Computer electricity costs

quoia

hansi wrote:

I accept that, but my PC is not on for 24/7 but I keep it on during the day when I am at home, and I reckon it does more harm than good to keep switching it on and off all day. That is my personal view.

Firstly, let me quote from the same article in PC Pro.

QUOTE:

COMPONENT LIFE - IS IT AFFECTED ?

A common notion propagated over the years is that switching your system off can reduce its life, due to the stress caused by the power-cycling process. However, there's little evidence for this. Hard disks, for instance, are usually rated for around 10,000 spin up/spin down cycles. And the reduced overall wear on bearings of both the hard disk and the system fans will, if anything, INCREASE their lifespan.

End quote.

From the quote, the 10,000 spin up/down cycles actually equates to a daily power up and power down of your system EVERY day for over 13.5 years !
Remember the most unreliable things tend to be those with moving parts and these all rely on bearings not wearing out.


From my own experience, whilst I would say it used to be a problem, I believe it no longer is.

I started working on computers before the PC was ever thought of.
They were machines the size of a couple of large wardrobes with the computing power of a digital clock or half of the 1st ever Sinclair pocket calculator.
Indeed several of them together would have struggled to meet the brains of the 1st ever microproccessor or Sir Clive's ZX80. (Remember them?)

Anyway in those days silicon chips were relatively new and manufacturing processes in their infancy compared with today. Components often failed if you sneezed near them. Air conditioned computer rooms didn't exist and our systems were marketed for installation in an "office environment". When occupied during the day, the offices would be baking hot, especially in summer, during the night they really would be freezing cold. Temperature swings exceeding 30 or 35 degrees centigrade was not unknown. (I doubt inside your home varies by more than 10 degrees all year round). The cooling and filtration on these computers was !!!!!!. The system was full of hot spots and the dirty environment led to filters becoming clogged and blocked.
The systems themselves literaly had 1000's and 1000's of components and millions of connections. Chips usually were not even soldered into PCBs, indeed many systems didn't even have PCBs (Printed Circuit Boards - these were just being "invented") but had WIRE-WRAP boards.
Chips (ICs - integrated circuits) usually were plugged into IC sockets mounted on a board, each socket had extra long legs sticking out the back and one or more wires would literally be wrapped around it. These wires would go to 1 or more other pins on other IC sockets.
Because of their overall physical build and design complexity, fault finding was a nightmare on large boards, so often many smaller boards were used in building an entire system. (I rememer boards with 600 or more IC's on them, probably 10,000 chip "legs" in total. That would mean 15,000 to 20,000 individual wires joining them all together with 30,000 to 40,000 hand wrapped pins. So a system easily had a million or 2.)

Splitting into sub-units helped one problem but caused another.
It allowed easier identification of a faulty component or wiring by board sustitution but for every separate board you introduced , you had to have a new interconnect between then - another PLUG and SOCKET !

So what's all this about I hear you ask?
Well every chip pin in a socket, wire to wire-wrapped pin and interconnect plug & sockets between boards, all relied upon a mechanical friction contact to maintain the circuit. Systems at that time definitely didn't like being disturbed and often developed a problem just AFTER preventative maintenance had been performed. Most faults tended to be seating problems on the interconnects and faulty boards taken back to the workshop worked after just taking a few chips out oif their sockets and putting the same ICs back in again. It became apparent that systems left on overnight suffered less failures on the non mechanical components - disc drives took a pounding though. The reason for the discs failing was usually head crashes. Unlike todays sealed units, these were open to the atmosphere and microscopic dust particles could get onto the disc surface. The read/write head instead of flying would crash into these like your car hitting a concrete bollard.

Anyway, the system boards became more reliable due to them not going through a daily heating / cooling cycle (perhaps 50 or 60 degrees swing inside) where each and every mechanical joint would expand and contract by the tiniest amount.
At this time we always recommended that systems be left turned on where possible.

Once systems got proper PCBs, the number of failures due to dodgy connections dropped dramatically but we were totally expecting this.
Chips themselves became more reliable and accordingly soldering them into PCBs directly became an option, saved costs and removed all those friction joints. 99% of all the wires became etched tracks and the number of joints dropped from 1 to 2 million down to about 50,000. There was so much less now that could be cause a problem and need "re-seating".

This in turn led to less component boards because they could be bigger and hence less interconnecting cables/plugs/sockets.
Chips ultimately got bigger (internally and externally) such that one 40 pin IC replaced a couple of hundred 16 pin ones.
Then we got the microprocessor and the PC and systems with only a few seperate boards and cards inside.
At this point leaving it on 24 hous a day became irrelevant regarding reliability.

Your current PC probably has less than 1000 "joints" altogether, they are far better made and are of a higher quality and design than the days of old. Most of these will be between the processor and motherboard itself, depending upon which Pentium or AMD chip you have, numbering around 2 to 5 hundred. Memory Simms or Dimms about 100 to 200 each and a dozen or so for the PSU connection to the motherboard. Anything ISA, PC or AGP is another 100 or so, and the hard drive, cd, dvd etc cables another 50ish per device.

The problem (if there ever was one with PCs) is defintely 99.999999999% in the mind and not borne out of current evidence.

The ONLY reason I don't turn mine off every night is the 5 minutes it takes to boot up from scratch when I turn it back on again due to it only being a Pentium II and a lot of software on it.
I'm aware of new P4 systems with faster discs and faster memory that will load XP in about 8 seconds from power on. When I get one of these I'll be turning mine off every night and saving about £3 a week!

hansi

WoW! Well that certainly shut me up !! (LOL)

quoia

hansi wrote:

Offices leave their PC's on 24/7 365 days per year and they don't seem to have problems. I like to keep mine on when I am home to check e mails, and if I'm reading the newspaper and there's something interesting on a website, I can check it straight away and being on broadband this is made very easy by being online continuously, bearing in mind that you must have a firewall enabled.

I agree there is no point turning it off if there is any likelyhood that you will be wanting to use it in a few minutes or so, and the longer it takes for your system to boot (and hence the time delay inconvenience factor) is partially in proportion to the period of switching it off, that would be considered a worthwhile amount of time to make any saving. However if you're tucked up in bed by 23:00 and don't use it again 8:30am the next day, why not give it a rest for 9 and a half hours, save yourself some money and reduce some CO2 greenhouse gas production at a power station?

RE Offices 24/7 365 - The users are not paying the electricity bill.
In the US this has actually become a serious energy issue. AMD responded first, and Intel have just followed suit, in the design of some new processors (just released) that signicantly reduce power consumption if they are "idling" to any degree. Specifically destined for SERVERS that usually have a 24/7 uptime, since they probably do 90% of their workload in a 9 to 5 environment and only 10% "when the office is empty", they reckon a company with 500 servers will save over $100,000 a year by upgrading.