AMD - Piledriver chitchat

[Terbinator]

Are people really derping all over the internet due to 40W under load? Why don't more websites test gaming loads?

You must be new here?

As for Aria,

[CAT-THE-FIFTH]

You must be new here?

The thing is that Batman and BF3 thread reasonably well IIRC,so I suspect in more lightly threaded cases,the difference will be less.

As for Aria,

Don't show them this then:

http://images.anandtech.com/graphs/graph4955/41704.png

http://www.extremetech.com/wp-conten.../11/RageVT.png

Another new engine,ie,id Tech 5 which threads well.

That might give them heart palpitations,and then they will have to go find some game that uses one thread based on a 15 year old engine. It is probably called superPI master.

I still wonder how 90% of WoW players play WoW, due to their lack of 5GHZ Core i7 CPUs and quad GTX690 SLI rigs.

[HalloweenJack]

hmmm interesting - apparantly my `old` AM3 board will work with a BD chip on the beta bios.... with a wall of vrm`s - would be fun to try a full on BD as well as a PD

[Terbinator]

There is a 95W 8300 too if you think your board may not be up to the full 125W.

[CAT-THE-FIFTH]

The GA-770TA-UD3 has an 8+2 phase VRM,so is quite a solid motherboard.

[Biscuit]

The Toms Hardware explained the issue over power usage the best IMO:

But then there’s power to consider. In the United States, we’re blessed to have relatively inexpensive energy. 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. Advantage: Intel.

Given the rate in which energy prices are increasing, this is becoming more and more important. Im going to be looking at switching my energy suppliers this year to try and save a few quid, not because i cant afford it but purely because its getting ridiculous!

[Terbinator]

The Toms Hardware explained the issue over power usage the best IMO:


Given the rate in which energy prices are increasing, this is becoming more and more important. Im going to be looking at switching my energy suppliers this year to try and save a few quid, not because i cant afford it but purely because its getting ridiculous!

I guess by the same rule, why buy an i5 'K' if you don't plan on going beyond its TDP anyway.

[watercooled]

Sorry to go off on one but I've seen complaints about superpi performance on a few forums. WHY??

Why does anyone still care about a completely irrelevant, limited-scope, catastrophically inefficient, x87 benchmark when deciding on what CPU to buy?

Edit: About power consumption, Trinity has proven PD needn't draw a load of power at idle, which is why I at least partly blame the AM3+ platform (sorry if I've already said this 10+ times in this thread). I wonder how much difference can be made with a more efficient uATX board?

[Biscuit]

I guess by the same rule, why buy an i5 'K' if you don't plan on going beyond its TDP anyway.

to a certain extent, however intel isn't affected quite so adversely as AMD by overclocking in a performance per watt basis.

[CAT-THE-FIFTH]

The Toms Hardware explained the issue over power usage the best IMO:


Given the rate in which energy prices are increasing, this is becoming more and more important. Im going to be looking at switching my energy suppliers this year to try and save a few quid, not because i cant afford it but purely because its getting ridiculous!

I am into SFF PCs,so tend to be more concerned with power consumption and peak TDP than most.

Yet,the benchmarks done are not really realworld though??? Look at the gaming power loads - 40W is nothing and that is with very high end parts.

If you want to reduce power consumption of a desktop there are two ways to do this:

1.)Choose components careful

The reviews use high end parts with often poor effiency ratings. Just changing between different motherboards can effect the power consumption at load and idle a decent amount,ie,less phases usually means better power consumption by a noticeable amount,or at least systems which can switch off phases when required. Even going for a smaller form factor does help,ie,mini-ITX against full ATX at times.

I remember X-bit labs testing an E350 motherboard with a 4 phase VRM which consumed 50% more power, than those with a more sane number of phases for a 18W TDP CPU.

The same goes with graphics cards - these are the singular worse component for power consumption. How many people chose the GTX460 over an HD6850 despite this:

http://www.anandtech.com/show/4002/a...asus-xfx-msi/7

With a GTX460 in the system,there was around 40W more power consumption measured at the wall and many GTX460 cards were pre-overclocked too.

Moreover,do people actively seek out the most efficient designs?? How many go for pre-overclocked designs with fancy coolers. Usually some of the reference ones can be more efficient like with the HD7850 and HD7870.

Other considerations include a PSU with good efficiency at low loads as this where your system idles. You ideally want a flat response.

Another is your monitor - this is often entirely ignored. Monitors can vary massively in power consumption by even 50W IIRC.

Less fans and drives also effect power consumption.

This is why this obsession with a few watts on the CPU side only is annoying. There is so much more to making an efficient system and yet all the other things seem to be ignored.

I also think that companies like Intel also do try to put pressure to emphasise certain things.

How many reviews made a bigger deal of Trinity and Llani desktop power consumption under uncommon loads, than its lower idle and low load power consumption when compared to the competition?? Surely that is even more important in the scheme of things??

2.)Don't overclock

This increases power consumption.

However,notice how many reviews do not bother to look at idle power consumption when overclocked. I noticed this in the two or three instances it was measured:

http://www.hardocp.com/article/2011/...mance_review/9

http://www.bit-tech.net/hardware/cpu...8150-review/10

These are two sites I don't even like,but its just lame that hardly any sites bothered to do so.


Other thoughts:

1.)Are reviews a correct indication of usage habits??

Do reviews actually consider the average usage habits of most people. Looking at most of the target crowd,I would think their PCs mostly run at idle.

Lets consider peak tasks - how often are they run and how long do they last?? Most people might do some video encoding and image editing during the year but is it at a sufficient level to really impact yearly power consumption?? 3 hours spent video encoding a month won't make much difference,would it?? 80 hours might.

However,gaming is most likely to be a more intense workload run more often - but then how many hours a week?? Would you also give up framerates and settings to use a slow CPU and GPU which consumes less power??

How many reviews have tested platform power consumption with different game loads??

I still think that people look at performance first and a more cursory glance at efficiency later.

2.)Usage habits

You want to save power - a PC which is switched off consumes nothing. So simply use the PC when it is needed.

Using a tablet or even a low powered notebook saves you loads of power for most purposes.

I have done this in the past when I had a overclocked Q6600 system with a 975X motherboard - idle power consumption on that system was poor.

3.)Does it matter in the scheme of all the other devices you own??

Think of your fridge,your freezer,your TV,etc?? People might have a look at the effiency of some of these devices,but do they really look into detail?? Do they calculate whether the Sony 40" TV is 20W more efficient than the £50 cheaper and prettier Samsung they want?? I doubt it.

How many people keep things on standby instead of switching things,off,etc??

I just tend to be very cynical I think at times.

[watercooled]

Something else about monitors, take some time to calibrate it; MFRs commonly set brightness and backlight to 100% at the factory, washing out colours and massively increasing power consumption. Realistic backlight levels are usually around 25% for general indoor use, and this can lower power consumption by 50% or more on some monitors, while improving picture quality and potentially extending the life of the backlight.

[Biscuit]

Component choice: In general i couldn't agree more, the most important one being chipset but as has been mentioned before, AMD has decided not to develop a new platform so they are stuck with any inefficiencies in 890/990 that are intrinsic to its design. I did look into this when i got my UD5 and tbh, with 990FX there didn't seem to be enough in it to go one way or another and it made more sense to go for the features that fit what i needed.
Im confused by your statement that people dont seem to consider power usage in GPUs as much... you're just wrong, so wrong and i think when you think about it and read GPU reviews you must know that. Think about the whole GTX480 power thing and how many jokes there were about it. Yeah people bought the 460 but we both know that there are a lot of people out there who will just buy Nvidia, regardless of the performance differences because they have this false illusion AMD are useless.
Regardless of this, and comparing to all the other components, we are talking about comparing CPUs. If you want efficiency in as many components whilst retaining maximum power in the widest range of applications, Intel wins hands down. Then add overclocking into the mix aswell, the case is still the same. Your individual requirements WRG to hard drives, fans or anything else is totally up to the user and selecting efficient models of them has no (rightly so) baring on CPU decision unless they have some kind of performance restriction.

Dont Overclock: If you are overclocking then you are irking out max performance so you are more likely to care about top end voltage/power rather than idle. Although yeah it is a shame its not measured in more interviews as I can think of any particular reason to leave it out... not like its difficult to measure once you have measured the other power draw figures.

Usage habits: I turn my PC on, game, turn it off. As you have done, I tend to use my netbook or tablet for browsing and pretty sure this kind of usage scenario is far from rare... so why would idle power matter to me or anyone else?
I dont really want to sacrifice FPS for power... so why would i look at AMD? They have worse/the same performance and higher/the same power usage as intel equivalents.

A lot of other points relating to other household components are utterly irrelevant in a CPU review though Cat. In reality whether someone does or doesn't look at power usage for a a fridge or toaster, even a television in many regards, has no relevance to a PC component review.
Many people (myself included) don't even have any say on what their white good are.
I agree people overlook these things in the reality of life then spend far too long worrying about it in their CPU, but peoples domestic life decisions are just irrelevant for reviewers.

[Terbinator]

I have my backlight on 10% lol.

[DDY]

I have my backlight on 10% lol.

So how is that 20 carrot a day diet going?

[CAT-THE-FIFTH]

Nvm.

[watercooled]

In regards to motherboard efficiency, I doubt I'll be buying another full ATX board as they tend to have efficiency low down on the priority list in my experience, and I only really need PCIe slots for GPU and sound card now. It is a shame you see relatively little attention paid to efficiency of the motherboard as it can have a huge impact on power draw; I think it's a large part of the reason my spec rig draws ~90W idle while my Llano rig draws about 25W, still on an ATX PSU. With some very rough estimation, I reckon the board is somehow burning 30W or so over the FM1 board; a small part can be attributed to the chipset, but still I reckon most is down to design. It's taught me to be more careful with motherboard selection in future.

I can't remember if I found this or if someone linked it in another thread, http://ssj3gohan.tweakblogs.net/blog...nglish%29.html
Shows how little power a PC really needs when idle.