I think what he means by work is the energy used to lower the barrier potential to allow the free movement electrons through the substrate (apply power to the gate/base of a transistor), but that is technically not work since work is a mechanical energy and transistors are based on electromagnetic energy so the unit would just be...well...power
yes, i used "work" wrongly but this ^
I think you're saying that since the actual instantaneous power fluctuates, there's no way to summarize it meaningfully. You're raising a key point, I think it's very important. At the same time, the TDP numbers given by CPU (not GPU) manufacturers are meaningful numbers for determining needed PSU capacity. Note my use of weasel words.
Of course, the TDP number is not the random number printed on the box… those numbers are made up.
It's all dissipated as heat, and that includes the dynamic power.
https://www.ece.cmu.edu/~ece322/LECTURES/Lecture13/Lecture13.03.pdf
By the way I came across a PSU estimator that's better than any I've seen before. It came up with a reasonable answer for my system and it's a good way to see many of the factors that go into the estimation. Also note the flashing red text.
http://www.extreme.outervision.com/psucalculatorlite.jsp
Yeah I used that a lot when planning a build just to make sure I had adequate wattage for my system. Recently purchased a watt meter and my computer consumes around 250W-270W while stressing CPU/GPU/RAM and fans at full speed.
Although it does not accurately show the actual system wattage its good for picking a reasonable power supply.