AMD - Piledriver chitchat

[CAT-THE-FIFTH]

8.5W idle is pretty impressive. I like how he considered low load efficiency which I think is very important if you want to save power.

[CAT-THE-FIFTH]

We tend not to flip out over 50 W unless dissipating that heat requires a noisy fan. But if you’re in Denmark paying $.40/kWh, just the 10 W difference between Core i5 and FX-8350 at idle costs you several bucks per month. Under load, you’re looking at up to a $15-a-month difference for a system running 24/7.

I decided to check the TH figures. I tried using this calculator:

http://www.ukpower.co.uk/tools/runni...s_electricity/

I tried with 10W extra,and assumed a calender month of 672 hours.

TH says Denmark pays $.40/kWh which is 25p:

http://www.xe.com/ucc/convert/?Amoun...rom=USD&To=GBP

10W more at idle 24/7 gives you £1.68 extra a month.

Now I calculated that 40W extra for 25% of the month,ie, 44 hours a week gaming would cost £1.68. So if the system stayed at idle for the remaining 504 hours that would cost you £1.26 more.

So the total would be £2.94 a month extra with a rig 24/7 on with 25% time spent gaming in Denmark.

UK version- part one

Now,I checked this website:

http://www.sust-it.net/energy_calculator.php

Using a UK average tariff
1.)10W for 672 hours comes to just under £1/month
2.)Using a mix of 75% 10W and 25% 40W,it comes to around £1.70/month

Lets look at the Hexus power consumption figures:

http://hexus.net/tech/reviews/cpu/46...er-cpu/?page=6





http://www.legitreviews.com/article/2055/13/



Before anymore tries to accuse me, saying I cherry picked the best gaming power consumption situations for the AMD CPUs,I actually had a quick look through the reviews in the reviews thread and those are the two instances I could find.

UK version - part two

Now lets assume the AMD system consumes 100W more at load and 25 watts more at idle. After we need to make the AMD system look worse. Lets look at a non 24/7 system.

Saying you have the PC on 8 hrs a day and spend 3 hrs gaming. You then keep it on the whole weekend(48 hours) and game for 12 hours a day. Basically you have nothing else better to do!!

So that is 49 hours idle and 39 hours gaming. That would be 196 hours idle and 156 hours under gaming load.

That comes to around £3 a month more. However,Hexus does not indicate this under a gaming load.

Maybe these costs are not correct for the tariffs,but I CBA looking at my electricity tariff at 3.30am in the morning.



Perhaps,if you are an accountant you can plan your paid for away break in Butlins in three years.

[HalloweenJack]

really? complaints over power consumption on an enthusiasts forum oO???

if your that worried - change the light bulbs , since they are switched on far longer than your pc is....

[Biscuit]

i have changed all my lightbulbs to energy saving ones

CAT - fair enough, when you do the sums it all adds up to not a huge amount, that ~£15 a year you save (arbitrary number) could be spent on intel where you get better performance per watt and make it back.

Given rising energy costs, every little helps,

[HalloweenJack]

which energy csaving ones? new LED ones or older style?

[Biscuit]

Older style unfortunately, im not in the flat often enough to warrent LED ones at the moment

[HalloweenJack]

bleh - theres your £15 a year saving then

[Biscuit]

Well not really, at ~£80+ to change all my light bulbs it would take me ages to save that back given im only in the flat for a couple of weeks before i go away for 3 months.

[CAT-THE-FIFTH]

i have changed all my lightbulbs to energy saving ones

CAT - fair enough, when you do the sums it all adds up to not a huge amount, that ~£15 a year you save (arbitrary number) could be spent on intel where you get better performance per watt and make it back.

Given rising energy costs, every little helps,

OTH,that could be wiped out,by not properly looking at monitor power consumption when getting a new one - how many reviews look at that?? The same goes at looking at low load PSU efficiency:

http://techreport.com/review/20401/a...-the-desktop/5

That is a more extreme example,but still proves my point. That is just changing PSU on a low load platform to one more efficient.

For example,I have tried to get people to use this PSU with mixed results:

http://www.kitguru.net/components/po...pply-review/6/

http://www.jonnyguru.com/modules.php...tory2&reid=313

The KitGuru tests are done at 230V,and the PSU is nearly 88% efficient at 88W.

Here is another popular one,the CX 430 V2:

http://www.kitguru.net/components/po...pply-review/4/

It is 82% efficient at 75W and is used in loads of pre-builds and DIY builds.

That 6% drop in efficiency,means that 10W idle figure can go either way in reality.

It is because it is "only" 360W even though it was systems which really did not require a 400W to 500W PSU.

Socket FM1 and FM2 CPUs already do better at idle and low load than many Intel CPUs too. Websites like TH,don't really try lame arguments for them do they?? For the Intel CPUs they do!!

Plus,the AMD CPUs tend to to better with undervolting if required especially the AM3+ ones from what I gather. I have had mixed results trying to undervolt newer Intel CPUs.

Now,don't get me wrong Intel makes some great low wattage CPUs like the E3-1220L V2 which costs silly money,but they are primarily more interesting for where they can plonked into,like old cases!!

[scaryjim]

If you've got GU10 fittings the savings are much more - I've recently knocked 188W off the power consumption of ONE light fitting in my house...

Of course, if you are the kind of person who worries about power consumption that much you shouldn't ever consider a performance desktop. A 65W A10-5700 should give you an adequate all-round computing experience and will sip power

[Biscuit]

OTH,that could be wiped out,by not properly looking at monitor power consumption when getting a new one - how many reviews look at that?? The same goes at looking at low load PSU efficiency:

Im aware of this, but i dont understand why you think this figure is not looked at. Both are generally covered in their respective reviews from reputable sites. I don't see why it would be mentioned in a CPU review though.

[CAT-THE-FIFTH]

Im aware of this, but i dont understand why you think this figure is not looked at. Both are generally covered in their respective reviews from reputable sites. I don't see why it would be mentioned in a CPU review though.

Because again,you are just obsessing about one metric which is of no relevance unless considered with all others. You keep on ignoring what several people in this thread have said.

You fail to realise,if you go around loads of forums,hardly anyone looks at these things like monitor power consumption or load effiencies,so your obsession with meaningless amounts is just weird.

Look at loads of the biggest forums. Look on Hexus . LOOK!

LOOK at the number people asking for monitor advice on Hexus and asking about power consumption

LOOK at the number people asking for PSU advice on Hexus and asking about low load power effiency

LOOK at the number people asking for motherboard advice on Hexus and asking about power effiency

LOOK at the number people asking for graphics card advice on Hexus and really fretting about 10W to 20W difference in power consumption in a standard rig

Then LOOK at a bigger forum like OcUK or Bit-tech or even Anandtech and see the number who obsess about them.

Go onto SPCR and yes you will see that more often - because that is a SFF and LOW POWER forum. That is more to with cooling.

So people then obsessing about CPU power consumption only, is hilarious,beacuse for all their OCD about it,it can account for NOTHING in reality.

So go and worry about a £1 to £2 difference which only an accountant would care about because they like spreadsheets.



PS:

I am not going to talk on any of this any longer. It is getting boring.

[watercooled]

Monitor power consumption is generally measured at default backlight settings, and low PSU load goes down to 80W or so, if you're lucky - modern systems idle far lower than that, and it can make an enormous difference to efficiency. Some smaller supplies like the 360W Seasonic one was tested down to ~40W (10%) on JG, but that's still a fairly high load for a basic system, and it's only tested at 120v, but PSUs tend to be more efficient running from 240. Basically, there's still not enough attention paid to truly real-world idle load scenarios IMO.

[CAT-THE-FIFTH]

If you've got GU10 fittings the savings are much more - I've recently knocked 188W off the power consumption of ONE light fitting in my house...

Of course, if you are the kind of person who worries about power consumption that much you shouldn't ever consider a performance desktop. A 65W A10-5700 should give you an adequate all-round computing experience and will sip power

Basically, there's still not enough attention paid to truly real-world idle load scenarios IMO.

No! No! System power consumption for a general purpose rig is defined as a function of Cinebench and Prime95 loads! :sarcasm:

[DanceswithUnix]

It is because it is "only" 360W even though it was systems which really did not require a 400W to 500W PSU.

The manual that came with the Gigabyte FM2 motherboard I just bought suggested a minimum of 500W psu. I think part of that is the problem that not all 500W psus are equal. The other problem is the sheer range of power requirements a system might see.

Not measured the A10-5800 on idle yet, but lets say it is 40W at the wall. As pointed out, that is probably 20W on the motherboard, and 20W lost in the PSU as they are toilet for efficiency at these low loads. Now I plug in a graphics card, and play a game on it. I am burning 300W and don't really want to buy a new PSU to do it.

I think the only way that can really be fixed is a new way of powering motherboards. It could probably idle on the 5V-SB line already, but then would require some time to bring up the PSU when going on load.

Edit to add: Bought my first LED lightbulb yesterday. Seems Tesco are replacing the Phillips ones they used to stock with own brand units that are slightly cheaper, so were selling off their remaining Phillips ones for just under £7. Not one to refuse a cheap toy I though I would try one. Not told anyone at home I replaced that bulb, I shall see how long it takes someone to notice

[watercooled]

I think the ATX and 80-Plus specs both need to be renewed if significant improvements are to be made with idle efficiency. Intel must be keen on it so maybe they'll throw their weight about to make it happen? Until then, we're stuck with MFRs focussing on PSUs capable of powering a small rail locomotive, which are monstrously inefficient at below about 200W and are designed to meet the spec to get a shiny 80-Plus badge, at the expense of decent efficiency where it's most important.

I think I remember Intel talking about powering PCs with just a 12v supply, which would at least simplify PSU design, but then you're leaving it up to the motherboard MFRs to design reasonably efficient DC-DC converters which might not necessarily be any better, unless there were decent rules in place.

On the subject of light bulbs, LEDs are where it's going, but after looking to get some myself, I don't think they're there yet. The Phillips Master series are like £40 each, but they only seem to do dimmable ones, which is probably where a lot of the money is going. Otherwise, they seem fairly good, good colour temp, CRI, etc.

LEDs have the potential to offer better quality light than CFLs, but cheap ones may ruin your opinion of them, so be careful. Also, everything needs to be heatsinked properly or they won't last long. Since LEDs have a Vf of around 3v, we'd probably be better off having a single, efficient, low-voltage PSU, say 12v, near the fuse box or something, then the fittings could drop that to the Vf of the LEDs. Current BC/ES light fittings were designed to power incans, and 240v AC is far from ideal for modern lighting, requiring every bulb to have an internal PSU which will be constantly baked from the heat of the lamp.
Of course, you have potential problems with voltage drop, especially in larger buildings, but even higher voltages (say 24-48VDC) would be better than mains AC.