A smaller PSU at 80% utilization should consume less power then a larger PSU powering the same load at 40% utilization.
That depends upon the efficiencies of the two power supplies you're comparing. Look at the plot in
this report in figure 7. For most of the units shown there, the efficiency peaked between 40 and 80% load and was pretty flat in that region. You can compare specific units and find significant deviations from the above plot but of the many I've seen, this is very representative of the average. So basically, your 40 to 80% comparison is simply not valid.
As you reach the upper limits of the PSU you loose at least some effeciency to heat and such, but bigger capacitors, larger power regulators, and bigger converters are inerrantly less effecient then smaller components of the same design and quality.
This is going from bad to worse. As far as I know, you lose to heat at low loads too. Large caps are not necessarily less efficient than small ones.
Did you actually read that graph? The one on page 14?
You do bring up a good point, if this magical 40-80% still holds true for modern power supplies, so my example may have been off by 4%. However, if you actually READ THE GRAPH you will see that I am conceptually right, and if I changed my example to 74% on a smaller PSU vs. 37% on a larger PSU I would be completely right. I'm all for being corrected, but when it's a minor technicality that is wrong don't be an asshat and try to claim that I'm completely wrong. Of course you loose electricity due to heat at lower loads too, but when you start to reach the maximum thermal design load of the system you start to loose a greater percentage. But when you're under the designed electrical load of the system you loose electricity due to using power components that are too big for the job too.
1. The PSU with the highest tested effeciency in the 40-80% load range was a 220w model. The 250w model was the 3rd worst of the 8 models tested. Quality > Maximum rated wattage, like I said.
2. If you're comparing the best item in the list, the Rush 2 220w (second generation model of their 220w PSU) to the worst one, the 175watt "Cobra" you're comparing apples to organges, like I said.
3. Let's compare the best PSU to the highest rated PSU: Rush 2 220w vs. the Prowler 250w. A load of 80% on the Rush 2 is 176w at 75% effeciency, which is a 70% load on the Prowler at 72% effeciency. Quality > Maximum rated wattage > "lower load = more effeciency".
4. Let's try comparing the two PSUs with the biggest difference in rated wattage: the 50w Shrimp to the 250w Prowler. 80% load on the shrimp is 40w @ 72% effeciency the Prowler is at 16% load and 60% effeciency. Larger components are inherrantly less effecient, especially for smaller-than-intended loads. It's called entropy and capacitance. They electrons simply get lost. If you tried to use gold bars (perfectly clean and insulated, perfectly welded together) as a long run of speaker wire for lower power speakers your speakers wouldn't make any noise. The gold bars would absorbe and disperse the electrons before they got to the speakers because they are too big despite being one of the best conductive materials available.
5. Lets try the example that most closely fits the 80% vs. 40% example I orginally gave: the 175w cobra vs. the 80w shrimp. 80% load on the shrimp is 64w @ 72% effeciency. 36% load on the cobra @ 64% effeciency. This isn't the best example as the cobra is obviously a crappy PSU. The 80w Shrimp at 100% load still beats the cobra.
5. THis graph shows 4 models by the same manufacturer. Two 200w and two 220w. They are the ones that are almost identical. The other >200w PSU achieves it's best effeciency at 60-80% load, not 40-80% load. The Only other PSU that hits 40% and then is nearly level until 80% is an 80w PSU that kicks the crap out of a 175w PSU. This is hardly difinitive data.
So, the 40-80% range may be important (although it's probably more like 60-80%, and really PSUs are generally benchmarked for effeciency at 80% load), thank you for pointing that out, but as we can see *quality* is by FAR the most important factor when choosing a PSU and real-world testing is far more valuable than manufacturers ratings.
So, does anyone have any benchmarks of *current* PSUs? There's gotta be someone who is irrate enough at me going against the "more is better" enthusiast grain who is just itching to find benchmarks that show some nice beastly PSUs that actually have better effeciency at a wide range of given loads then some of the smaller ones that are available. Those are the gems we should be looking for and reccomending, not telling people "get a 1kw PSU". When we tell people that "uber wattage" is more important than quality, and they have a budget, they will buy a lower-quality PSU with a bigger number on it. Unless a gem of a high-end PSU can be found you're far better off speccing out your system, speccing out some possible future upgrades and buying the best PSU in your budget that will run that system at ~80% load then you are buying a lower-quality PSU with a big number on the label. It should be more reliable, have more stable rails, and should use less electricity. PC power supplies are fairly complicated and involve numerous design decisions that affect performance ranges and multiple components of varying quality and design. The maximum rated wattage of the unit is the first thing you shoud look at, but it is NOT the most important attribute of the unit.
Enthusiast and gaming PCs tend to have very large budgets but wether you're spending $600 or $6000 on a PC you should spend your money wisely... or just buy from a system builder.