Just LoL
They aren't bad in terms of power usage to be honest, the same amount of transistors on the previous process, would have used more power, and it uses marginally more than a 295, but is quite a bit faster in some instances, which is fair enough.
Its also got comparable power usage to the 5870, unfortunately, thats per transistor and per size, its 50% or so bigger, and about 60% higher power usage.
It would be a decent amount higher power on 55nm, this is the problem Nvidia dont' have worse problems than AMD, the Fermi isn't bad, its got the exact same power usage per transistor as AMD, the problem is AMD is getting 90% of the performance in a much smaller number of transistors.
TSMC aren't great, thats a given, but the idea that Fermi is somehow not a bad design responsible for its own power usage irks me. Next gen will be equally as bad for Nvidia if they insist on using 50% more transistors, for 10-15% more performance, they'll always have a higher power, less efficient part. People seem to forget that the 280gtx, the 285gtx both had yield, size, and power issues aswell, the low end part in each series, the 260gtx, when forced to compete with the equivilent profitable part from AMD, couldn't turn a real profit. The Fermi is no different because its the same style of design, Fermi 2 will have the same problem and they'll continue to blow it until they increase efficiency per transistor.
The 280gtx used something like 225W didn't it, officially 250W isn't bad, but really its 300W, the difference there, the 70W, is because TSMC have got leakage and power problems, the fact that it would still be a very high power card without it, is the main issue.
