Ummm, no. That would be almost logarithmic. Its not. Its linear.
How could it be 50% less heat with only 5% increase in efficiency? My calculous is a bit rusty but if your statement were true, you would be talking steel melting temperatures compared to your basic, generic 70% efficient PSU.
Remember, it is very rare for a properly sized PSU to be maxed out. This is because a properly sized PSU must assume the CPU, GPU, motherboard, all drives, all fans, and everything else the PSU is powering are all demanding maximum power at the exact same point in time. And the odds of that are slim at best and even if it occurs, it will only be for a second or two at a time.
I don't know where you are getting 5%.
Gold is 87% at 20% load. 90% at 50% load. 87% at 100% load.
Titanium is 92% at 20% load. 94% at 50% load. 90% at 100% load.
But lets use your 5% and assume 50% load on that 750W supply.
50% load on a 750W supply = 375 watts.
At 90% efficiency, that means the PSU must pull from the wall 417W (417 x .9 = 375.3).
At 95% efficiency, that means the PSU must pull from the wall 395W (395 x .95 = 375.25).
As you can see that 5% better efficiency means just 22W difference - 22W in the form of heat. Hardly 50%, right? And a 22W space heater sure wouldn't do much to keep closet warm on a cool day. Depending on the fuel and who you talk to, the heat from on candle is 40 to 70W! So even if the demand on that PSU was its full 750W capacity, you are still only talking 44W difference - actually less than that since the
80PLUS cert says 3% difference between Gold and Titanium at 100% load.
[ ... ]