Frankly, I don't think either temp or power consumption is the ideal thing to look for . . . although power consumption probably comes closest.
Temps depend in part to the fan profile used, and the efficiency of the gpu's heatsink. So the temp of Card_A could be lower than Card_B, but B might throw off more heat . . . maybe into your case . . . maybe not.
IMO, power consumption is more likely to dictate the "hotter" card . . . though its possible for one chip to use power more efiicently than another . . . so power consumption could be misleading . . . up to a point.
I'd suggest reading qualified reviews - like here and Anand's site among a few others - and pay attention to what these reviewers indicate about the hotter - or cooler - card, and whether the difference is enough to think about.
The op is right Tom's Hardware charts can be really helpful in there you can see the same GPU from different manufactures they have different temperatures so if you go the manufactures web site you see the cards using different cooling systems and that must be looked out also. 3d power draw must be looked out and this is an issue that really counts on me.
Finally I’ve chosen HD5770 for these reasons later on I discovered there is no 3d party cooling system for this cards tested at this time, new GPU and the problem is if you don't check the 3d noise also in charts you may end up playing some demanding games with head phones over your head until something new and effective comes out.