Anandtech system power usage chart

[Wave101]

Anandtech always does system power information in all of its reviews. I would like to know if the power usage is from the wall or within the system itself. I am sure there is a article somewhere on anandtech that gives us an idea of how they find system power. Please inform me.

 

[spittledip]

From the wall=power within the system (if by "power within the system" you mean "power being used by the system").

 

[Wave101]

There are 2 different type of power measurments. One being from the wall, which includes the power used by the power supply since it is not 100% efficient. The second being by the system components excluding the power supply. Since Anandtech does not tell us what type of power supply is used in his tests, we are not given information on what the power supplies efficiency is, thus possibly skewing the results.
So my question remains, does anandtech get its numbers from the wall (including power supply inefficiencies), or from the components excluding the power supply.

Unless that is what you just answered haha. Either way, how are we to tell what the power supply efficiency is.

 

[Wave101]

Let me just go into my problem. I just find that the anandtech values are too vague.

I will use the SLI GTX 580s for an example and an example power supply efficiency of 80%.

The chart says 850w. That could mean this:

850w at the wall socket which means the system uses 850*0.8 = 680w

Or it could mean:

850w system use, so 850/0.8 = 1062w at the wall socket.

The chart anandtech uses is vague in that it does not tell us which he is measuring or what the efficiency of the power supply is. Really, this chart is almost useless in determining anything other than some graphics cards use less power than others.

 

[Sunny129]

Let me just go into my problem. I just find that the anandtech values are too vague.

I will use the SLI GTX 580s for an example and an example power supply efficiency of 80%.

The chart says 850w. That could mean this:

850w at the wall socket which means the system uses 850*0.8 = 680w

Or it could mean:

850w system use, so 850/0.8 = 1062w at the wall socket.

The chart anandtech uses is vague in that it does not tell us which he is measuring or what the efficiency of the power supply is. Really, this chart is almost useless in determining anything other than some graphics cards use less power than others.

i'm not 100% sure b/c i haven't verified it, but i have a feeling that the figures Anand provides us with in his reveiws are measurements taken from the wall - that is, they have the PSU's inefficiency factored in. otherwise the reviews would be under-reporting real-world power consumption, or at the very least uner-reporting the amount we'd get charged for by the electric company. so it would also be my guess that when an AT review states the test rig's power consumption at 850W w/ a PSU that's 80% efficient at that power draw, it means that the system is really drawing 680W not including the PSU's inefficiency...

...would love to know for sure though myself.

 

[Zap]

I'm 99% sure it is power draw from the wall.

I'm not sure PSU efficiency matters a whole lot because the first thing is that I believe most people have a normal 80Plus or 80Plus Bronze PSU because Silver/Gold/Platinum are rare and expensive, and have only become common in the past year. Thus the difference in power may be in the range of a couple watts per hundred. The other thing is that power supplies are not exactly the same efficiency at all wattages, so even if they are using the same exact PSU and tell us which PSU they use, the results they get will not be linear and will be skewed slightly. Heck, a slightly differing AC voltage can affect PSU efficiency, so even if you build an EXACT replica of their test rig your numbers may be a hair off.

This goes for everything else that can be measured or benchmarked. Due to variances in, well, everything... you will NEVER be able to EXACTLY duplicate numbers. You can, however, get in the ballpark. Thus, it may be best to take those power draw numbers as a ballpark figure as a measure of relative power draw, and not the exact power draw you should expect.

 

[Wave101]

My issue with the numbers is that anandtech has power supplies that it has tested and knows the efficiencies at various states of power draw. It would be nice if they could use that information to give us a very close estimate of power used in the system rather than at the wall socket. That information would be very helpful in choosing the proper power supply for a video card.

 

[dclive]

My issue with the numbers is that anandtech has power supplies that it has tested and knows the efficiencies at various states of power draw. It would be nice if they could use that information to give us a very close estimate of power used in the system rather than at the wall socket. That information would be very helpful in choosing the proper power supply for a video card.

To me this means people are over-spec'ing (overbuying) power supplies. A quick check suggests that for all non-CF/SLI systems, 500-600 W PSUs would be plenty, and for most people (i.e. nVidia 460/560 and lower) a 400W PSU would be more than enough.

 

[Wave101]

To me this means people are over-spec'ing (overbuying) power supplies. A quick check suggests that for all non-CF/SLI systems, 500-600 W PSUs would be plenty, and for most people (i.e. nVidia 460/560 and lower) a 400W PSU would be more than enough.

Well that is pretty obvious to almost anyone who can read this chart. What is not obvious is what power supply one should get if they want a viable upgrade path for the future by SLI or Crossfire.
Heck, Anandtech could be using a 90% efficient power supply or %70. Difference being 765 watts vs 595 watts system use. Makes a big difference in price for a ~200 watt difference.

 

[Zap]

What is not obvious is what power supply one should get if they want a viable upgrade path for the future by SLI or Crossfire.
Heck, Anandtech could be using a 90% efficient power supply or %70. Difference being 765 watts vs 595 watts system use. Makes a big difference in price for a ~200 watt difference.

Two things.

1) Extremely unlikely they are using a 70% efficient PSU. Thus, your range would be more like 80-90% efficient. In fact, unless they are in the habit of changing their PSU, I would say you are looking in the 80-85% range. Then again, it may not matter because if they use the same PSU for all the tests because a PSU is not exactly the same efficiency at all outputs. Thus, the GTX 580 SLI may be closer to 80% while the same PSU and system running a single GTX 580 might be closer to 85%.

2) If the extra cost for another 100W completely breaks the bank, how can that person afford SLI/Crossfire?

 

[SparkyJJO]

Also bear in mind that if the system consumption on the DC side of the PSU is 350W, do you really want to run a 400W PSU for that? No. The closer you run a PSU to its max, the more stress that is on it and the shorter its lifespan potentially is.

Also, as PSUs age, they can't provide as much power. So my 620W Corsair I have now probably is able to do only 600W or a bit lower now. This isn't a problem since I have more than enough room to spare, but if you were cutting it close with a PSU, as it ages it may not be able to reliably power the system anymore in just a couple of years.