One note about AMD - they undervolt very well, and default voltages selected by BIOS is in most cases very high. My X4 620 had a default voltage around 1.35V, yet my undervolting stopped at 1.12V (real, 1.1V in BIOS).
I may have my physics wrong, but (from the first page) if a particular product uses 50 per cent more power, but it's also twice as fast (100 per cent) at completing the task at hand, then doesn't this mean this is 33 per cent more efficient, not 50?.
For this article, Efficiency = Speed/Power (electrical efficiency is actually [effective energy]/[power consumed]. I assume that [effective energy] is proportional to speed.)
Letting a represent speed and b power,
a/b = E
2A/1.5B = 1.33 E
Originally Posted by sotu1 Any chance we can get some numbers on how much it costs to run a high powered PC per hour vs a low powered PC per hour? That would be pretty cool :)
It depends on how much are you charged by your electrical company. They can't really do it, but perhaps they could do it in their context, that is, the London context. :D
What about GPUs? was really looking forward to that bit... :( nevermind. The rest was quite interesting. I'm looking forward to more powersaving specials. :0)
Originally Posted by der_george What about GPUs? was really looking forward to that bit... :( nevermind. The rest was quite interesting. I'm looking forward to more powersaving specials. :0)
Just look at the reviews? They have the power consumption charts in each and every review, don't they?
For the UK, as a quick and dirty way to estimate things, I use 1 Watt per year = £1.
So, the difference between leaving on a machine all the time drawing 50 Watts and one that draws 75 Watts is roughly £25 per year.
So saving 10 Watts on a machine that you run for 4 hours a day is less that £2 over the course of a year. If this costs you £40 to achieve, you have to have over 20 years of use in order to justify it. The price of electricity almost certainly will go up, but I'm guessing it won't go up enough to justify the cost given the average lifespan of the equipment (whether through redundancy or failure).
Efficiency does not equate to ultimately saving money, leaving the payback time out of equations is misinformation at best.
(NOT having a go a Bit-tech or anyone in particular, it just seems likes it's fashionable within the "Green" industry to use possible savings rather than actual costs)
if a particular product uses 50 per cent more power, but it's also twice as fast (100 per cent) at completing the task at hand, then this means its 50 per cent more efficient.
Something fishy there... If product A uses 100W (0.1kW) of power and does 1 unit of work per hour, we can say its efficiency (workrate / power) is 1 / 0.1 = 10 units/kWh. If product B uses 50% more power (150W = 0.15kW) and is twice as fast (2 work units per hour), then its efficiency can be said to be 2 / 0.15 = 13.33 units/kWh, making it 33.3% more efficient, NOT 50% more efficient.
if a particular product uses 50 per cent more power, but it's also twice as fast (100 per cent) at completing the task at hand, then this means its 50 per cent more efficient.
Something fishy there... If product A uses 100W (0.1kW) of power and does 1 unit of work per hour, we can say its efficiency (workrate / power) is 1 / 0.1 = 10 units/kWh. If product B uses 50% more power (150W = 0.15kW) and is twice as fast (2 work units per hour), then its efficiency can be said to be 2 / 0.15 = 13.33 units/kWh, making it 33.3% more efficient, NOT 50% more efficient.
Which is something Jezcentral and me have mentioned. ;)
if a particular product uses 50 per cent more power, but it's also twice as fast (100 per cent) at completing the task at hand, then this means its 50 per cent more efficient.
Something fishy there... If product A uses 100W (0.1kW) of power and does 1 unit of work per hour, we can say its efficiency (workrate / power) is 1 / 0.1 = 10 units/kWh. If product B uses 50% more power (150W = 0.15kW) and is twice as fast (2 work units per hour), then its efficiency can be said to be 2 / 0.15 = 13.33 units/kWh, making it 33.3% more efficient, NOT 50% more efficient.
Which is something Jezcentral and me have mentioned. ;)
Originally Posted by Jipa Good article. I just noticed that my stupid HD4870 consumes over 70 W on idle. Loving how they cut the idle consumption with the HD5-series cards.
I know AMD isn't as power efficient as Intel but you can still have some fun with reducing power usage. Just keep default CPU clock settings and create smartprofiles in AOD for demanding games. I'm using this(2.8GHz ->3.5GHz) for Crysis, CoD, Fallout 3 and it works like a charm. If only it was possible to turn on/off Cool'n'Quiet feature on-the-fly...
Originally Posted by Kúsař I know AMD isn't as power efficient as Intel but you can still have some fun with reducing power usage. Just keep default CPU clock settings and create smartprofiles in AOD for demanding games. I'm using this(2.8GHz ->3.5GHz) for Crysis, CoD, Fallout 3 and it works like a charm. If only it was possible to turn on/off Cool'n'Quiet feature on-the-fly...
If your PC is idle most the time (like if you browse the web, listen to music, etc), then idle power draws are much more important to focus on. Athlon 240e is the clear winner here. It has the lowest idle power draw of any CPU I have seen. It uses LESS THAN 5 W !!! That includes VRM losses so its pretty impressive. http://www.lostcircuits.com/mambo//index.php?option=com_content&task=view&id=70&Itemid=1&limit=1&limitstart=4 "the power numbers are un-adjusted gross input into the motherboard's voltage regulator module and are, therefore, inflated by roughly 30%. meaning that the CPU itself draws somewhere around 3-3.5W power in idle mode"
Originally Posted by faugusztin One note about AMD - they undervolt very well, and default voltages selected by BIOS is in most cases very high. My X4 620 had a default voltage around 1.35V, yet my undervolting stopped at 1.12V (real, 1.1V in BIOS).
indeed, if you disable CnQ in bios and use this handy bit of software instead
You can run your machine with high voltages for full power overclocks for gaming and have it go down to very low power state at idle and have it ramp up in three stages.
My machine is quite happy with 0.8v at 1Ghz and so I use that all the time, plenty for most (well nearly all) tasks to be honest, but when gaming I want cooking mode 1.6v for the High CPU Overclock.
You can set what mode it boots up in via a start up shortcut which is handy, Auto (like CnQ but with your custom settings) , low, mid, high.
Originally Posted by sandys My machine is quite happy with 0.8v at 1Ghz and so I use that all the time, plenty for most (well nearly all) tasks to be honest, but when gaming I want cooking mode 1.6v for the High CPU Overclock.
I found the same thing. 0.8v at 1GHz. Under load I undervolt as well so it's using somthing like 1.1 or maybe a bit more.
It's 100% stable but if I tired setting it to even 0.9v at 1GHz in the BIOS, it won't even POST :(
Lol, I think anybody who builds their own PCs is kidding themselves as to how "green" they may be. I mean, upgrading is not a "green" process in itself, even if you sell your own kit, it just means there's more kit out there requiring more resources to be used.
Lol, come on, we ain't "green", the best we can attribute to ourselves is a label of being electricity bill conscious. :)
Comments 1 to 25 of 46
ReplyLetting a represent speed and b power,
a/b = E
2A/1.5B = 1.33 E
So yes, math error in the article. ;)
It depends on how much are you charged by your electrical company. They can't really do it, but perhaps they could do it in their context, that is, the London context. :D
Just look at the reviews? They have the power consumption charts in each and every review, don't they?
So, the difference between leaving on a machine all the time drawing 50 Watts and one that draws 75 Watts is roughly £25 per year.
So saving 10 Watts on a machine that you run for 4 hours a day is less that £2 over the course of a year. If this costs you £40 to achieve, you have to have over 20 years of use in order to justify it. The price of electricity almost certainly will go up, but I'm guessing it won't go up enough to justify the cost given the average lifespan of the equipment (whether through redundancy or failure).
Efficiency does not equate to ultimately saving money, leaving the payback time out of equations is misinformation at best.
(NOT having a go a Bit-tech or anyone in particular, it just seems likes it's fashionable within the "Green" industry to use possible savings rather than actual costs)
Good article. I just noticed that my stupid HD4870 consumes over 70 W on idle. Loving how they cut the idle consumption with the HD5-series cards.
EDIT: Also the sound card seemed to draw more power than I thought.
Which is something Jezcentral and me have mentioned. ;)
;)
We cover GPUs in depth, so I didn't bother :)
E8400+HD4870 -> i5 750 + HD4870 = 190->160W idle power usage
i5 750+HD4870 -> i5 750 + HD5870 = 160->117W idle power usage
It is funny that moving from dual to quad and from less to more powerfull graphics card results in smaller idle power usage. :)
But you can reduce power usage with Intel too!
Idle means it just sits at the desktop DOING NOTHING.
Unlike what you think, browsing the web does consume CPU cycles, and so does listening to music.
indeed, if you disable CnQ in bios and use this handy bit of software instead
http://crystalmark.info/software/CrystalCPUID/index-e.html
You can run your machine with high voltages for full power overclocks for gaming and have it go down to very low power state at idle and have it ramp up in three stages.
My machine is quite happy with 0.8v at 1Ghz and so I use that all the time, plenty for most (well nearly all) tasks to be honest, but when gaming I want cooking mode 1.6v for the High CPU Overclock.
You can set what mode it boots up in via a start up shortcut which is handy, Auto (like CnQ but with your custom settings) , low, mid, high.
Yes but when the web page is loaded, the PC is just sitting there idle while you read it. I did say if it was idle most of the time, not all the time.
Playing music is trivial. It must be using the bare minimum of CPU cycles. Nothing that's going to make the cores clock up.
I found the same thing. 0.8v at 1GHz. Under load I undervolt as well so it's using somthing like 1.1 or maybe a bit more.
It's 100% stable but if I tired setting it to even 0.9v at 1GHz in the BIOS, it won't even POST :(
Lol, come on, we ain't "green", the best we can attribute to ourselves is a label of being electricity bill conscious. :)
-
« Previous
-
1
-
2
-
Next »
Discuss in the forums