Or if you read the response without confirmation bias:
If you run the chip at 125C, an abnormally high running temperature, you can dissipate the heat with room temperature air (25C).
Wrong way round. If you run the chip under normal conditions with a room temp of 25C, the chip loads at 125C.
tldr: They have zero thermal headroom. You can say 'worst case' all you like, but look what happened with BFL and others. Power consumption is always more than expected, so these aren't 'worse case', these are realistic simulations. I really can't imagine them being able to run at these temps consistently, stably.
No! Read the response. It's a
simulation, and they set the chip's temperature
artificially high at 125.
I was reminded earlier to never initially ascribe malice when simple human tribe instincts can explain as much of someone's behavior.
After all, Chicago Bears fans will badmouth the Packers; they hate the Packers because they love the Bears. Buick owners would badmouth Oldsmobiles (easy to do, now, that they're no longer made) because the Olds was not a Buick.
So it is with those who go onto one thread or another and badmouth that vendor. Sure, there are some whose job is to shill and counter-shill, and there are those who deeply believe that one or the other vendors are a con-game, if not a criminal fraud. But they're quite rare. Most posters are likely just people who identify strongly with one of the {BFL/Avalon/AM/KnC/bitfury/[scam of the week]} tribes.
All others, therefore, suck, and they're going to explain why to us.
Now, I'm going to go where I should have gone directly, and that's to catch up on zerohedge dot com, which is part of my Saturday afternoon routine. That where the real FUD lives; you guys are amateurs compared to that tin foil crowd.
Read the first line, the rest was just gibberish :/
This is thermodynamics, you don't 'set' a temperature for the chip. You set airflow characteristics, ambient temps, all the material and conduction properties, convection properties blah blah, and most importantly: the heat output of the chip.
The temp it settles at is the result of your simulation and is determined by the thermodynamic relationship of the system.
UNLESS, this is a faked simulation which really isn't a simulation - but a fixed temp heat dissipation exercise. Set chip temp to 125C fixed, 2D analysis, ambient temp in -> what happens to the airflow? Means absolutely nothing, provides absolutely nothing, tells you absolutely nothing but would produce a pretty graph like we see here.
And that is an auto scaled legend (min and max, spread linearly), if its showing 223 as the max then somewhere on that simulation is a spot temp of 223. He can say otherwise but....
supertldr: Simulation is either still wrong and they don't know it is/don't care/don't understand/know its wrong and just did it for the pretty picture. The results of the simulation are questionable.