in germany) use a [URL=http://www.elv.de/Main.asp?Menue=Shop&Artikel=303-76&Gruppe=HT-EM]ELV EM 800[/URL] to measure.

Quite accurate and it measures real power, reactive power and apparent power for the whole computer system. It's not 100% accurate to measure CPU only, because increased CPU load normally also gererates increased thermal dissipation in the PSU and the mainboard power converters.

There is a link to a non-german device here somewhere in the forum but i *err* just can't find it...

The description of the device in your link does not contain any accuracy information. My EM 800 has a 0.1 W resolution and a 1% +- 3 digits accuracy whitch translates roughly to about 1.5% overall accuracy in the 100-150W range. But the EM 800 operates at 230V and costs 50 EUR (about 64$ today)


All times are UTC. The time now is 01:01.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.